# 404

tr.rr:

TRANSA(TIONS

O"'J SYSTFMS. MAN, AND C\'BrRNETI('S,

VOl ..

20,

No.2,

MARCHI

APRJI.

1990

**Fuzzy Logic in Control Systems: Fuzzy Logic Controller-Part I
**

CHUEN CHIEN LEE,

STUDENT MEMBER, IEEE

Abstract -During the past several years, fuzzy control has emerged as one of the most active and fruitful areas for research in the applications of fuzzy set theory. especially in the realm of indnstrial processes, which do not lend themselves to control by conventional methods because of a lack of qnantitative data regarding the input-output relations. Fuzzy control is based on fuzzy logic-a logical system which is mucb closer in spirit to human thinking and natural language tban traditional logical systems. The fuzzy logic controller (FLC) based on fuzzy logic provides a means of converting a linguistic control strategy based on expert knowledge into an automatic control strategy. A survey of tbe FLC is presented; a general methodology for constructing an FLC and assessing its performance is described; and problems that need further research are pointed out. In particular, the exposition includes a discussion of fuzzification and defuzzification strategies, the derivation of the database and fuzzy control rules, the definition of fuzzy implication, and an analysis of fuzzy reasoning mechanisms.

I.

INTRODUCTION

URING the past several years, fuzzy control has emerged as one of the most active and fruitful areas for research in the application of fuzzy set theory [141]. The pioneering research of Mamdani and his colleagues on fuzzy control [63]-[66]. [50] was motivated by Zadeh's seminal papers on the linguistic approach and system analysis based on the theory of fuzzy sets [142]. [143], [145], [146]. Recent applications of fuzzy control in water quality control [127], [35], automatic train operation systems [135], [136], [139]. automatic container crane opera· tion systems [137]-[139], elevator control [23]. nuclear reactor control [4], [51], automobile transmission control [40], fuzzy logic controller hardware systems [130]. [131]. fuzzy memory devices [107], [l08], [120], [128], [129], [133]. and fuzzy computers [132] have pointed a way for an effective utilization of fuzzy control in the context of complex ill-defined processes that can be controlled by a skilled human operator without the knowledge of their underlying dynamics. The literature in fuzzy control has been growing rapidly in recent years. making it difficult to present a comprehensive survey of the wide variety of applications that have been made. Historically, the important milestones in

Manuscript received May 27, 1988; revised July I, 1989. This work was supported in part by NASA grant NCC-2-275 and AFOSR Grant 89-0084. The author is with the Electronics Research Laboratory, Department of Electrical Engineering and Computer Sciences, University of California. Berkeley, CA 94720. IEEE Log Number 8932013.

D

the development of fuzzy control may be summarized as shown in table I. It should be stressed, however, the choice of the milestones is a subjective matter. Fuzzy logic, which is the logic on which fuzzy control is based, is much closer in spirit to human thinking and natural language than the traditional logical systems. Basically, it provides an effective means of capturing the approximate, inexact nature of the real world. Viewed in this perspective, the essential part of the fuzzy logic controller (FLC) is a set of linguistic control rules related by the dual concepts of fuzzy implication and the compositional rule of inference. In essence. then, the FLC provides an algorithm which can convert the linguistic control strategy based on expert knowledge into an automatic control strategy. Experience shows that the FLC yields results superior to those obtained by conventional control algorithms. In particular. the methodology of the FLC appears very useful when the processes are too complex for analysis by conventional quantitative techniques or when the available sources of information are interpreted qualitatively, inexactly, or uncertainly. Thus fuzzy logic control may be viewed as a step toward a rapprochement between conventional precise mathematical control and human-like decision making, as indicated by Gupta [30]. However. at present there is no systematic procedure for the design of an FLC. In this paper we present a survey of the FLC methodology and point to the problems which need further research. Our investigation includes fuzzification and defuzzification strategies, the derivation of the database and fuzzy control rules, the definition of a fuzzy implication, and an analysis of fuzzy reasoning mechanisms. This paper is divided into two parts. The analysis of structural parameters of the FLC is addressed in Part I. In addition, Part I contains five more sections. A brief summary of some of the relevant concepts in fuzzy set theory and fuzzy logic is presented in Section II. The main idea of the FLC is described in Section III, while Section IV describes the fuzzification strategies. In Section V. we discuss the construction of the data base of an FLC. The rule base in Section VI explains the derivation of fuzzy control rules and rule-modification techniques. Part II consists of four sections. Section I is devoted to the basic aspects of the FLC decision-making logic. Several issues including the definitions of a fuzzy implication, © 1990 IEEE

0018-9472/90/0300-0404$01.00

wc shall briefly summarize some of the basic concepts of fuzzy set theory and fuzzy logic which will be needed in this paper. [41]. u B of A the union A U B is pointwise defined for all II E V by J1. xu. Fukami. Finally. a fuzzy set F can be written concisely as F = /uJ1.. A fuzzy set may be viewed as a generalization of the concept of an ordinary set whose membership function only takes two values (O. and J1.F(u))lu E V).. Some of the representative applications of the FLC. intersection and complement for fuzzy sets are defined via their membership functions. Definition 1: Fuzzy Set: A fuzzy set F in a universe of discourse U is characterized by a membership function J1. ( u) .B(u)}.F: U --> [0. which could be discrete or continuous. More specifically. A more detailed discussion may be found in [141]. Crossover Point. Togai and Watanabe Yamakawa Dubois and Prade
[SO]
compositional operators. Komolov et a/.AnB(u)
=
min{J1.F which takes values in the interval [0. I].. the composition of Rand S is a fuzzy relation denoted by
.B(U)}. The set theoretic A operations of union.. 1). J1.l' respectively.LEE: FUZZY
LOGIC II\: CONTROl
SYSTEMS: H_.. Sugeno and Murakami Kiszka.x of the complement of a fuzzy set A is pointwise defined for all u E V by J1. Miyamoto et 0/. [42]. the interpretations of sentence connectives "and" and "also.' • " u.J1. I] namely. respectively.x . the Cartesian product of A I" .F(UJjU.u.A(u). ( u) x
=
1. see the following.
Definition 5: Complement: The membership function J1.J1. and Fuzzy Singleton: The support of a fuzzy set F is the crisp set of all points u in V such that J1." and fuzzy inference mechanisms.A(u). " A" is a fuzzy set in the product space VI X .). A" are fuzzy sets in UI. When U is continuous.f = 0. Ostergaard Willaeys et a/.IC
CONTROLLFR-I'ART
1
4115
1 A rationale fur fuzzy control [145] Linguistic approach'[146] Steam engine control [64] Analysis of control algorithms [5J.0 is referred to as fuzzy singleton.ZZY LO(.J1. [7J Heat exchanger and cement kiln control Optimal fuzzy control [121] Finite automaton [57] Wastewater treatment process [113] Fuzzy conditional inference [24] Probabilistic fuzzy sets (control) [33] Derivation of fuzzy control rules [103] Predictive fuzzy control [135] Parking control of a model car [97] Fuzzy system stability [55] Fuzzy chip [107] Fuzzy controller hardware system [130] Approximate reasoning [21]
TABLE 1972 1973 1974 1976 1977 1977 1979 1980 1980 1983 19R3 1983 1984 1985 1985 1986 1988 Zadeh Zadeh Mamdani & Assilian Rutherford et a/.. [149] and [21].
Definition 4: Intersection: The membership function J1." '. Mizumoto and Tanaka Hirota and Pedrycz Takagi and Sugeno Yasunobu. Gupta et a/. we describe some unsolved problems and discuss further challenges in this field. In particular.B' respectively.
Definition 2: Support.
Fuzzy
SETS AND
Fuzzy
LOGIC
For the convenience of the reader.J1. X V" and is expressed as Ru. Set Theoretic
Operations
Let A and B be two fuzzy sets in V with membership functions J1.
B.• . [148]. are investigated. A. X U" with the membership function
or
L J1. from laboratory level to industrial process control.F(U) > O.5.AlIB(II)
=
II. Tong et al. U is called the universe of discourse and u represents the generic element of U. When V is discrete. a fuzzy set F is represented as F
=
max{J1. Fuzzy Sets and Terminology Let U be a collection of objects denoted generically by {u}. is called the crossover point and a fuzzy set whose support is a single point in V with J1. n B of the intersection A n B is pointwise defined for A all u E V by J1.
i= I
n
Definition 7: Fuzzy Relation: An n-ary fuzzy relation is a fuzzy set in VI X . Section II discusses the defuzzification strategies...= ((uI. are briefly reported in Section III. A
Definition 6: Cartesian Product: If A I" .F = 1./u)ju.. Thus a fuzzy set F in U may be represented as a set of ordered pairs of a generic element u and its grade of membership function: F = {(u.
Definition 8: Sup-Star Composition: If Rand S are fuzzy relations in U X V and V X W. the element u in U at which J1. Definition 3: Union: The membership function J1.

e.
R
0
S = {[ (u. Here we introduce fuzzy sets A.
(normal)
JLF(Aul+(1-A)u2)
p min(JLF(ut). Fig. algebraic product. }
[1441 y=xo
0
R
where x R is the sup-star composition of x and R.. if speed is interpreted as a linguistic variable. is a fuzzy set F in V which is normal and convex. is closely related to the forward data-driven inference which is particularly useful in the FLC.g. and x is a fuzzy set in U. very slow. Linguistic Variables and Fuzzy Sets Definition 9: Fuzzy Number: A fuzzy number F in a continuous universe V. y instead of crisp sets in the traditional logic.
No. . (GMT) The fuzzy implication inference is based on the compositional rule of inference for approximate reasoning suggested by Zadeh in 1973 [146]. Fuzzy Logic and Approximate
Reasoning
where * could be any uperatur in the class of triangular norms."
\PROCESS\
R S and is defined by
0
Fig.WE:W}
D. A'.. fast.JLF(u2
». consequence: x is A'. If the star represents the minimum operator.. which comprises four
.. The GMP. or drastic product (also see Part II [150]). then the "sup-star compositional rule of inference" asserts that the fuzzy set y in V induced by x is given by
= 1. i.5
o
40
Fig. and M is a semantic rule for associating with each value its meaning. e.
55
70
Speed
"Speed" is linand "high. then its term set T(speed) could be T(speed) = {slow.." "medium. To highlight the issues involved. we present the main ideas underlying the FLC. which rcduces to "modus poncns" when A' = A and B' = B. For example.G. there are two important fuzzy implication inference rules named the generalized modus ponens (GMP) and the generalized modus tollens (GMT): premise 1: x is A'.
The use of fuzzy sets provides a basis for a systematic way for the manipulation of vague and imprecise concepts. w) )] .1001. ANI> CYBEKl\:fTrcs. Tt x) is the term set of x.
UEV.VEV. consequence: premise
1: y is B'. A linguistic variable can be regarded either as a variable whose value is a fuzzy number or as a variable whose values arc defined in linguistic terms. C.0 medium fast
0. B. the set of names of linguistic values of x with each value being a fuzzy number defined on U. 2 shows the basic configuration of an FLC. Tt x). y is B' (GMP) premise 2: if x is A t:. Definition II: Sup-Star Compositional Rule of Inference: If R is a fuzzy relation in U X V. s~p (JL u( u . bounded product.en y is B. MAIN IDEA OF THE FLC
In this section.
JLA u)
premise 2: if x is A then y is B.U\EV. I. B' via linguistic variables x. minimum.I]. G is a syntactic rule for generating the names of values of x. V. a real line. Diagrammatic representation 01 tuzzy speeds." and "fast" as "a speed above about 70 mph. moderate.
1\11\:"1.
III. which reduces to "modus tollens" when 8' = not B and A' = not A. we can employ fuzzy sets to represent linguistic variables.
MARCH/APRIL
1990
J. especially in the realm of medical diagnosis. More specifically: see the following.
20. We might interpret "slow" as "a speed below about 40 mph. 2. max
UEU
In fuzzy logic and approximate reasoning.lspeed
slow 1. namely.406
[ITT
THANSA<TIONS
()~
SYSTEMS.2.
VOl. v)
* JL."
PROCESS OUTPUT & STATE
CONTROllED
SYSTEM
ACTUAL CONTROL NONFUZZY
I. then this definition reduces to Zadeh's compositional rule of inference
where each term in T(speed) is characterized by a fuzzy set in a universe of discourse V = [0. Definition 10: Linguistic Variables: A linguistic variable is characterized by a quintuple (x. that is. is closely related to the backward goal-driven inference which is commonly used in expert systems.
[1461. The GMT.
Basic configuration
of fuzzy logic controller
(FLC)." These terms can be characterized as fuzzy sets whose membership functions are shown in Fig. w). more or less fast.
(convex)
AE[O. guistic variable with three terms: "slow. In particular. M) in which x is the name of variable." "moderate" as "a speed close to 55 mph.
U\.( v .

in U X V. decision-making logic.. Fuzzy Conditional Statements and Fuzzy Control Rules In an FLC. and W. and an implicit sentence connective also links the rules into a rule set or.LR. there are many ways in which a may be defined.
C. " n.. in the case of twoinput-single-output (MISO) fuzzy systems. The decision making logic is the kernel of an FLC. such as "if (x is Ai and y is B.LEE:
FUZZY
l.
.) then (z is C).
consequences Since
the antecedents and the consequents of these rules are associated with fuzzy concepts (linguistic terms). which converts the range of values of output variables into corresponding universes of discourse. the dynamic behavior of a fuzzy system is characterized by a set of linguistic description rules based on expert knowledge. it has the capability of simulating human decisionmaking based on fuzzy concepts and of inferring fuzzy control actions employing fuzzy implication and the rules of inference in fuzzy logic.U{-J>ART
I
407
principal components: a fuzzification interface. Furthermore. b) defuzzification." is implemented by a fuzzy implication (fuzzy relation) R. fuzzy conIJ'-THEN
also ( RI. and z are linguistic variables representing two process state variables and one control variable.I. Basically. the overall behavior of a fuzzy system is characterized by these fuzzy relations.. R
=
(a set of conditions
are satisfied)
THEN
(a set of can he inferred). .LBi( v)
1 .(u. they are often called fuzzy conditional statements. A fuzzy control rule. Symbolically. and C. When this is the case. Operator
We are now ready to describe the main ideas undcrlying the FLC in terms of fuzzy logic. Fuzzification
B. a knowledge base. Since each fuzzy control rule is represented by a fuzzy relation. and B) U X V X W. which are used to define linguistic control rules and fuzzy data manipulation in an FLC. a fuzzy control rule is a fuzzy conditional statement in which the antecedent is a condition in its application domain and the consequent is a control action for the system under control. is a fuzzy set Ai X B.
2)
R
Il:
if x is A" and y is BII then z is C"... and a defuzzification interface.L(Ai"ndBi~ei. These fuzzy control rules are combined by using the sentence connectivcs and and also. later. the system will be referred to as a multi-input-multi-output (MIMO) fuzzy system.
R. ~ (A. The knowledge base comprises a knowledge of the application domain and the attendant control goals.Lel
w)
4)
where Ai and R.
trol rules provide a convenient way for expressing control policy and domain knowledge. and [uzzifier represents a fuzzification operator.w)
= [f. a rule base. Symbolically. a sentence
Ri"". C. and As will be seen fuzzy implication
B. x = fuzzifieri XII) where xI) is a crisp input value from a process.2. and is defined as follows:
f. y. with i = 1. fuzzy control rules have the form: RI: if x is AI and y is B. is a fuzzy implication (relation) in --> denotes a fuzzy implication function.)
where also represents
connective. The combination in question involves the sentence connective also. The structural parameters involved in the design of an FLC will be discussed at a later point. f. b) performs a scale mapping that transfers the range of values of input variables into corresponding universes of discourse..···. In other words. several linguistic variables might be involved in the antecedents and the conclusions of these rules.L Ai( u)
and f. ~f.' Rz: if x is Az and y is
B2
then z is C2. l) The fuzzification interface involves the following functions: a) measures the values of input variables.' . Sentence Connective Operators
An FLC consists of a set of fuzzy control rules which are related by the dual concepts of fuzzy implication and the sup-star compositional rule of inference.r . y.. b) the rule base characterizes the control goals and control policy of the domain experts by means of a set of linguistic control rules. a fuzzy system can be characterized by a single fuzzy relation which is the combination of the fuzzy relations in the rule set. which yields a nonfuzzy control action from an inferred fuzzy control action.v. are linguistic values of the linguistic variables x. The defuzzification interface performs the following functions: a) a scale mapping. respectively. then z is C. V. In our terminology. equivalently. and z in the universes of discourse U. For example. It consists of a "data base" and a "linguistic (fuzzy) control rule base:" a) the data base provides necessary definitions. c) performs the function of fuzzification that converts input data into suitable linguistic values which may be viewed as labels of fuzzy sets.OGIC
]'J
CONTROL
SYSTEMS:
FUZZY
LOGIC
CONTROI. A. X is a fuzzy set. R2. The expert knowledge is usually of the form
IF
A fuzzification operator has the effect of transforming crisp data into fuzzy sets.
3)
where . Ai' Bi.

1'0. the observed data are usually crisp.\NS. namely:
Zo
=
defuzzifier( z). an isosceles triangle was chosen to be the fuzzification function. we form a triangular fuzzy number which is convenient to manipulate [42]. Observed data are disturbed by random noise. In this way. the sup-star compositional rule of inference is applied z=yo(xoR) where
a
In fuzzy control applications.
MAN. i. Consequently. The strategy of fuzzification in this case is to use the concept of "hybrid numbers" [42].
FUZZIFICATION STRATEGIES
3)
Fuzzification is related to the vagueness and imprccision in a natural language.\(T[ONS
ON SYSTFMS. It is a subjective valuation which transforms a measurement into a valuation of a subjective value. a fuzzification operator should convert the probabilistic data into fuzzy numbers. I) A fuzzification operator "conceptually" converts a crisp value into a fuzzy singleton within a certain universe of discourse. Experience with the design of an FLC suggests the following principal ways of dealing with fuzzification. In this connection. Fuzzification plays an important role in dealing with uncertain information which might be objective or subjective in nature. fuzzy (possibilistic) data. completeness of fuzzy control rules. inter activity. rule base: a) choice of process state (input) variables and control (output) variables of fuzzy control rules. In this case. V. Operator
E. and some. In large scale systems and other applications. data base: a) discretization/normalization of universes of discourse. Basically. c) interpretation of the sentence connective also. decision making logic: a) definition of a fuzzy implication. Basically. d) consistency. a fuzzy singleton is a precise value and hence no fuzziness is introduced by fuzzification in this case. Since the data manipulation in an FLC is based on fuzzy set theory.
MARCil/APRil
1990
D. F. Based on the method of Dubois and Prade. while the base is twice the standard deviation of the data set. Defuzzification
The output of the inference process so far is a fuzzy set. some observations relating to the behavior of such systems are precise.
where Zo is the nonfuzzy control output and defuzzifier is the defuzzification operator. b) interpretation of the sentence connective and. y and the fuzzy relation R.40H
JUT
TR.
DATA
BASE
The knowledge base of an FLC is comprised of two components. one must defuzzify the fuzzy control action (output) inferred from the fuzzy control algorithm.
is the sup-star composition. Compositional
Operator
To infer the output z from the given process states .
ANI) CYBDC"-. c) types of fuzzy control rules." require both probabilistic and possibilistic modes of characterization.. In [76].lI'TICS. the histogram of the measured data may be used to estimate the membership function for the transformation of probability into possibility [17]. namely. b) source and derivation of fuzzy control rules. a data base and a fuzzy control rule base. c) completeness. IV. d) choice of the membership function of a primary fuzzy set. while others are measurable only in a statistical sense.
2. In this way. specifying a possibility distribution of control action. a nonfuzzy (crisp) control action is usually required. The vertex of this triangle corresponds to the mean value of a data set. it should be noted that Dubois and Prade [20] defined a bijective transformation which transforms a probability measure into a possibility measure by using the concept of the degree of necessity. fuzzification is necessary during an earlier stage. We shall address some issues relating to the data
. which involve both uncertainty (fuzzy numbers) and randomness (random numbers).ix) equal to zero except at the point xo' at which /-L)xo) equals one. is the added probability of elementary events in E over the probability assigned to the most frequent elementary event outside of E. the necessity of an event. The use of hybrid number arithmetic in the design of an FLC suggests a promising direction that is in need of further exploration.
VOL
20. In the on-line control. referred to as "hybrids. It interprets an input Xo as a fuzzy set A with the membership function J1. d) definitions of a compositional operator. e) inference mechanism. Design Parameters of the FLC The principal following: 1) 2) design parameters for an FLC arc the
2)
3)
4)
5)
fuzzification strategies and the interpretation of a fuzzification operator (fuzzifier). defuzzification strategies and the interpretation of a defuzzification operator (defuzzifier).e.. E.r . and hence it could be defined as a mapping from an observed input space to fuzzy sets in certain input universes of discourse. b) fuzzy partition of the input and output spaces. This strategy has been widely used in fuzzy control applications since it is natural and easy to implement. computational efficiency is enhanced since fuzzy numbers are much easier to manipulate than random variables.

In effect.0 0.0 0. represents an appropriate scaling mapping. The choice of quantization levels reflects some a priori knowledge.0 0. XII
-0.0 0.0 0. A look-up table based on discrete universes.0 0.0 0.
K2ie(k).0 OJ) 0.LEE: FUZZY
l. One example is shown in Table III.7 1.4 -0. These concepts are subjectively defined and based on experience and engineering judgment.0. and change of error (de) is C. In the case of an FLC with continuous universes.0 0.0 0. + 4.0 0.0 0. Fuzzy Partition of Input and Output Spaces A linguistic variable in the antecedent of a fuzzy control rule forms a fuzzy input space with respect to a certain universe of discourse. coarse resolution could be used for large errors and fine resolution for small errors.7 0. B.0 0.3
PM
0.7 1.0
PS
0.0 0. Discretization / Normalization of Universes of Discourse
For the purpose of discretization.0
-0 -5 -4 -3
-2 -I 0 I 2 3 4
-3.R < XII '" < Xo '" < XII '" XII <. we need a scale mapping. The scale mapping can be uniform. In this relation. a discrete universe may be formed by a discretization of the continuous universe. nonuniform.3 0.1 < +0. it should be noted that the correct choice of the membership functions of a term set plays an essential role in the success of an application.0 0. 2) Normalization of a Universe of Discourse: The normalization of a universe requires a discretization of the universe of discourse into a finite number of segments. we shall discuss some of the important aspects relating to the construction of the data base in an FLC. or both. can be implemented by off-line processing in order to shorten the running time of the controller [90). The concepts associated with a data base are used to characterize fuzzy control rules and fuzzy data manipulation in an FLC. as will be seen at a later point in this section.o<xlI".0 0.4 + 0.K3de(k)j.2
base in this section and to the rule base in the next section.6 1.0 0.7 0.0 0.0 0. In this setting a fuzzy set is then defined by assigning an explicit function to its membership function.0 0.2 + 0.4 -0.0 0. A. in which as a special case F is a linear function of its arguments. In what follows.: if error (e) is A.0 0.7 1..7 0..0 0.0
PM
0.2 -0.2 < XII '" -l.8 + 1. or both. quantization discretizes a universe into a certain number of segments (quantization levels).0 0.0 0.0 0.3 0.0 0.0 0. Thus.0 0. In general.0 0.0 0. then output is D.0 0.0 0.3 0.2". A simple instance of an FLC can be represented K4[U(k)]
=
by
The representation of uncertain information with fuzzy sets brings up the problem of quantifying such information for digital computer processing. [. Each segment is labeled as a generic element.0 0. is transformed into the normalized closed interval [-1.. For example.2 < XII <.: + 3.0 0.0 0. The mapping can be uniform (linear).7 1.0 0.0 0.3 0.7 1.0 0. + I). A fuzzy set is then defined by assigning grade of membership values to each generic element of the new discrete universe.1 <XII'" +0.6
3.2.0 0. a continuous universe may be normalized. which defines the output of a controller for all possible combinations of the input signals.:
NB
-1.2 -0.-O.3 0.0 0.0 0. the representation depends on the nature of the universe of discourse.3.: +0. i = 1.3 0.0
NS
0.0 0. nonuniform (nonlinear).3 0. where the universe of discourse.0 0.1 +0.0 0. we may have control rules of the form: R.0 0.0 0.7 0.2
5
6
-0. The choice of quantization levels has an essential influence on how fine a control can be obtained.6. If the universe is continuous.0 0. Furthermore. the peformance of an FLC is less sensitive to small deviations in the values of the process state variables. fTZZY
LOlil(
l"UNTRUI. with each segment mapped into a suitable segment of the normalized universe.4 < XII'" + 0.8 < XII '" + 1.
where F denotes the fuzzy relation defined by the rule base and K. An example of discretization is shown in Table II.0 0. For example.0 0.
F[ K1e(k)..7 0.3 0.0 0.0 0.OCiIC IN C01\TROL
SYSTEMS.LI:f{-P/\KT
I
4m
II
USING A NUMERICAL DEFINITION
TABLE
QUANTIZATION AND PRIMARY FUZZY SETS
Level No. In this connection. The normalization of a continuous universe also involves a pnon knowledge of the input/ output space.0 0.3 0.0 0. sum of errors (ie) is B.0 0.3 0.6 < SII <. the number of quantization levels should be large enough to provide an adequate approximation and yet be small to save memory storage. in a three-input-one-output fuzzy system. where a universe of discourse is discretized into 13 levels with seven terms (primary fuzzy sets) defined on it.0 0. and forms a discrete universe.3 0.0
ZE
0.0 0. 1) Discretization of a Universe of Discourse: Discretization of a universe of discourse is frequently referred to as quantization.0
NM
0.8 -0. In general. [105). we see an analogy to the parameters of a conventional PID controller [63). while that in the consequent
. if a universe is quantized for every five units of measurement instead of ten units.0 0. A universe of discourse in an FLC is either discrete or continuous.1 +0.0 0. then the controller is twice as sensitive to the observed variables.7 0.5).
Range
xo'" -3. due to discretization. which serves to transform measured variables into values in the discretized universe.7 1.0 0.4.

. the cardinality of a term set in a fuzzy input space determines the maximum number of fuzzy control rules that we can construct. 1) Numerical Definition: In this case. a fuzzy control algorithm should always be able to infer a proper control action for every state of process.5. or both.0. The primary fuzzy sets (linguistic terms) usually have a meaning.5]
-0. a dominant rule always exists and is associated with the degree of belief greater than 0. a linguistic variable is associated with a term set.x) and T(y) are 3 and 7.4 0.2. .5] [+ 2. depicting two fuzzy partitions in the same normalized universe [.0] [-1.0
.0
[-4. and P. (b) Finer fuzzy partition with seven terms: NB. In the case of two-input-one-output fuzzy systems. say 0.2.0. negative.s
N
. zero.
should input/ unique usually
C. +0. An additional rule is added whenever a fuzzy condition is not included in the rule base. If this is not the case.
·1 +1
(b)
Fig.9.1]
u. zero. Since a normalized universe implies the knowledge of the input/output space via appropriate scale mappings. implying that we have a strong belief in the positive sense of the fuzzy control rules which arc associated with the FLC.
D.3. The number of primary fuzzy sets determines the granularity of the control obtainable with an FLC. Diagrammatic representation of fuzzy partition. or whenever the degree of partial match between some inputs and the predefined fuzzy conditions is lower than some level.2 0. a well-formed term set can be achieved as shown. In this sense.0 1-0. + 1. The latter indicates that no dominant rule will be fired. A fuzzy partition.2] [-2. we choose the level E at the crossover point as shown in Fig.0. negative small.+ 1. + 0.
M/\RCII/APRII. NM: negative medium.3] [.
+1
respectively.5.0 0.5. It be noted that the fuzzy partition of the fuzzy output space is not deterministic and has no solution. + 1). A typical example is shown in Fig. the membership function of each primary
. the terms could be asymmetrical and unevenly distributed in the universe. The property of completeness is incorporated into fuzzy control rules through design experience and engineering knowledge. PS. 2) Rule Base Strategy: The rule base strategy has to do with the fuzzy control rules themselves. The union of these supports should cover the related universe of discourse in relation to some level set E. the maximum rule number is 3 x 7.0.0] [+ 1.
NO. positive. In this case. 3. with each term in the term set defined on the same universe of discourse. Furthermore. the grade of membership function of a fuzzy set is represented as a vector of numbers whose dimension depends on the degree of discretization. Membership functions having the forms of triangle-shaped and trapezoid-shaped functions are used here. (a) Coarse fuzzy partition with three terms: N. determines how many terms should exist in a term set.
MAN. 1) Data Base Strategy: The data base strategy is concerned with the supports on which primary fuzzy sets are defined. In general. A heuristic cut and trial procedure is needed to find the optimal fuzzy partition.2 0. In general.5 1.5. -0. depending on whether the universe of discourse is discrete or continuous: a) numerical and b) functional.2] [+0.
1
·1
xl><
coarse
(a)
I.5] [. The former shows that no control action will result. -0. positive small. PS: positive small. ZE.
of the rule forms a fuzzy output space.2 0. PM. This property is called "completeness. such as NB: negative big.0.0]
-1.
finer
.
"i Primary Fuzzy Sets
0. and PB.+2.410
irr-r:
TRANSACTIONS
ON SYSTEMS. if the cardinalities of Tt.2 0.1.6. negative medium.0. NM.0.
1990
TABLE III
NORMALIZATION AND PRIMARY FUZZY SETS
USING"
FUNCTIONAL
DEFINITION
Normalized Universe
Normalized Segments
Range
[-6. two dominant rules are activated with equal belief 0.1.O. rule base. This is equivalent to finding the number of primary fuzzy sets. ZE: zero. -4.0.4 NB NM NM ZE PS PM PB
[-1. PM: positive medium. ZE. NS: negative small.
2. positive big.5.2 0. This property of an FLC is called s-completeness.
AND
CYBFRNETH'S.•
20.6] [+ 0.2 0. or a nonnonnalized universe is used.~'. + 1. + 4. An illustrative example is shown in Table II.0. and PB: positive big. then. Completeness
Intuitively.0] [ . positive medium. Membership Function of a Primary Fuzzy Set There are two methods used for defining fuzzy sets.0. 3. In the extreme case. -2." The completeness of an FLC relates to its data base. NS.2
D.
VOl . negative big. 3.

0.LEE:
FUZZY
LO(ilC
Il'.1. the selection of the linguistic variables has a substantial effect on the performance of an FLC. For optimized performance. Many experts have found that fuzzy control rules provide a convenient way to express their domain knowledge. 4. and it seems likely that a combination of them would be necessary to construct an effective method for the derivation of fuzzy control rules. Either a numerical definition or functional definition may be used to assign the grades of membership to the primary fuzzy sets.0. Another approach includes an interrogation of experienced experts or operators using a carefully organized questionnaire. This explains why most FLCs are based on the knowledge and experience which are expressed in the language of fuzzy if-then rules [64]. Typically. a human operator employs-consciously or subconsciously-a set of fuzzy
2) Functional Definition: A functional definition expresses the membership function of a fuzzy set in a functional form. This issue is discussed in greater detail in [58]. Seen in this perspective. the linguistic variables in an FLC are the state. and properties of consistency. B. which affects the robustness of an FLC. This raises the issue of the fuzziness or. 2) Based on Operator's Control Actions: In many industrial man-machine control systems. In this section.3. state error.
RULE BASE
A fuzzy system is characterized by a set of linguistic statements based on expert knowledge. typically a bell-shaped function.
fuzzy set has the form of /J.0. as we mentioned before. source and derivation. etc. as reported in [103]. the specificity of a membership function. the input-output relations are not known with sufficient precision to make it possible to employ classical control theory for modeling and simulation.' CONTROl
S't'STfMS:
FUZZY
(_()(IIC
CO:-JTROLLER-P. Table III and Fig.7. [118].
. [59]. [58]. [80]. The formulation of fuzzy control rules can be achieved by means of two heuristic approaches. triangleshaped function. we shall examine the following topics related to fuzzy control rules: choice of process state (input) variables and control (output) variables. [47]. [82]. it should be noted that in our daily life most of the information on which our decisions are based is linguistic rather than numerical in nature. In this connection. And yet skilled human operators can control such systems quite successfully without having any quantitative models in mind. Choice of Process Stale Variables and Control Variables of Fuzzy Control Rules Fuzzy control rules are more conveniently formulated in linguistic rather than numerical terms. In particular. As was stated earlier. [4].0. experience and engineering knowledge play an important role during this selection stage.3]. VI. 1) Expert Experience and Control Engineering Knowledge: Fuzzy control rules have the form of fuzzy conditional statements that relate the state variables in the antecedent and process control variables in the consequents. The functional definition can readily be adapted to a change in the normalization of a universe.\RT
r
411
A. fuzzy control rules provide a natural framework for the characterization of human behavior and decisions analysis. more accurately. trapezoid-shaped function. etc. Such functions are used in FLC because they lead themselves to manipulation through the usc of fuzzy arithmetic. The collection of fuzzy control rules that are expressed as fuzzy conditional statements forms the rule base or the rule set of an FLC. the membership functions should be sufficiently wide to reduce the sensitivity to noise. In this manner. The proper choice of process state variables and control variables is essential to the characterization of the operation of a fuzzy system. [50]. state error derivative. state error integral. In particular. types of fuzzy control rules. interactivity. The expert knowledge is usually in the form of "if-then" rules. [127]. which are easily implemented by fuzzy conditional statements in fuzzy logic. [113].
Fig. In effect. we can form a prototype of fuzzy control rules for a practicular application domain. 4 show an example of a functional definition expressed as:
Note that if the normalized universe is changed. The most common one involves an introspective verbalization of human expertise. Furthermore.f(U) where a
= =
i~ I
L ai/u"
5
[0. Source and Derivation of Fuzzy Control Rules There are four modes of derivation of fuzzy control rules. The choice of grades of membership is based on the subjective criteria of the decision. the use of cut and trial procedures is usually a necessity. A typical example of such verbalization is the operating manual for a cement kiln. These four modes are not mutually exclusive. and completeness. the choice of linguistic variables and their membership function have a strong influence on the linguistic structure of an FLC. if the measurable data might be disturbed by noise.7.
Example
of functional
definition
of primary
fuzzy sets. the parameters U f' (If should be changed accordingly. justification.

The rule justification is done by referring to a closed system trajectory in a phase plane. but few are focused on human learning. the ability to create fuzzy control rules and to modify them based on experience. and has the effect of shortening the rise time. with a common term {negative. [101]. [94]. 5 shows the system response of a process to be controlled. [95]. namely. 3) Based on the Fuzzy Model of a Process: In the linguistic approach. A knowledge of parameter-adjusting based on phase plane analysis (e. The output is the change of the process input (CI). The prototype of fuzzy control rules is tabulated in Table IV and a justification of fuzzy control rules is added in Table V. The SOC has a hierarchical structure which consists of two rule bases. [106]. and provides a more tractable structure for dealing theoretically with the FLC.
Rule justification
by using phase plane. [103]. Procyk and Mamdani [87] described the first selforganizing controller (SOC). the linguistic description of the dynamic characteristics of a controlled process may be viewed as a fuzzy model of the process. zero. [6]..
MAHCllj
APRil. Fig.. 5. The set of fuzzy control rules forms the rule base of an FLC. Sugcno's fuzzy car can be trained to park by itself. AND CYBFRNFTICS. [99]. rise time) and an intuitive feel for the behavior of the closed loop system are required. it yields better performance and reliability. The principle of global rule modification in symmetry and monotonicity is also employed. such rules can be deduced from the observation of human controller's actions in terms of the input-output operating data [97]-[99].2. if (E is negative and DE is negative) then CI is negative. The first is a heuristic method in which a collection of fuzzy control rules is formed by analyzing the behavior of a controlled process. However. For example.
VOL
20. further studies relating to the SOC have been carried out at Queen Mary College and elsewhere [60]. 3. The second one is constructed by "meta-rules" which exhibit human-Iikc learning ability to create and modify the general rule base based on the desired overall performance of the system.g. [102]. The control rules are derived in such a way that the deviation
from a desired state can be corrected and the control objective can be achieved. Recently.-112
IFrT
TI-L\~S/\CTI()~S
ON SYSTEMS. In practice.
No. the convergence of the prescriptive method requires a careful analysis. We assume that the term sets of input/output variables have the same cardinality. King and Mamdani [49] introduced another useful method for rule justification. The first one is the general rule base of an FLC. MAN.. Although this approach is somewhat more complicated. it is expedient to express the operator's control rules as fuzzy if-then rules employing linguistic variables.
1990
DE
decrease
oWKIhoof
iii
. positive}.
. The second approach is basically a deterministic method which can systematically determine the linguistic structure and/or parameters of the fuzzy control rules that satisfy the control objectives and constraints [111]. Justification of Fuzzy Control Rules There are two principal approaches to the derivation of fuzzy control rules. Mamdani [1] proposed a prescriptive algorithm for deriving the "best" control rules by restricting system responses to a "prescriptive fuzzy band" which is specified by fuzzy control rules. Ri:
Ri/:
if(E is positive and DE is negative) then CI is positive. Based on the fuzzy model. However. [49]. to automate such processes. [104]. [7].
II
vi
viiviiiix
x xi
E
DE
+
time
(a)
(b)
Fig. C. (3) Phase-plane
trajectory. Several methods of adjustment of rule selection have been studied [1]. More specifically. A very interesting example of a fuzzy rule-based system which has a learning capability is Sugeno's fuzzy car [97]. The corresponding rule of region i can be formulated as rule R. The derivation is purely heuristic in nature and relies on the qualitative knowledge of process behavior. As was pointed out by Sugeno. where the input variables of the FLC are the error (E) and error derivative (DE). A brief review of these results is given in the following. overshoot. 4) Based on Learning: Many FLCs have been built to emulate human decision-making behavior. Rule Rii for region ii decreases the overshoot of the system's response. we can generate a set of fuzzy control rules for attaining optimal performance of a dynamic system.
if-then rules to control the process.. this approach to the design of an FLC has not as yet been fully developed. So-called "scale mappings" should be adjusted first so that the system trajectory can terminate on a desired state.
(h) System step response.

Braae and Rutherford assumed that the fuzzy control rules of an open system (process) and a desired closed-loop system were initially given.
8
9 10 11
P N
N P
Z Z
i (rise time).ZE.\RT
r
413
decrease ollershoot
TABLE
PROTOTYPE OF F1.NM. The main idea is to invert the low order linguistic model of a certain open loop system." The main idea is that scale mappings should be adjusted first to yield approximately a desired trajectory behavior.0(ilC
CONTROLLFR-I'. [101]. Rule justification by using a linguistic tic trajectory with initial rules." As in conventional identification. which is somewhat heuristic and subjective.
An additional advantage of this approach is that the measurement noise appearing in the linguistic phase plane is less of a problem than that in the nonlinguistic phase plane. (From Braae and Rutherford [7]. An approach to generating the rule base of an FLC.
POSITIVE) CI Reference Point
E N
Z
DE
Z N Z
NBr::M NB
NS ZE ZE ZE ZE
ZE ZE
ZE ZE
P
5
Z Z
P
Z
P N N P
Z
a. [103]. PB)
Rule No. or both.ZE.PM. for example. However. and used a correlation analysis of the LE to determine the linguistic structure. linguistic inversion mappings are usually incomplete or multivalued.
ZERO. However. PM}
Reference Range i (rise time) i (overshoot> iii iii
DE NS NB PS PB NS PS
CI
15
16 17
I~
19
PR PS NB NS PS NS
PM NM NM PM
ZE ZE
ix
xi
Better control performance can be obtained by using finer fuzzy partitioned subspaccs.g. 6. NM. The systematic rule justification has recently been proposed and studied by means of fuzzy relational equations [13]. The studies in question deal with one.k d.j c. The purpose is to synthesize a linguistic control element (FLC) based on the fuzzy models described above.f. Tong [Ill] introduced the concept of "logical examination" (LE) for converting process input-output data into a set of fuzzy control rules. is described in [6]. Rule No. v ii (overshoot). (b) Linguistic rules. This method is restricted to relative low order systems but it provides an explicit solution for rule generation of the FLC.
S"STEMS:
FUZZY
1.h.viii
ix
xi
VI
RULES WITH TEnM SETS
PROTOTYPE
OF
Fuzzy
CONTROL
(NB. The basic notion of these two approaches is so-called "fuzzy identification. is necessary to complete the inverse mapping which has a reasonable singled-valued solution. Takagi and Sugeno [103] proposed a fuzzy identification algorithm for modeling human operator's control
. Rule 6 7 No. The prototype and the justification of fuzzy control rules are also given in Table VI and Table VII.PM. assuming that fuzzy models of the open and closed systems are available. PM. structure identification and parameter estimation. This can be inferred from the linguistic trajectories.i b.t set point
E
TABLE
RULE JUSTIFICATION
V
WITH TERM SETS
IE
(a)
decresse rIse lIme
IE
(b) phase plane.
IV
RULES WITII TERM SETS
(NeGATiVe. the fuzzy identification comprises two phases. PS. [84]. PS.}ZZY CONTROl. 6. Tong tackled both identification problems simultaneously. vi
iii. namely. [125] and linguistic control rules [Ill].
POSITIVE) CI Reference Range
E P N N
p
DE N N P P N P
TABLE
P
N
Fig. vii
iv. PS.e. it is still somewhat heuristic and subjective. and encounters difficulties in the identification of rnultivariable fuzzy systems. which is analogous to the conventional controller design by pole placement. I 2 3 4
E PB PM PS ZE ZE ZE NB NM NS ZE
ZF
DE ZE ZE ZE NB NM NS ZE ZE ZE PB PM PS ZE
CI
Reference a e i b f
Point
5
6 7
~
9 10 II 12 13
ZE ZE
PB PM PS NB NM NS NB NM NS PB PM PS ZE
c g k d h I set point
TABLE VII RULE JUSTIFICATION WITH TERM SETS {NB. So. This approximation has substantial effect on "linguistic substitution" which further determines a fuzzy controller. A slightly modified method was suggested in [7].
ZERO. [104]. ZE.'-FE:
FUZZY
10Gle
IN COl'\"TRol. NS. An example is shown in Fig. with the term set {NB. PB}. (a) Linguistrajectory with modified
(NEGATIVE. Then rule modification can be accomplished by using the linguistic trajectory behavior to optimize the system response in the linguistic phase plane. 14 E
NS. Rule I 2 3 4 No. an "approximate" strategy. [15].NM. NS. It tracked the linguistic trajectory of a closed loop system in a "linguistic phase plane.

when a control command u is chosen to be Ci.
and y is B. At issue is the problem of structure identification.y. In this way. depend on trial-and-error. Ai" .: . with i = 1. " y.' .») then u is Ci•
("I
then z is Cz
R." A typical rule is described as R
j:
if (u is C.···. " n. the problem is reduced to parameter estimation.' . Sugeno has successfully applied this method to the design of an FLC for navigating a model car through a crank-shaped curve [98] and for parking a model car in a garage [97]." The most likely control rule is selected through predicting the results (x. and the choice of the membership functions of primary fuzzy sets.2. [136].'"
. and z in the universes of discourse U. y.r
THr\NS. [139] as well as to automatic container crane operation systems [137]-[139]. The structure identification requires the determination of the system order and time delays of discrete-time fuzzy models. D.: if x is A". then z
=
f. state error.' . y. state integral) at time t and compute a fuzzy control action at time t as a function of (x. R2: if x is A2.2..\ClloNS
ON SYSTEMS.V.g. are linguistic terms of the linguistic variables x. nature. a suitable linguistic structure is easy to find since one can observe and/or ask for the kind of information which the operator needs.[125] for further details.
i." '. are currently in use in the Fuzzy Control Rules: Most FLC's fuzzy control rules which. the consequent is represented as a function of the process state variables x. i= 1.414
rrr.
""IAN.. the rule is interpreted as: "if the performance index x is Ai and index y is B. 1) State Evaluation have state evaluation case of MISO systems. from the input-output data of the system. " Bi. are referred to as "object evaluation fuzzy control rules. and z are linguistic variables representing the process state variables and the control variable. and z in the universes of discourse U." The rules in question. a typical control rule is if the control notch is not changed and if the train stops in the predetermined alluwance zone.2.y) and the control rules in the rule set.
and y is B. which are derived from skilled operator's experience.···. " V." evaluate the process state (e. two types of fuzzy control fuzzy control rules and object evalrules. The reader is referred to [13]. Further research on this problem has been reported by Sugeno and Kang in [101]. . which is done by optimizing a least-square performance index via a weighted linear regression method. It is called "object evaluation fuzzy control. rules of the form RI: if x is Ap"'. which is partly addressed in this paper.e. Types of Fuzzy Control Rules Depending on their rules. respectively. By contrast. and z are linguistic variables representing the process state variables and the control variable. Sugeno's method provides a more systematic approach to the design of an FLC.··
'." '. MARCHI APRIL
1990
actions.···." In automatic train operation. " B. However.: ..
VOL
20. Another approach based on fuzzy relational equations is directed at the same problems. state. such as the ehoice of process state variables. In this case. No. and C. In such circumstances.v)
where x. and Ihara [135] proposed another algorithm which predicts present and future control actions and evaluates control objectives. [99]. .[15].y)."'. The fuzzy control rules to be identified have the form of
Ri: if
x is
Ai. then z is Cn
A control command is inferred from an objective evaluation of a fuzzy control result that satisfies the desired states and objectives.(x. However. respectively.n. while the parameter estimation reduces to the determination of the overall fuzzy relation matrix. some steps of this algorithm. then this rule is selected and the control command Ci is taken to be the output of the controller. Takagi and Sugeno [104] improved their algorithm so that parameter estimation can he fully implemented. y) corresponding to every control command Ci• In linguistic terms.: . 2) Object Evaluation Fuzzy Control Rules: Yasunobu. in the arc characterized as a collection of and y is BI then z is and y is
B2
where x." ·. which are referred to as "state evaluation fuzzy control rules. are the linguistic values of the linguistic variables x. state evaluation uation fuzzy control design of the FLC. and z is a logical function of the process state variables such as a linear function of x.
->
(x is Ai and y is B. the predictive fuzzy control provides a mechanism for evaluation so that the desired states and control objectives can be achieved more easily. the fuzzy partition of input spaces. It is well known that systems control encounters difficulties in satisfying multiple performance indices simultaneously and in achieving accurate control in the presence of disturbances.
R
j:
if x is Ai. It should be noted that predictive control has been successfully applied to automatic train operation [135].' . and x. Ai" ." or "predictive fuzzy control. such as process state variables.. and the experimental results are quite remarkable. and W.. then the control notch is not changed.'[) CYBFRNrTH'S. " y.
and y is B.[84].
Fuzzy control rules of this type.y.. Al\. A control command u takes a crisp set as a value. Miyamoto. and W. . the state evaluation fuzzy control does not evaluate the computed control actions as human operators do. then z
=
Mx. taking values such as "good" or "bad." '. yare performance indices for the evaluation of the ith rule. Tests have shown that this type of control is capable of
. Recently. fuzzy control provides an effective framework for solution. In a more general version. The inference mechanism of this FLC will be discussed in Part II [150].

.n. 35. This may happen as a consequence of interaction between the rules. Usluel. e. vol. 47-65. Y. IS. Man.
2) Number of Fuzzy Control Rules: There is no general procedure for deciding on the optimal number of fuzzy control rules since a number of factors are involved in the decision. F. Expert Systems Knowledge. 1987.
. Pedrycz. Man Cybem. "Modelling of a fuzzy controller with application to the control of biological processes. A. [15] __ " "Fuzzy rule generation for fuzzy control. Newell. 26. The reported research in [12]. The problem of interaction is complex and not as yet well understood. pp.• vol. and C. Mach. 9.1(' CONTROLLER-PART
I
415 [5] M. If an input Xo is A. [26] S. 1982.[12]. J." Int. "Use of rule-based system for process control." in Proc. [20] __ . [16] E. 217-232. 13. 1985. Tanaka. pp. depending on the definition of fuzzy implication and sup-star composition. Baaklini and E. 215223. and J. Kandel. Trojan. Cologne: Verlag TUV Rheinland. [19]. 255-273." Fuzzy Sets Sys:. 13-22. vol. vol. Pedrycz." Fuzzy Sets Syst..thenzisC. [6] "Theoretical and linguistic aspects of the fuzzy logic controller." Int. Joint Automatic Control Conf. [10] __ . [101]. no 5. and J. 145-158. The consistency of rules may be improved through the use of the concept of a fuzzy clustering of fuzzy control rules..:ifxisA. pp. pp. pp. no." Kybernetes. W. pp. 1986. [18] D. 73-83. [19] indicates that interactivity of rules can be controlled by the choice of fuzzy implication and sup-star composition. Graham and R. "Modeling controllers using fuzzy relations. vol. W. "On identification in fuzzy systems and its applications in control problems. Rawlik. 130] M.• vol. pp. [18]." Fuzzy Sets Syst. Theory Adv.. 1988. 3. Studies.. Czogala and W. 4) Interactivity of Fuzzy Control Rules: Assuming that a collection of fuzzy control rules has the form R. Baldwin and N. 31.. M. 1-68.
REFERENCES
[1] N. Arai. Gupta." Cont . I. "Moving mark recognition and moving object manipulation in fuzzy controlled robot. Baldwin and B. Mamdani. 7. Pilsworth. it is important to check on the consistency of fuzzy control rules in order to minimize the possibility of contradiction. [26]. vol. 3. 1988.. Ltd . 43-59. "Unfair coins and necessity measures: Toward a possibilistic interpretation of histograms.
E. Berkeley. 26. pp. J. pp. Tomizuka of the University of California. pp. [23] F. Isik. 11. 193-219. vol. 1979. we would expect that the control action z is C. J.. 3-13. Probabilistic Sets in Decision Making and Control. [31] M. pp. "Section of parameters for a fuzzy logic controller. 576-582. Fei. 243-273. 7. pp. Technol." Fuzzy Sets Syst. and the reviewers for their helpful comments and suggestions. i=I. [14] __ ..1988. M. 275-293. Amsterdam: North-Holland. 1980. vol. "Axiomatic approach to implication for approximate reasoning with fuzzy logic. 393-405. 1980. Mizumoto. [18]. M. [17] B. no. 9.. 19~2. Bandler. "Prescriptive methods for deriving control policy in a fuzzy-logic controller. Braae and D. vol. "Some problems concerning the construction of algorithms of decision-making in fuzzy systems. Ed.. "The fuzzy decade: A bibliography of fuzzy systems and closely related topics. Syst. Gottwald and W. Syst." Fuzzy Sets Syst.. Syst. Possibility Theory: An Approach to Computerized Processing of Uncertainty. San Francisco. A. pp. G. 1975. Syst. K." IEEE Trans. Efstathiou.. 1980. 1988. "Estimation of fuzzy memberships from histograms. 31. 1984. [3] J. 1983. Ciliz. pp. Dubois and H. Prade. Kohout. the control action z may be a subset or a superset of C. New York: Plenum Press.g. and K. 201-221.• vol." Fuzzy Sets Syst. Czogala.." Fuzzy Sets Syst. L. 133] K." Fuzzy Sets Syst." in Approximate Reasoning in Expert Systems. In this connection. Kiszka. 185-188. Studies. 1989. Probabilistic Sets: Fuzzy and Stochastic Approach to Decision.. B. Interactivity." in Approximate Reasoning in Intelligent Systems. [4] J. pp." Electron." in Approximate Reasoning in Expert Systems.LEE:
FUZZY
LOGIC IN CONTROL
SYSTEMS: FUZZY
1-0(. 2. pp. M. Eds. M. Bandler. The assistance of Professor Zadeh is gratefully acknowledged. The author would like to thank Professor M. 1985. Mag. V. pp. vol. the rules may be subjected to different performance criteria. M." Cybem. Hirota. efficiency of computation. 1986. New York: Pergamon.1989. 1985.. 3. 1984. and the choice of linguistic variables. 15. "FLEX·~800 series elevator group control system. Graham and P. [12]. W. human operator behavior. [19] _~. "Fuzzy adaptive control of a first-order process. no." Kybemetes. 10. and J. vol. Pedrycz. 3-4. Zadeh. Japan." Fuzzy Sets Syst." in Proc. Cologne: Verlag TUV Rheinland. 1978. Amsterdam: North-Holland. Tsukamoto. 1-14. "Fuzzy logic controllersA perspective. no. 15-20. B. A. 5. P." Cybem. "Probabilistic fuzzy controller as a generalization of the concept of fuzzy controller. [28] _~. no I. Aug.• Osaka. no." Inform. Gupta and Y. Sanchez and L. pp. 3. Gaines and L. 3. no 4. Zadeh of the University of California. 1981. Lett. pp. 553-577. "Fuzzy relations in a control setting. M. pp. Control and Recognition Processes. Gupta." Fuzzy Sets Syst.. and S. S. B. "Pratical aspects of the knowledge-based control of a mobile robot motion. Sarma. Kandel. pp. Berkeley. E." Fuzzy Sets Syst. In fact. on Circuits and Systems. 1988. Decision and Control. Kiszka. [8] K. [2] J. [27] B. no. Kiszka." Automat. "The generalized modus ponens under sup-min cornposition-A theoretical study. R. 1986. 4. H. Rutherford. pp. vol. [85]. for his encouragement of this research. Properties of Consistency. M. [32] K. "Fuzzy identification and control of a liquid level rig. A. [13] __ . [25] B. "Some considerations of fuzzy conditional inference. [22] J. "Problems of the design of fuzzy controllers. Syracuse. 10. vol. vol. [85]. pp. it should be noted that Sugeno's reasoning and identification algorithm provides an alternative solution to these problems [104]. 1980. [64]. 30th Midwest Symp. Fukarni. Bharathi Devi and V. pp. [9] E. vol 3. 1987. pp. vol. 1981. Mach.. vol. 625-626. 257-274. FAIO·C." IEEE Contr . 6. no. vol. [12] E. Jones. 15. 1985. "Fuzzy logics and the generalized modus ponens revisited...
operating trains and cranes as skillfully as an experienced operator. Hachisu. "Analysis and synthesis of fuzzy systems by the use of fuzzy sets. J. Eds. [24] S. performance of the controller. S. 399-418. 185-199. In practice. "Control problems in fuzzy systems.. [7] __ . Sci . 223-229. [11] Eo Czogala and K. New York: Chaman and Hall.
ACKNOWLEDGMENT
am greatly indebted to Professor Lotfi A. vol.. I. "Controllability of fuzzy control systems. [21] _~. and Completeness 1) Completeness:
Please refer to Section V of this pa-
per. B. "Rule-based process control using fuzzy logic. Hirota and W. 1979. Fujitec. pp. 2. Gupta. Hirota. vol. 1988.···. NY. [29] I. SMC-16. pp. 1977. Bernard. Man. [26]. 3) Consistency of Fuzzy Control Rules: If the derivation of fuzzy control rules is based on the human operator experience." Fujitec Co. Uncertainty and Decision. Guild. vol. A. no. Czogala and T.

M. Boston: D. pp. Satoh. J." in Industrial Applications of Fuzzy Control. New Yurko Academic. "An experiment in linguistic synthesis with a fuzzy logic controller. K. 1981. Fuzzy Reasoning and if> Applications. "The application of fuzzy control systems to industrial processes. vol. 63-72. Larsen. "Energctistic stability of fuzzy dynamic systems. vol. pp. Kandel. K." IEEF Trans. Mamdani. 153-167. __ . and D. S. Furtv Logic in Knowledge Engineering. pp. pp. ISXS-1SSH. 105-124. 2nd IFSA Congress. Kickert. 10. NJ: Prentice Hall. 4. "A fuzzy logic controller for aircraft flight control." Fuzzy Sets Syst. 1985.D. vol. "Applications of fuzzy set theory to control systems: a survey. 1985. R. Fukuzaki. SMC-7. 1979. Japan. pp. Miyake. R. Kochanska. 34. Murakami. 1YH6. no. London: Academic. "A fuzzy logic controller for a traffic junction. Y. M. "Application of a fuzzy controller in a warm water plant. thesis. Mamdani and S.. S. vol. no." IEEE Trans. W. Li and C. "Approximate reconstruction of mapping functions from linguistic description in problems of fuzzy logic applied to system control. 8. I. Amsterdam: North-Holland. 1988." Cybern. F. pp. Nakayama. 12. S. 1985. Int. Negoita. vol. 1. Kiszka. Procyk and E.. J. Ins.
""O_
2. 65-72. V. A. "Applications of fuzzy relational equations for methods of reasoning in presence of fuzzy data. 2-7. 1977. 1978. A linguistic self-organizing process controller. Kaufmann. Mamdani .... __ . A. of the IFAC Symp. vol. pp. M. 1981. Reidel Publishing Company. vol." Fuzzy Sets Sl'sl . 669-678." in Proc. 29-44. Eds. Amsterdam: North-Holland. Fuzzy Sets Theory and Applications. __ . O. 04. and M." in Fuzzy Automata and Decision Processes. 6. __ . Ed. 67-69. 1984. 345-349. 77-88.. pp. M. 1. 19HI. "Optimal control of a finite automation with fuzzy constraints and a fuzzy target. no. M. Di Nola and A. and I. "On the stability of fuzzy systems. PA. Jones. "Application of fuzzy control tu activated sludge process. "Further analysis and application of fuzzy logic control. Univ." Ph. vol. Gesarntherstellung: Verlag TUV Rheinland. Introduction to Fuzzy Arithmetic. Mizumoto and H Zimmermann. Control. Rutherford and G. K. Fuzzy Techniques in Pattern Recognition.
V()I. Introduction to Theory of Fuzzy Subsets. 805-810. K. Syst . 1981. 2.
20.
[63]
[64] [65]
[66]
[67]
[68] [69]
[70] [71]
[72] [73]
[74]
[75]
[76]
[77] [78]
[79]
[80]
[81]
[82]
[83] [84]
[R5]
[86] [87] [88]
[89]
[90]
E. pp. Kelemen. J. New Delhi: Wiley Eastern Limited.41h [34]
IUT. Mag. pp. N. Teranu. Studies. no. 1980. Amsterdam: North-Holland. 1975. vol. pp. 1987. 12. S. P. F /WK2/75. Mizumoto. P. Fedrizzi. no. 223-240. pp.
TRANSACTIONS
01'" SYSTLMS. Sugcno. J. Syst. Boston. C. Gupta. vol. Studies. Knowledge Representation and Decision Analysis. 1982. Man Cvbern. 936-937. 1-13. "Applications of fuzzy set theory. W.
. "Development of fuzzy algorithms for servo systems. un Fuzzy Information Processing. pp. 9. Gaines. Saridis: and B. Orlovski. J. 293-307. 1988. Specialists' Meeting on In-Core Instrumentation and Reactor Core Assessment." IEEE Trans. T. "The implementation of fuzzy algorithms for control." Int. pp." IEEE Trans. __ . Gupta." Automut: vol. Expert Systems and Fuzzy Systems. nu 1. I. Maeda. Ventre." Internal Rep. J. pp. G. S. New York: Van Nostrand. Mizumoto. Amsterdam: North-Holland. pp.." Fuzzy Sets Syst. 1988. vul. J.. on Cybernetics and Society. Philadelphia. 1976. Isik. A. R.. 4. 19H3." in Industrial Applications of Fuzzy Control. Ostergaad." in Proc. The Mathematics of Fuzzv Systems. Ed. M. 1974. M. A. S. Makeev. Bloore. Gotoh. vol. c." Fuzzy Sets Syst. 1986. __ .
MAN.rIdentification and fuzzy rule-based control of a mobile robot motion. Kinoshita. no." IEEE Trans. Man Cybern. H. Znd IFSA Congress. and B. G. Eds. July. "Application of fuzzy controller to automobile speed cuntrol system. vol." Automat. Kaufmann and M. Assilian." in Industrial Applications or Fuzzy Control. July. Conf. lEE. Negoita. vol. pp. 1. "Dynamical learning behavior of a rule-based self-organising controller. 1989. 6. Lau. __ . vol. "Comparison of fuzzy reasoning methods. France. Kacprzyk and S. IEEE Int. 235-242." Automat. 16. S. Man Mach. C. 22S-236. Majumder.. Syst. Fuzzy Mathematical Approach to Pattern Recognition.
/\ND
CYBI'. D. Cadarache. Amsterdam: North-Holland. no. Y. Folger." lnt . 12." Cybemet . 2. "The influence of some fuzzy implication operators on the accuracy of a fuzzy mudel-Part I. 1986. Svst . __ . G. Sugiyama. E. no. "Analysis of a fuzzy logic controller. 12. Nikiforuk. pp. D. vol. Conf. pp. A. MIT. J. IEEE Int. 1979. vol. II. J. M. 175-189. F." in First Int. A Kaufmann. P. pp.. Murayama and T. 187-230. 5. no." in Industrial Applications of Fuzzy Control. J. 7. 13. pp. E." in Proc. Fuzziness and structural stability. vol. "An automatic operation method for control rods in BWR plants. Rao and D. "Advances in the linguistic synthesis of fuzzy controllers. 2nd IFSA Congress. Dearborn." Fuzz. vol. Kasai and Y. "The influence of some parameters on the accuracy of fuzzy model. "Towards an analysis of linguistic modeling. King and E.. A." in Fuzzv Automata and Decision Processes. G. A. G. 3. . pp. H. no. Menlo Park: Benjamin/Cummings Publishing Company. 1985. R. Mamdani. 8. Sugeno. 1976. no. 1985. Van Nauta Lemke. H. SMC-15. M. Pal and D. 1985. pp. 707-717. Itoh. pp. pp. pp. C-26. Shaknov. Queen Mary College. Optimization Models using Fuzzy Sets and Possibility Theory. 1977. 15. Gupta. Sugeno. "Applications of fuzzy algorithms for simple dynamic plant. Fuzzy Sets. Reidel Publishing Company. 16. 1977. and Information.RNFTICS. Congress on Transportation Electronics. Boston: D. 247-306. Computer. "An identification of fuzzy relational systems. 3-10. Marseille. 282-285." Sets Syst.
Mi\f{CI!/!\l'RII
1990
[3S]
[36]
[37]
[38]
[39] [40]
[41] [42] [43] [44] [45] [46] [47]
[48] [49]
[50] [51]
[52]
[53]
[54]
[55]
[56] [57]
[58]
[59] [60]
[61]
[62]
C. 1987. S. 1981. Japan."Comparison of various fuzzy reasoning methods. "Application of fuzzy controller to automobile speed control system. 12. Man Mach. Man Mach. N." 1111. vol. V. Syst. "Human operator's fuzzy model in man-machine system with a nonlinear control object. 1984. L. E. Pedrycz. M. Japan.. Mamdani. 1977. Ray and D. Komolov. July 1987. France." Proc. Fuzzy Mathematical Models in Engineering and Management Science. pp. Studies. SMC-15. 1182-1191.87-104. nn. __ . M. Kickert and H. 301-308. 15-30." in IFAC World Congress.. pp. Syst." in Proc . 1975. Gupta.. Mamdani. L II. M. Dutta Majumder. J. Mamdani and B. Sardis. M. J. N. Gesamtherstellung: Verlag TUV Rheinland. 15. London. 1985. J. 1982. "Note on the arithmetic rule by Zadeh for fuzzy conditional inference. pp. J. and H. T.1985. pp. vol. no. H. H. c. 19H4. S. Symposium Intelligent Control. vol. MI. 3. pp. nu. A. 253-283. and S. "Electronically controlled continuously variable transmission. 121. Tokyo. M. V. Klir and T. Larkin. Kickert and E. M. Matsushima and H." in Proc. Takarnizawa. 28S-320. 3. 1. M. Mamdani. Lembessis. Tokyo. I. 111-128. H. Y." in Proc. "The influence of some fuzzy implication operators on the accuracy of a fuzzy model-Part II.143-146." IEEE Contr . 12. no. Ed. B. and P." in Proc. 3. 1984. Kacprzyk and M. Sugeno. Uncertainty. 1. Pappis and E. Rutherford. Kiszka. W. B. 403-421. 43-48. no. P. "Optimizing control of diesel engine. 1977. M. 783-792. "The application of fuzzy control systems to industrial processes. Murakami and M. Prade and C. vel. 1975." 1111. 1987. "Application of circle criteria for stability analysis of linear SISO and MIMO systems associated with fuzzy logic controller. vol. no. SMC-14. "Industrial applications of fuzzy logic control. H. M. Amsterdam: North-Hulland. pp. pp. OIl Fuzzy Information. M. pp. Amsterdam: NorthHolland. of London. Man Cybem . pp. Queen Mary College. "Application of fuzzy logic to approximate reasoning using linguistic synthesis. W. 1988. Morimoto. Tokyo. Ed.1985. M. T. 1. Maier> and Y." Cybem . Zimmermann. __ . M. Sherif. Man Cybern. 163-175. 10. 13. IEEE. K. pp. Gaines. vul. 1985. Englewood Cliffs.. New York: Springer-Verlag. 15. "Fuzzy controls under various approximate reasoning methods.. "An approach to the analysis of fuzzy systems. 1979." in Proc. 1988. P. T." Fuzzy Sets Syst. Sliwinska. Combining Fuzzy Imprecision with Probabilistic Uncertainty in Decision Making. "Fuzzy logic control of a heat exchange process. J. E. 1987. Gaines." Kybemetes . 1986." in Proc. 1978. A. New York: Wiley-I nterscience.. pp. 1985." Fuzzy Sets and Systems.

Miyamoto. [92) Y. [124) __ . Sasaki. Mulholland." in Proc. Univ. 59-83. 1988." in Proc.. O. Man Cybern. 26. E." Fuzzy Sets Syst. Xu and Y. Zadeh. M. M. "A simple fuzzy computer hardware system employing min and max operations-A challenge to 6th generation computer. pp.. Ed. (133) T. "A fuzzy controller in turning process automation. no. Sugeno. 551-555. Ed. Japan." Automa. 231-240.LEE: FUZZY
LOGIC IN CONTROL
SYSTEMS: FUZZY
LOGI(
CONTROI. Sugeno. Ed. 103-113. Sugeno. Amsterdam: North-Holland." Cybern . [131) __ . Yager. M. vol. pp. J. M. K. 1985.. "Fuzzy self-organizing controller and its application for dynamic processes. vol. Umbers and P. Fuzzy Sets and Applications: Selected Papers by L. Sugeno. no... "Fuzzy control of model car. Sci. Znd IFAC Symp. [127) O. vol. R. [143) "Toward a theory of fuzzy systems.. R. 1978. France. "The current mode fuzzy logic integrated circuits fabricated by the standard CMOS process. SMC-15. pp. Sugeno. and J. pp." Fuzzy Sets Syst. [136) S. Marseilles. Japan. Amsterdam: North-Holland." Informal. Computer. no. M. Carter. "An analysis of human-decision making in cement kiln control and the implications for automation. "Expert system on a chip: An engine for real-time approximate reasoning. "Application of fuzzy reasoning to the water purification process. pp. 1976. M. M. Ragade. 1976." IEEE Trans. 1987. pp. 2. Yasunobu. Amsterdam: North-Holland. 8.. July 1987. 1985. of London. July 19~3." in Industrial Applications of Fuzzy Control. Valverde. and T. vol. Hohannesburg. 1085-1087. M. "Fuzzy memory device. Japan. 147-151. "A retrospective view of fuzzy control systems. SMC-17. July 1987." Control Theory Adv. pp. no. G. Japan. __ . pp. on Decision and Control. 49-50. 2nd IFSA Congress. [142) "Fuzzy algorithm. Ihara. 2nd IFSA Congress. vol. IEEE Int." IEEE Expert Syst." Fuzzy Sets Syst. Sugeno. M. and M." in Approximate Reasoning in Expert Systems. [118) I. Tokyo. Aug. 1985. Amsterdam: North-
Holland. Sugeno. [130) T. 6. R." Automal. Y. Sugiyama. S. [132) __ . 4. Yamakawa. 419-432.
[104) (105) [106) [107) [108) [109) [110) [111) [112) (113) [114) [115J [116)
[117)
"Fuzzy identification of systems and its applications to modeling and control. S. "Automatic train operation by predictive fuzzy control. Technol. Amsterdam: North-Holland." Int. Nguyen." in Proc. Latten. "Fuzzy robot controls. Zadeh. Hasegawa. 12. M. M. pp. M. no. S. Tokyo. 36." in Proc. Amsterdam: North-Holland. Ovchinnikov. R. on Cybernetics and Society. M. Tokyo. 1968. [128) T. "Comparing fuzzy logic with classical controller designs. 125-138. no. Yasunobu and T. Sekino. Malvache. [122) __ . 338-353. "Use of fuzzy model for process control. I. no. [129) T.86-89. Yamakawa." in Proc. 835-838. Sugeno and K. pp. Zadeh. Tokyo. [91) D. July 1987. 94-102. 14. Scharf. "Decision making with possibilistic dynamic systems. [98) M. [101) M. Sugeno and K." in Proc. 1985.. vol. "Derivation of fuzzy control rules from human operator's control actions." Ph. Tokyo. King. [95) S. Yasunobu and S. [123) D. [96) M. B. 4. 1988. Syst. [140) M. Japan. Znd IFSA Congress. BadenBaden. Japan. and H. pp. Kandel. Tong. [103) T. 1986. [126) R. A. pp. Control. "On mode and implication in approximate reasoning. Scharf and N. Murakami. 1980. 19-40. Zailu. Studies. E." Fuzzy Sets Syst." in Froc. "A Heuristic adaptive controller for a sinter plant. 39-64.. "Reconfigurable fuzzy logic processor: A full custom digital VLSI. [139) S. 1986. pp. Gupta. SMC-17. 1984. pp. 6. RSA. 12. Tang and R. Sugeno and M. Dec. Ed. 1985. Int. 157-166. Gupta. M. 942-946. Gupta. M. pp. [141) L. San Diego. pp. K. 139-152. 679-686." IEEE Trans. [134) T. 1985. Tong. pp." in Industrial Applications of Fuzzy Control. R. 1987. pp. pp. Amsterdam: North-Holland. 1976. pp. 1985. T.LER-PART
I
417
572-573. [100) M. Eds. 1984. 122-130. 3. 26. and R. K. Yamakawa and T. Studies. vol. Mineral and Metal Processing. Control. Knowledge Representation and Decision Analysis. "Analysis of fuzzy control algorithms using the relation matrix. C-35. J." in Industrial Applications of Fuzzy Control. "The evaluation of fuzzy models derived from experimental data. Ed. Japan. 1979. 28. M. 2nd IFSA Congress. pp. vol. no. 1965. of the IFAC
Symp." in Aspects of Network
." in Advances in Fuzzy Set Theory and Applications. M. Syst. Kochanska. M. Syst . 55-62. 1-12." Fuzzy Sets Syst. Man Mach. Las Vegas. [93) E. Togai and S.. July 1987. Congress on Applied Systems Research and Cybernetics. [137) S. pp. Takagi and M. 4. 41-62. Man Cybem." in Froc. Tsukamoto. vol. Togai and H. R. "An approach to fuzzy reasoning method. 16. pp. Ed." in Froc. pp. Trillas and L. Man Mach. 1987. vol. vol. "Fuzzy parking control of model car. vol. vol. 1. "Experiments with the use of a rule-based self-organising controller for robotics applications. 1. S. 8th IFAC World Congress.. B. pp. Yasunobu. Murakami. 41h IFAC / IFlP / IFORS Int. [125) C. Queen Mary College. M. 1977. Eds." in Proc. pp. Tokyo. vol. 481-500. "Synthesis of fuzzy models for industrial processes. Ed. 1988. 1980. on Fuzzy Information. Yamakawa and K. Amsterdam: North-Holland." Int. 1976. 195-214. Workshop on Fuzzy System Applications. 199-210. and K. "High speed fuzzy controller hardware system. on Automation in Mining. Beck. A. 1988." in Int. A. Japan. [97) M. 15-33. pp. vol. Sugeno and G. Decision and Control. pp. [99) M. 1979. "A control engineering review of fuzzy systems. 2nd IFSA Congress. M. 1986. "Predictive fuzzy control and its application for automatic container Crane operation system. 151-164." IEEE Trans. Sugeno. 349-352. Stachowicz and M. Watanabe and W. 6. T. 559-569. 1981. Approximate Reasoning in Intelligent Systems.D thesis.. Sugeno." in Advances in Fuzzy Set Theory and Applications. Tanscheit and E. 1988. Yonekara. "An experimental study on fuzzy parking control using a model car. 695-701." Fuzzy Sets Syst. R. Tong. A. New York: Pergamnon. "Automatic train operation and automatic crane operation systems based on predictive fuzzy control. 20th IEEE Conf. 6. J. 143-162. Rutherford and G. "Fuzzy model identification and selflearning for dynamic systems.." in Froc. 1-18. no. pp. Conf. 51-52. "Analysis and synthesis of the rule based selforganising controller. Miyamoto. 683-689. Ragade. Sakai. I. [135J S. Aug. Tong. Sugeno. Mandie. 1978. pp. "A fuzzy accelerator and a programming environment for real-time fuzzy control." Int. 1986. "Fuzzy control for automatic train operation system. "Fuzzy microprocessors-Rule chip and defuzzifier chip. L. Japan. pp. Tokyo. Man Cybem. A. Miki. M. M. Gen. 2nd Fuzzy System Symp. 1985. Tokyo. 6. vol." in Industrial Application of Fuzzy Control. M." IEEE Trans. W. 161-167." in Proc . Amsterdam: NorthHolland. "An annotated hihliography of fuzzy control. 2nd IFSA Congress. "Fuzzy control of the activated sludge wastewater treatment process. Workshop on Fuzzy System Applications. pp. Kiszka. 2nd IFSA Congress. "Evaluation of an automatic container Crane operation system based on predictive fuzzy control. "Optimal control of fuzzy systems. "The application of fuzzy sets theory to the temperature of box annealing furnaces using simulation techniques. pp. 249-269. vol.. Syst. W. 13. Mizumoto. Sugeno. Willaeys and N. A." in Froc. pp. 1988. pp. vol. 1985. 1986. and H. Watanabe. J. vol. 55-60. "A microprocessor based fuzzy controller for industrial purposes. Yagishita. 11-23. Dec. pp." in Proc. Yamazaki and M. "Structure identification of fuzzy model. Chiu. Willaeys. Kang. [102) K. "The application of a fuzzy controller to the control of a multi-degree-freedom robot arm. 1980. "An introductory survey of fuzzy control." in Industrial Applications of Fuzzy Control. 8. R. [120) II. 19~O. pp. "Fuzzy controller hardware system. lizuka. Tananka." in Proc. July 1987. Amsterdam: North-Holland. Hasegawa. Eds. July 1987." Inform. R. and A. April 1983. Sanchez and L. on Decision and Control. 1985. [119) M. pp. Yager. vol." in Industrial Applications of Fuzzy Control. Bandler. "Fuzzy modeling of the process. Uragami. 1985." in Industrial Applications of Fuzzy Control. 1981. Sept. July 1987. Acapulco. "Fuzzy sets. vol. "Contribution of the fuzzy sets theory to man-machine system. Congress on Control in Transportation Systems. [13S) __ . no.. Mag. 116-132. 2. 16. pp. Nishida. and R. Japan. pp. [94) E." in Int. Itoh." Informat.. Dettloff." in 23rd IEEE Con]. J. New York: Wilcy-Interscience. [121) D. lizuka. Shao.. Yager. R.

H. 1975. "A rationale for fuzzy control. the IEEE Control Society.S. "The concept of a linguistic variable and its application to approximate reasoning I.
VOL.
2. __ . vol. degree in electrical engineering from the University of California. pp. Man Cybern. Taiwan. SMC-3. His current research interests lie in the areas of learning systems. approximate reasoning.418
IEEE TRANSACTIONS
ON SYSTEMS." Trans. 1. 1971. Fuzzy Sets. 1973. pp. 419-435. Zimmermann. ASME. Mr. Berkeley. Berkeley.
. 20. Lee is a member of Phi Tau Phi. Syst.
20." TEEE Trans. qualitative system analysis. pp.. vol. J. the IEEE Computer Society. pp.
\'IAN. 3-4.D. H.
AND
CYBERNETICS. 43-80. E. Boston: Kluwer Academic Publishers. Lee. 2. vol. Sci. Since 1986. in 1983 and the M. no. __ . III." Informat. Mar. 1985.." lnformat . vol. Man Cybern. "Similarity relations and fuzzy orderings. Decision Making. Syst. Fuzzy Set Theory and Its Applications. pp. 1971. __ .S. He is currently a candidate for the Ph. Ed. 8. Measur. DeClaris. Zimmermann. "Outline of a new approach to the analysis complex systems and decision processes. and Cybernetics Society. Boston: Kluwer Academic Publishers. C. and the IEEE Systems. 9. and artificial intelligence.y Apr. in 1987. 1987. Berkeley. fuzzy control. Dynam. II.
'v1. 177-200. 1990. "Fuzzy logic in control systems: Fuzzy logic controller -Part II. 199-251. J. Rinehart and Winston. R.. pp. c. __ .. He is also a student member of the American Association for Artificial Intelligence. 469-490.
NO. Sci." IEEE Trans. neural networks. 1972. 301-357. New York: Holt. pp. Man. Syst. vol. degree in the Department of Electrical Engineering the Computer Sciences at the University of California. intelligent control. He spent the summer of 1988 at the NASA Ames Research Center. 94. 28-44. Kalman and N. Control. vol. and Expert Systems. he has been a Research Assistant in the Electronics Research Laboratory al the University of California. pp. working on a fuzzy control systems project.r\RCl~/ APRIL
1990
[144] [145] [146] [147] [148] [149] [150]
and System Theory. 3. degree in control engineering from the National Chiao-Tung University.
Chuen Chien Lee (S'88) received the B.