Professional Documents
Culture Documents
An S4PR Class Petri Net Supervisor For Manufacturing System
An S4PR Class Petri Net Supervisor For Manufacturing System
Mowafak H. Abdu-Hussin
Abstract — Petri Nets are a family of formalisms that allow to effectively model manufacturing systems. The experimental
approach method is provided to ensure formally in elementary siphons to distinguish deadlock conditions for a class in terms of
generalized Petri nets, namely S4PR. A siphon is a kind of special structural objects of a Petri net and plays an important role in
synthesizing a live Petri net controller of flexible manufacturing systems FMS. An FMS corresponds to a class of concurrent
systems called Resource Allocation Systems (RASs). The application of Petri Nets (PNs) in RASs is an active research field devoted
to defining and exploits different subclasses of Petri Nets allowing modeling the widest set of RAS. A monitor is added for a derived
minimal siphon such that it is max-controlled if it is elementary with respect to the siphons that have been derived. Simulation PN
to the structural analysis and reachability graph analysis are employed to synthesize and control of FMS. An example is used to
illustrate this method.
(Simple Systems of Simple Sequential Processes with a formalism that allows the modeling of systems involving
Resources) is equivalent to an S4R [4]. The benefits of the concurrency, resource sharing, conflict, which allows to the
deadlock prevention policy, by means of adding a control effectiveness of the correctness requirement are performed
place (monitor) for each siphon to remain marked for all on the qualitative properties of the net modeling. Qualitative
reachable markings. Furthermore, the author [4] studies a properties of Petri nets (such as siphons, liveness, deadlock,
deadlock prevention and deadlock avoidance policy to conflict, and safeness) are useful in discrete manufacturing
control the reinforcement S4R nets using a resource applications. The formalism employed to ensure the desired
allocation policy based on the unsafe marking concept. The pattern of the flow control, particularly the deadlock
resource allocation decision can be structured computation avoidance of FMS. In this work, the monitors are added to
the plant model such that the elementary siphons in S4PR
the controller depended on siphons. Recent researchers [10],
are all max-controlled and no insufficiently marked siphon
[15]−[17], [21]−[28] are developed a novel approach also is generated.
suffers from such problems as computational complexity
[10], [11], [36], behavior permissiveness, and structural The remainder of this paper is organized as follows. In
complexity. section II, provides the preliminaries and notations of Petri
The structural object of siphons is extensively used to nets and S4PR nets which have the focus of this work. A
characterize and analyze deadlock situations in FMS that are deadlock prevention policy for an S4PR is developed in
modeled on Petri net (PN) [13]. The generalized PN is Section III. In Section IV, an S4PR example is proposed
considerably more complex than that in an ordinary one on method to show our application efficiency and simulation
account of a marked siphon might not be sufficient in the PN of the proposed method. Section V, concludes the paper.
absence of dead transitions. An iterative method to deal with
the deadlocks in S4PR nets is developed in [6], II. –II. PRELIMINARIES
[13],[15]−[17], [23], [24], [27], [28], [31]−[34] . The Petri nets have been mainly used for studies in computer
iteration, it typically uses an integer programming to find operating system behavior, but recently used to model and
out approaching an insufficiently marked siphon that can visualize behaviors comprising concurrency, sequential
cause dead transitions and then designs control places for processes with resource sharing found in FMSs.
the siphon by solving an integer programming problem.
Recently, there exists an increase in Petri net models for A. Petri Net
modelling of RAS in the states of FMS, that is, lead to some A generalized Petri net [1], [19], [21] is a four-tuple
of the syntactical limitations of the S3PR class G ( P, T , E ,W ), where P and T are finite, non-empty
[1]−[3],[10],[19], [24], [25], [35]−[36]. The classes of an and disjoint sets. P is a set of places and T is a set of
S4PR is solicitude the modeling of concurrent cyclic transitions with P T , and P T . E
sequential processes sharing common resources, where an
operation places can use simultaneously multiple resources ( P T ) (T P) is called a flow relation of the net,
of different types. Classes of an S4PR net model [6], represented by arcs with arrows from places to transitions or
[15]−[17], [26], [28], [31-34] generalizes the earlier, that from transitions to places. W : (P T ) (T P) ,
allowing multiple simultaneous allocations of resources
before the process. One of the formal, recent FMS where {0,1, 2,... } is a mapping that assigns a weight
representations is a Petri net model, the classical model for to an arc: W ( x, y ) 0 if ( x, y ) E , and W ( x, y ) 0
both pictorial and a formal description of the systems where
otherwise, where ( x, y ) ( P T ) (T P ) . A net is
concurrent processes are performed. The production Petri
nets S3PR in [12], Linear S3PR (L-S3PR) [11], S4R [4], said to be ordinary, denoted by G ( P, T , E ), if
S4PR [6], [27, [28], [33], extended the capabilities of S3PR ( x, y ) E , W ( x, y ) 1 . A net G ( P, T , E ,W ) is
to be S4PR models beyond Sequential RAS (ES3PR) [21],
[22], and weighted S3PR (systems of simple sequential
pure (self-loop free) if W ( x, y ) 0 implies
processes with resources) WS3PR [9], [26] etc. Chao [8] W ( x, y ) 0 , ( x, y ) (T P ) . A pure net
proposed deadlock control approaches from S3PGR2 G ( P, T , E ,W ) can be alternatively represented by its
(systems of simple sequential processes with general
incidence matrix [C ] that is a | P | x | T | integer matrix
resources necessity) experience in incorrect or restricted
liveness of the system is depending on max-controlled with [C ]( p, t ) W (t , p ) W ( p, t ) .
siphons. Shih et al. [30] proposed a sequence of control in
S3PMR. Given a net G ( P, T , E ,W ), and a node x P T ,
The Petri net has an appealing graphical representation The preset of x is defined as
of a powerful mathematical formulation and is favorite's
tools suitable for modeling of FMSs with parallel and
x { y P T | ( y, x) E. While the post-set of a
concurrent activities. Thus, it has generated intense interest node x P T is defined
in many researchers [1]−[28], and [30]−[36]. Petri nets are
p P, | p | | p | 1 . I : P ( J : T )
indexed by P(T ) , where is the set of integers. We
A marking M of a net G is a mapping from P Z , denote a column vector whose entries equal 0 or (1) by
where Z {0, 1, 2, … }. M ( p) denote the number of 0(1) . I T and [G ]T are their transposed versions. I ( J ) is
tokens contained in place p. Place p is marked by M iff a P(T ) -invariant if I 0 ( J 0) and
M ( p) 0. A subset S P is marked at M if at least [G ]T I 0, ([G ] J 0) .
one place in S is marked at M . M ( S ) is denotes the sum || I || { p P | I ( p) 0} (|| J || {t T | J (t ) 0}) is
of tokens contained in S at marking M , where called the support of I (J ) , where
M ( S ) p S M ( p) . (G, M 0 ) is called a net system or
|| I || {p P | I ( p) 0}
marked net. For economy of space, we use pP
M ( p) p (|| J || (t T | J (t ) 0}) is called the positive support
while
to denote vector M . For instance, M (7, 3, 0, 2, 0, 4)T
in a net with seven places p1−p7 can be rewritten as: M0 = || I || { p P | I ( p) 0}(|| J || {t T | J (t ) 0})
(7p1 + 3p2 + 2p4 + 4p5). A place p is marked at M if is called the negative support. A P (T ) -invariant is
M ( p) 0 . minimal if it contains no P (T ) -invariant as a proper
subset. P (T ) -invariant is said to be a P (T ) -semiflow if
Transition t T is enabled at M if (p t ) ,
none of its components is negative. An invariant is called
M ( p ) W ( p, t ). this fact is denoted as M [t ; when minimal when its support is not a strict superset of the
fired, this gives a new marking M ' such that p P, support of any other, and the greatest common divisor of its
M ' ( p) M 0 ( p) W ( p, t ) W (t , p) , which is elements is one. If I is a P-invariant of net (G , M 0 ) , then
M n such that M 0 [t1 M 1[t 2 M 2 , ... , where C is the incidence matrix of G. A non-zero integer
M n1[tk M ' holds. The set of markings reachable from vectors y 0, (resp. x 0).
M in net G is called the reachability set of Petri net A nonempty set S P is a siphon if S S and
(G, M ) and denoted as: R(G, M ) . Transition t T is Q P is a trap if Q Q). A siphon is minimal if
live at M 0 if M R (G , M 0 ) , M ' R (G, M 0 ) , there is no siphon contained in S as a proper subset. A
M '[t holds, t is friable under M '. A transition t T is minimal siphon containing no support of any trap is a strict
minimal one.
dead under M 0 if M R(G, M 0 ), where t is firable.
A siphon S is said to be invariant- controlled by P-
A marking M R (G , M 0 ) is a (total) deadlock iff
invariant I if I T M 0 0, and || I || S . A siphon S is
t T , t is dead.
said to be max-marked at M R (G , M 0 ) if p S
Step 1. p PS P PR , k S ( p ) : 0;
0
p y SP ( p , p ) ,
m
i 1
0
i
i
S
Tow types of parts must be processed, where a part of type1 M0(p21) = 2. While places M0(p15) = M0(p20) = M0(p22) = 3,
arrives the system by conveyor I1, and leave it by conveyor and others places are zero. Places p17 , p18 p24 , denote
O1. Machines M1 or M3 can be processed first. While
the shared resources R1, H1, H2, M3, R2, H3, H4, and R3
Machine M1 needs to get tools H1 to achievement its work respectively. In the system, there are two H1, one H2, three
and M3 need to occupy tool H1 and H3. Machine M2 has a
H3 and three H4 tools. The process idle places are: P
0
part to be processed which is needed to use tools H2 and H4
{p5, p6}, and the resource places are: {p14 – p24}. The places
(arrival to operations of production routing J 1 .
p14 , p15 and p16 is denoted the resources M1, M2, and
M4 respectively. The PN is bounded, not live and not
reversible but repetitive so that it can be made live. The
initial marking of PN can be denoted as: M0 = {10p5 +10p6
+ 2p14 + 3p15 + 2p16 + p17 + 2p18 + p19 + 3p20+ 2p21 + 3p22 +
p24}. We are assumed that the weight of arcs: W(p21, t4) =
W(p21, t5) = W(p21, t11) = W(t6 , p21) = W(t12 , p21) = 2, in
order to take specifications of weight arcs to PN. A net is a
format to the generalized PN.
The net shown in Figure 2 is pure and bounded and there
exist dead reachable states and not live. The modeling and
simulation of FMS are used in our example with Petri net
Toolbox with Matlab [29]. There are eleven resources in
this system leading to eleven minimal P-invariants, we
would like mention as follow: I1 = p2 + p14, where M0(p14) =
2, I2 = p7 + p15, where M0(p15) = 3, I3 = p12 + p16, where
M0(p16) = 2, I4 = p1 + p9 + p17 , where M0(p17) = 1, I5 = p2 +
p4 + p10 + p18, where M0(p18) = 2, I6 = p4 + p10 + p19, where
M0(p19) = 1, I7 = p4 + p10 + p20, where M0(p20) = 3, I8 = 2p3
+ 2p11 + p21 , where M0(p21) = 2, I9 = p7 + p12 + p22, where
M0(p22) = 3, I10 = p7 + p12 + p23 , where M0(p23) = 3, I11 = p8
+ p13 + p24, where M0(p24) = 1.
4 Simulation and structural analysis of the behavioral
Fig. 2. Petri net model for an S PR (G 0 , M 0 ) properties of Petri net model use the PN-tool with
This cell produced two products types, i.e. part-1 and part- MATLAB [29], starts with coverability tree keys. We can
2 (p1, and p2), is performed to show how to model and see the original net system has (1330) reachable states with
control of an FMS using methodologies Petri net initial marking, among which there are (18)-deadlock states.
implemented throughout this example. There are two part In order to solve the deadlock problem, one of the most
product types, namely P1, and P2, to be performed. For widely is exercised the Petri nets tool to design a controller
these raw product types the production cycles are as shown to avoid deadlock. We can find out minimal siphons of Petri
in Figure 1(b), belonging to the S4PR class. They perform net shown in Fig. 2. The net system of FMS is an S4PR, and
two types of parts must be processed achieved to their own contains deadlocks. Analysis structured of PN, there is 44
production routings that are depicted in Figure 1(b). An strict minimal siphon SMS as shown below. Among the
S4PR (Simple Systems of Simple Sequential Processes with set of siphons from S1−S3 are elementary siphon and S4−S44
Resources) is showing in Fig. 2 belong to generalized Petri which are referred to the element dependent ones are
nets have competitive processes of the Resource Allocation marked by *.
Systems RAS modeling, and design of FMS. Figure 2, S1 = {p3, p10, p18, p21}, S2 = {p8, p12, p23, p24},
shows an example of a Petri net which models an FMS S3 = {p3, p10, p20, p21}, S4* = {p2, p4, p9, p17, p18},
problem that was designed using the Petri tool application. S5* = {p7, p11, p21, p23}, S6* = {p8, p12, p22, p24},
The Places (p1– p7) represent the operation of R1, M2, R2 S7* = {p3, p10, p19, p21}, S8*= {p7, p11, p21, p22},
and M3 respectively, for production cycle's part type-P1, S9* = {p8, p9, p14, p17, p20, p21, p23, p24},
while production cycles part type-P2 (p13 – p9) are S10* = {p8, p9, p14, p17, p19, p21, p22, p24},
performed on the operation place of M3, R2 and M2 S11* = {p8, p9, p14, p17, p19, p21, p23, p24},
respectively. The Petri net is illustrating an example of and S12* = {p8, p9, p14, p17, p20, p21, p22, p24},
by this place specification that uses a multi-set of resources S13* = {p8, p10, p20, p21, p23, p24}, S14* = {p8, p10, p20, p21, p22,
at a production routing. Marking is acceptable tokens in Fig. p24}, S15* = {p8, p10, p19, p21, p23, p24},
2, where is token M0(p5) = M0(p6) = 10, places M0(p17) = S16* = {p8, p10, p19, p21, p22, p24}, S17* = {p8, p9, p17, p18, p21,
M0(p19) = M0(p24) = 1, places M0(p14) = M0(p16) = M0(p18) = p23, p24}, S18* = {p8, p9, p17, p18, p21, p22, p24},
S19* = {p8, p10, p18, p21, p23, p24}, checking three siphons and added three monitors (or control
S20* = {p8, p10, p18, p21, p22, p24}, places) VS1−VS3 are satisfying the max cs-property.
S21* ={p8, p11, p21, p23, p24}, S22* The complementary sets of siphons S1 – S3 are computed
= { p8 , p11 , p21 , p22 , p24 } , {p8, p11, p21, p22, p24}, S23* = as:
{p7, p9, p14, p17, p20, p21, p23}, 1)- For Siphon S1 = {p3, p10, p18, p21}, is contain two
S24* = {p7, p9, p14, p17, p20, p21, p22}, resource places (i.e. p18, p21).
S25* = { p8 , p9 , p14 , p15 , p16 , p17 , p20 , p21 , p24 } ,
S26* = {p3, p9, p14, p17, p20, p21}, S27* = {p7, p9, p14, p17, p19,
Let l1 I p18 I p 21 I
r S 1 R
p18 I p 21 ,
p21, p23}, S28* = {p7, p9, p14, p17, p19, p21, p22},
S29* = {p8, p9, p14, p15, p16, p17, p19, p21, p24}, l1 {p2 + 2p3 + p4 + p10 + 2p11 + p18 + p21};
S30* = {p3, p9, p14, p17, p19, p21}, S31* = {p7, p9, p17, p18, p21, Th(S1) = [H(p18) + H(p21)] \ S1 ;
p23}, S32* = {p7, p9, p17, p18, p21, p22}, S33* = {p8, p9, p15, p16, Th( S1 ) p2 + p4 + 2p11, where applied definition 1.
p17, p18, p21, p24}, S34* = {p3, p9, p17, p18, p21},
S35* = {p7, p10, p18, p21, p23}, S36* = {p7, p10, p18, p21, p22}, k S1 p1 + p2 + p4 + 2p11 + 2p12 + 2p13 ;
S37* = {p8, p10, p15, p16, p18, p21, p24}, g S1 k S1 VS1 ,
S38* = {p8, p10, p15, p16, p19, p21, p24},
S39* = {p8, p10, p15, p16, p20, p21, p24}, g S1 {p1 + p2 + p4 + 2p11 + 2p12 + 2p13 + VS1}; Clearly,
S40* = {p8, p11, p15, p16, p21, p24},
S41* = {p7, p10, p20, p21, p23}, S42* = {p7, p10, p19, p21, p23},
hS1 l1 g S1 .
S43* = {p7, p10, p20, p21, p22}, S44* = {p7, p10, p19, p21, p22}. hS1 2p3 + p10 + p18 + p21 – VS1 – p1 – 2p12 – 2p13; is a P-
In Fig. 2, there are eleven resources exist in this system
leading to eleven minimal P-semi-flows in table: invariant. Noticing that we say that S1 is max-controlled by
TABLE 1
the P-invariant hS1 . M 1 (VS1 ) M 0 (S1 )
Eleven resources : (P-
Invariant equations)
The holders of resource r:
H (r ) I r \ {r}
pS1
(maxp 1) 1 (2 2 1) 1 2 .
H(p14) = {p2 + p14}\ p14 = p2 ;
I p14 p2 + p14, 2)- For Siphon S2 = {p8, p12, p23, p24} is contain two
resource places (i.e. p23, p24).
I p15 p7 + p15, H(p15) = {p7 + p15}\ p15 = p7 ;
G ( P, T , E ), if e E,W(e) 1.
While PN is said to be a generalized net iff e E , 3)- For Siphon S3 = {p3, p10, p20, p21}, is contained two
resources (i.e. p20, p21).
W (e) 1.
Let l3 I
r S 3 R
p 20 I p 21 ,
Having applied the Proposition 1, in order to make all
elementary siphons has a max-controlled. Our experiment is
l3 {2p3 + p4 + p10 + 2p11 + p20 + p21}, proposition 1, 2, and definitions 9, 10, and 11 to adding
monitors or control places (VS1VS3) to the net of Fig. 3,
Th( S3 ) p4 + 2p11 the Petri net is live. A Petri nets are suitable to describe the
k S 3 2p1 + 2p2 + 2p3 + 2p4 + 2p7 + p13 sequential and parallel execution of assignments to or
without synchronization; it is possible to define loops and
g S 3 k S 3 VS3 , the conditional execution of tasks in FMS. Conditional
transitions are required at every region in the Petri net where
g S 3 2p1 + 2p2 + 2p3 + 2p4 + 2p7 + p13 + VS3; two or more activated transitions shares the same input
hS 3 l3 g S 3 , place and compete for the token that are located to that
place. Petri nets possess special characteristics that can be
hS 3 p10 + 2p11+ p20 + p21 − 2p1 −2p2 − p4 − 2p7 − p13 − defined mathematically and are used to analyze and control
VS3; of FMS. Terms like conflict, confusion, contact, liveness
and deadlock are well-defined properties of Petri nets that
M 1 (VS3 ) M 0 ( S3 ) pS (max p 1) 1 may be helpful when optimizing FMS. The reachability
3
Petri nets is suitable for FMSs with dynamic structures. A of more general Petri nets. A structural non-liveness of Petri
simulation of Petri net models is the execution of a model net condition is realizing in the local structure of siphons
of RAS in FMS, represented by a computer program that which is containing no trap. The cs-property highlights the
gives structure information in the location of deadlock on conditions are markings since it provides an explanation of
the non-featureless of liveness property.
the net. The popular software is training and research where
the installation Petri Net Toolbox in MATLAB, which can
be to draw Petri net V .2.3 software's with the exploitation References
of Graphical User Interfaces (GUIs), depended on this [1] Abdul-Hussin M. H.,"Elementary Siphons of Petri Nets and Deadlock
prevalent software. Control in FMS", Scientific Research Publishing , J. Computer &
Communications, 2015, (3), 1-12. DOI: 10.4236/jcc.2015.37001
[2] Abdul-Hussin M. H.,"Design of a Petri Net Based Deadlock
The performance results show that is our proposed method Prevention Policy Supervisor for S3PR", in Proc. IEEE-ISMS2015,
reduces the reachability graph of a class of an S4PR net for 6th Int. Conf. Intelligent Systems, Modelling & Simulation, Kuala
modeling, simulation, and control FMS complexity and it Lumpur, (2015), pp. 46-52.
will help managers to understand the changing behaviors of [3] Abdul-Hussin M. H. ,"Synchronization Competitive Processes of
FMS. Adding the controller places (monitors) to the net of Flexible Manufacturing Systems Using Siphons Petri Net”, in:
Proc. IEEE 5th National Symposium on Information Technology:
Figure 4 is live and obtained to M20 reachable states Towards New Smart World 2015, Riyadh, Saudi Arabia , pp. 1-6.
marking so that the Petri net is deadlock-freeness. A siphon [4] Abdallah, I.B., and H. A. ElMaraghy, "Deadlock prevention and
is generating even though the monitors are added. The avoidance in FMS: a Petri net based approach’, Int. J. Adv. Manuf.
reachability graph has reduced to (21) states with initial Tech., 1998, 14, (10), pp. 704–715
marking shown in Figure 4. This example gives as a Petri [5] Barkaoui K. and J. F. Pradat-Peyre,“ On liveness and controlled
siphons in Petri nets,” in Application and Theory of Petri Nets 1996:
nets tool dynamic siphon are used for the purpose of control 17th Int. Conf. Osaka, Japan, 1996 Proc., vol. 1091 of LNCS ,
of the network, which is representing the system such as Springer, Berlin, pp. 57–72,
FMS. The visualization of Petri net simulation is helped us [6] Cano, E. E., C.A. Rovetto, and J. M. Colom,” On the Computation of
to structure analysis of Petri nets, and the MATLAB the Minimal Siphons of S4PR Nets from a Generating Family of
software can show the PN manner of the simulation Siphons”. In: Emerging Technologies and Factory Automation
(ETFA), (2010) IEEE Conf. on, pages 1 –8.
analytical structure of siphons are correct results. The
[7] Chu, F, and X. L. Xie (1997),"Deadlock analysis of Petri nets using
resulting models can be visualized using Petri net with siphons and mathematical programming", IEEE Trans. on Robotics
MATLAB toolbox. and Automation, Vol. 13, no. 6, pp 793 – 804
[8] Chao, D. Y., (2010) ," Fewer Monitors and More Efficient
V. CONCLUSION Controllability for Deadlock Control in S3PGR2 (Systems of Simple
Sequential Processes with General Resource) Requirements”,
Computer Journal, (British Computer Society), vol. 53, no.10, pp.
The use of the maximal control siphon (cs)-property to 1783-1798
control a generalized Petri net is still an open problem [9] Chao, D.Y.," Weighted Characteristic P-vector and Deadlock Control
requiring a further study. Also, a sufficient and necessary of WS3PR", J. of Info. Sci. and Engi. Vol. 26, (2010), 1121-1136
siphon control condition for S 4 R, and S4 PR family remains [10] Chen Y.F., Z. W, Li, K. Barkaoui, et al.,"Compact Supervisory
Control of Discrete Event Systems by Petri Nets With Data Inhibitor
open. A deadlock prevention policy of FMSs is proposed, Arcs", IEEE Trans. Syst., Man, Cybern. A, Syst., Hu., (99), pages 1-
which can obtain a live Petri net controller based on a 16, 2016.
controllable siphon basis. Petri net models of FMS provide [11] Ding Z., Y. Zhou, and M. C. Zhou, “Modeling self-adaptive software
us the possibility to consider integrating the planning phase systems with learning Petri nets,” IEEE Trans. Syst., Man, Cybern.,
is enabled to drive different elements of resources, by Syst., vol. 46, no. 4, pp. 483–498, Apr. 2016.
frequently changing assortment production, requires [12] Ezpeleta. J; J.M. Colom, and J. Martinez (1995)," A Petri net based
concurrent, random order processing of various workpiece deadlock prevention policy for flexible manufacturing systems,"
IEEE Trans. on Robotics and Auto. , vol. 11, n. 2, pp. 173-184.
types. Through, the efficient experimental studies used a
[13] Ezpeleta, J., and L, Recalde (2004)," A Deadlock Avoidance
class of an S4PR based siphons property of Petri net in order Approach for Non-sequential Resource Allocation Systems", IEEE
to obtain the liveness of FMS. Petri net based siphon is used Trans. Syst., Man, Cybern. Part A, 34, 93–101.
to detect the deadlock privation in RAS, that determines the [14] Fanti M.P., and M. C. Zhou. 2004. Deadlock control methods in
properties Petri net and superior bounds for the minimum automated manufacturing systems. IEEE Tran. Sys., Man and
make range, maximize the productivity, etc. Based on Petri Cyber., Part A, 34(1), 5-22.
nets to deal with the deadlock problem with a class of Petri [15] Hou, Y. F., M. Zhao, and D. Liu (2016),” Deadlock Control for a
nets, namely S4PR, and has developed a novel and effective class of generalized Petri nets Based on Proper Resource Allocation,
deadlock-prevention policy. By explicitly controlling Asian Journal of Control, Vol. 18, No. 1, pp. 206–223, Jan. 201
elementary siphons via adding monitors, a liveness- [16] Hou, YiFan, Zhiwu Li, Mi Zhao, and D. Liu (2014),” Extraction of
elementary siphons in a class of generalized Petri nets using graph
enforcing controlled system can be found in the reachability theory", In: Engi. Computations, Int. J. for Computer-Aided Engi. &
graph can fully reflect the behavior of a system. The Software Vol. 31 No. 2, pp. 331-352
extension of the proposed a class of an S4PR net, in our
future works includes developing deadlock control policies
[17] Hou, Yi Fan, Li, Zhi Wu, Zhao Mi, and Ding Liu (2013),"Extension Emerging Technologies & Factory Automation (ETFA'2014). pp. 1-
of Elementary Siphons in a Class of Generalized Petri Nets", Applied 12
Mechanics and Materials, vol. 284-287, pp. 2238-2243 [28] López-Grao, J.P, and J.M. Colom (2006),"Lender processes
[18] Li J., X. Yu, M. C. Zhou, and X. Dai ,"Lean Reachability Tree for competing for shared resources: Beyond the S4PR paradigm", In:
Unbounded Petri Nets", IEEE Trans. Syst., Man, Cybern., Society, Proc of 2006 IEEE Int. Conf. on Systems, Man, and Cybernetics,
Jan. 2016, 10 pages Taipei, Taiwan
[19] Li Z.W., and M. C. Zhou, “Elementary siphons of Petri nets and their [29] Mahulea, C, M., Matcovschi and O. Pastravanu,"Home Page of the
application to deadlock prevention in flexible manufacturing Petri Net Toolbox for MATLAB", ( 2014 ),
systems,” IEEE Trans. Syst., Man, Cybern. A, Syst., Humans, vol. 34, http://www.ac.tuiasi.ro/pntool
no. 1, pp. 38–51, 2004 [30] Shih, Y.Y. and D.Y. Chao (2010), “Sequence of control in S3PMR,”
[20] Li Z. W., G. Liu, H. M. Hanisch, and M. C. Zhou, Deadlock Computer Journal, (British Computer Society), vol. 53, no.10, pp.
prevention based on structure reuse of Petri net supervisors for 1691-1703. Published by Oxford University Press.
flexible manufacturing systems, IEEE Trans. Syst. Man and [31] Tricas, F, Garcia-Valles, J. M. Colom, and J. Ezpeleta (2005)," A
Cybern., A, vol.42, no.1, pp.178-191, 2012 Petri Net Structure Based Deadlock Prevention Solution for
[21] Li Z. W., M. Uzam, and M. C. Zhou (2008),“Deadlock Control of Sequential Resource Allocation Systems. In: Proc. Of the 2005 IEEE,
Concurrent Manufacturing Processes Sharing Finite Resources,” in: Int. Conf. Rob. & Aut., pages 272–278, Barcelona, Spain
Int. J. Adva. Manuf. Tech., vol. 38, No. 7-8, pp. 787-800 [32] Tricas F., F. Garcia-Valles, J.M. Colom, and J. Ezpelata (1998)," A
[22] Li, Z. W., and N. Wei,” Deadlock control of flexible manufacturing structural approach to the problem of deadlock prevention in
systems via invariant-controlled elementary siphons of Petri nets”, processes with resources, in Proc. of WODES’98, pp. 273-278
Int. J. Adv. Manuf. Technology Vol. 33(1–2):24–35, 2007 [33] Tricas F., and J. Ezpeleta, “Computing minimal siphons in Petri net
[23] Li Z. W., and M. C. Zhou (Eds.)," Deadlock Resolution in models of resource allocation systems: A parallel solution,” IEEE
Automated Manufacturing Systems: A novel Petri net Approach", Tran. Sys., Man and Cybern., A: Humans, vol. 36, no. 3, pp. 532-
(Springer, London) 2009 539, 2006
[24] Liu G.Y., and K. Barkaoui," A survey of siphons in Petri nets", in: [34] Zhao M., Li Z. W., and S. H. Hu ,"Suboptimal liveness-enforcing
infor. Sci., 363, pages 197-220, 2016 supervisor design for a class of generalized Petri nets using partial
[25] Liu H.X., K. X. W. Wu, M.C. Zhou, and H. Zou," Deadlock siphon enumeration and mathematical programming", Int. J. Sys. Sci.
Prevention for Flexible Manufacturing Systems via Controllable , Vol. 41(9), (2010) ,pp. 1013 – 1026 ,
Siphon Basis of Petri Nets", IEEE Tran. Sys.,, Man and Cyber., vol. [35] Zhang X. Y. , Z.W. Li , C.F. Zhong , A.M. Al-Ahmari,"Reachability
45, n. 3 (2015) 519-529 analysis of a class of Petri nets using place invariants and siphons",
[26] Liu D., Z. W. Li, and M.C. Zhou," Hybrid Liveness-Enforcing Policy Maejo Int. J. Sci. Technol. 7 (2) (2013) 278–290 .
for Generalized Petri Net Models of Flexible Manufacturing [36] Zhong, C. F., Li, Z.W., Chen, Y.F., and Al-Ahmari,"On non-
Systems", IEEE Trans. Syst., Man, Cybern., Syst., vol. 43, no. 1, pp. existence of a maximally permissive liveness-enforcing pure net
85–97, 2013 supervisor", IEEE Tran. Sys., Man and Cybern. Syst. , (2013),(43),
[27] López-Grao, J. P, J. M. Colom, and F. Tricas (2014)," The deadlock 29–37
problem in the control of Flexible Manufacturing Systems: An
overview of the Petri net approach", in: 19th IEEE Int. Conf. on