Professional Documents
Culture Documents
To cite this article: Mohammad Tabatabaei & Reza Barati-Boldaji (2017) Non-overshooting PD
and PID controllers design, Automatika, 58:4, 400-409, DOI: 10.1080/00051144.2018.1471824
REGULAR PAPER
1. Introduction
eliminate overshoot in closed-loop system step response
Simple structure and easy tuning of PID controllers was proposed in the literature [11]. Moreover, state
have made them as the most popular controllers for feedback design to reach a non-overshooting step
industrial applications [1]. A variety of approaches have response has been reported [12]. Characteristic Ratio
been proposed to design PID controllers in the lit- Assignment (CRA) method is one of the common
erature. PID controller coefficients could be adjusted approaches for transient response control. In this
to minimize a closed-loop system performance index method, the characteristic ratios appropriately related
by means of some optimization methods [2]. Internal to the denominator coefficients of a transfer function
Model Control (IMC) method has been employed for are assigned to reach a non-overshooting step response
analytical design of PID controllers [3,4]. Tuning PID [13]. Manabe introduced Coefficient Diagram Method
controllers based on some frequency specification such (CDM) to assign characteristic ratios [14]. A Butter-
as gain margin, phase margin and gain crossover fre- worth filter pattern for characteristic ratios to achieve
quency has been considered, too [5]. In [6], a method the desired transient response was proposed, too [15].
to design PID controllers based on the phase margin Designing PID controllers to reach a non-
and gain crossover frequency requirements has been overshooting step response has been taken into consid-
proposed. In this work, the crossover frequency has eration in the literature. In [16], the relation between
been selected based on an integral performance index the location of transfer function poles and zeros and
criteria. Moreover, adaptive control and auto-tuning step response overshoot has been extracted. This rela-
methods have been utilized for PID controller design tion has been utilized to design a PID controller to avoid
in the literature [7]. overshoot in the closed-loop system step response. In
On the other hand, transient response control has another work, a non-overshooting PI controller for
been considered by the control engineers. Attain- variable-speed motor derives has been provided [17].
ing a non-overshooting or minimum overshoot step In [18], the Tabu search optimization method has been
response is required in some real plants. Thus sev- employed for adjustment of the PID controller parame-
eral methods have been introduced to achieve this ters to attain a desired transient response. Widder’s and
goal. In [8], necessary and sufficient conditions for Markov–Lucaks theorems have been employed to attain
state space models to achieve a non-overshooting a non-negative error response [19]. In [20], a fuzzy PID
step response was extracted. In [9], a min–max opti- controller to achieve a zero overshoot step response
mization approach to determine the optimum loca- has been presented. The CRA and CDM approaches
tion of zeros to attain a minimum-overshoot transient have been employed to design non-overshooting PID
response was employed. In [10], a compensator for controllers for some real plants [21–23]. In [24], a mod-
special case of minimum phase systems to attain a non- ified version of the PID controller, called I+PD con-
overshooting closed-loop system step response was troller, has been designed using process step response
designed. A rational two-parameter controller to and damping optimum criterion. In this approach, the
CONTACT Mohammad Tabatabaei tabatabaei@iaukhsh.ac.ir Department of Electrical Engineering, Khomeinishahr Branch, Islamic Azad
University, Isfahan, Iran
© 2018 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted
use, distribution, and reproduction in any medium, provided the original work is properly cited.
AUTOMATIKA 401
The PID controller with the following transfer func- is that the first and third terms in (11) are greater than
tion is considered: or equal to zero. Thus we have
1 Ti (T − Td )(T + Td + 2KKp Td ) ≥2TTd (T + KKp Td ).
C(s) = Kp 1 + + Td s , (3)
Ti s (12)
where Kp , Ti and Td are the proportional gain, integra- Ti (1 + 2KKp ) ≥ 2KKp T. (13)
tor and derivative time constants, respectively.
The closed-loop transfer function is calculated as The realization of (12) requires the satisfaction of the
following two relations:
KKp (Ti Td s2 + Ti s + 1)
H(s) = . Td ≤ T. (14)
Ti (T + KKp Td )s2 + Ti (1 + KKp )s + KKp
(4)
2TTd (T + KKp Td )
According to (4), the closed-loop system is stable if the Ti ≥ . (15)
following relations are satisfied (T − Td )(T + Td + 2KKp Td )
To establish (13), the following condition should be met
KKp > 0, Ti (1 + KKp ) > 0, Ti (T + KKp Td ) > 0.
(5) 2KKp T
Ti ≥ . (16)
Since K > 0, the following relations could be derived 1 + 2KKp
from (5) Finally, attaining a stable closed-loop system with non-
Kp > 0, Ti > 0, T + KKp Td > 0. (6) overshooting and non-undershooting step response
requires the following conditions:
This means that for the closed-loop system stability, Kp 0 < Td ≤ T, Kp > 0, Ti > 0 (17a)
and Ti must be positive but Td could be negative. On
the other hand, to avoid undershoot in the closed-loop 2KKp T
system step response the numerator coefficients of (4) Ti ≥ (17b)
1 + 2KKp
should be positive. This means that
T 2 (ω24 − ω14 ) + Td2 T 2 ω12 ω22 (ω22 − ω12 ) The monotonically decreasing condition (1) gives
+ (2KKp (Td − T) + 1)(ω22 − ω12 ) > 0. (24) Ti2 (ω26 − ω16 ) + Ti4 Td2 ω14 ω24 (ω24 − ω14 )
Considering (21), the sufficient condition for establish- + Ti3 (Ti − 2Td )ω12 ω22 (ω24 − ω14 )
ment of (24) is
+ Ti2 (2AKKp Td + A2 − 2B − 2KKp )(ω24 − ω14 )
2KKp T − 1
Td ≥ . (25) + Ti3 (−2Td (A2 + AKKp Td − 2B − 2KKp )
2KKp
+ Ti (A2 + 2AKKp Td − 2B − 2KKp
Finally, if the following conditions are fulfilled, a closed-
loop system step response with zero overshoot and − B2 Td2 − 2BKKp Td2 ))ω12 ω22 (ω22 − ω12 )
undershoot could be obtained + Ti (Ti (B2 + 2BKKp ) − 2AKKp )(ω22 − ω12 ) > 0.
Kp > 0, Td > 0 (32)
2AKKp rewritten as
Ti ≥ (34d)
B2 + 2BKKp
Kp > 0, Ti > 0, Td > 0 (40a)
2KKp + 2B − A2 A2 − 2B
Td ≥ (34e) Kp ≤ (40b)
2AKKp 2K
AKKp
A2 + 2AKKp Td − 2B − 2KKp Td <
B2 + 2BKKp
− B2 Td2 − 2BKKp Td2 > 0 (34f)
A2 K 2 Kp2 + B(B + 2KKp )(A2 − 2B − 2KKp )
+
B2 + 2BKKp
(40c)
2Td (A2 + AKKp Td − 2B − 2KKp )
Ti ≥ .
A2 + 2KKp Td (A − BTd ) − 2B − 2KKp − B2 Td 2
(34g) KKp
Ti > (40d)
(A + KKp Td )(B + KKp )
Inequality (34f) could be rewritten as
2AKKp
(B2 + 2BKKp )Td2 − 2AKKp Td Ti ≥ (40e)
B2 + 2BKKp
2
− A + 2B + 2KKp < 0. (35)
controller with three design parameters, the following selected. Only the value of Td should be selected such
isodamping property should be fulfilled, too. that inequality (d2) will be realized. However, this
parameter determines the settling time of the closed-
d
∠[G(jω)C(jω)] ω=ωc = 0. (42) loop system step response and the maximum amplitude
dω of the control signal.
This means that the phase variations around the Finally, for the second-order plant (27), the gain
crossover frequency are negligible. This yields a closed- crossover frequency and isodamping property lead to
loop system robust to gain variations. the following relations between PID controller param-
Applying conditions (41) and (42) on plant (2) yields eters:
the following relations between PID controller param-
eters Ti ωc (B − ωc2 )2 + A2 ωc2
Kp = . (47)
K (1 − Ti Td ωc2 )2 + Ti2 ωc2
Ti ωc 1 + T 2 ωc2
Kp = . (43)
K (1 − Ti Td ωc2 )2 + Ti2 ωc2
−b1 ± b21 − 4b0 b2
Ti = , (48)
−a1 ± a21 + 4a2 T 2b2
Ti = , (44)
2a2 where
where A(B + ωc2 )
b0 = ,
(B − ωc2 )2 + A2 ωc2
a1 = 1 + T 2 ωc2 + 2TTd ωc2 , b1 = −(1 + 2b0 Td ωc2 ), b2 = ωc2 (b0 (1 + Td ωc2 ) − Td ).
(49)
a2 = ωc2 (T − Td )(TTd ωc2 − 1). (45)
Incorporating (17), (43) and (44) yields the PID con- Conditions (40) and constraints (47) and (48) yield a
troller parameters for plant (2). This means that the one-dimensional parameter region in terms of parame-
three-dimensional region obtained from conditions ter Td which could be considered as the solution region.
(17) is converted to a one-dimensional region in terms
of parameter Td . Several PID controller parameters Remark 6: The gain crossover frequency ωc should be
may fulfil these relations. One of these controllers chosen such that inequalities in (40) are satisfied. Sub-
could be selected by the designer. The obtained solu- stituting (47) and (48) in (40) will not lead to straight-
tions could vary by changing the crossover frequency forward inequalities in terms of Kp , Ti , Td and ωc . Thus
parameter. finding an appropriate value for ωc for satisfaction of
conditions (40) could be performed by software.
Remark 4: The gain crossover frequency ωc should In the next section, the ability of the so-designed
be appropriately selected such that inequalities (17) non-overshooting PD and PID controllers for the con-
are fulfilled. This could be realized by trial and error. trol of the mentioned plants is verified through some
Moreover, this parameter determines the transient experimental and simulation tests.
response speed of the closed-loop system. Increasing
ωc decreases the settling time of the closed-loop system
3. Experimental and simulation results
step response. Moreover, increasing ωc increases the
control signal amplitude. Thus this parameter should In this section, the performance of the proposed PID
be selected such that a satisfactory transient response and PD controllers is investigated. Three examples are
with permissible control signal will be obtained. given to show the effectiveness of non-overshooting
For plant (18), applying (41) causes the following PID controllers.
constraint on the PD controller parameters in (19)
Example 1: Consider a modular DC servomotor sys-
ωc 1 + T 2 ωc2 tem in Figure 1. As shown in Figure 2, a permanent
Kp = . (46)
K 1 + Td2 ωc2 magnet DC motor coupled with a tachometer to mea-
sure its angular velocity and a position potentiometer
This converts the admissible two-dimensional param- to measure its position is considered. An Advantech
eter region obtained from (26) to a one-dimensional A/D and D/A interface is employed to implement the
region in terms of parameter Td . designed controllers in MATLAB real-time environ-
ment. To verify the effect of the load disturbance, a
Remark 5: For the PD controller design for plant (18), magnetic brake causing angular velocity decrement is
any arbitrary gain crossover frequency ωc could be provided.
406 M. TABATABAEI AND R. BARATI-BOLDAJI
where ω(t) is the motor angular velocity and v(t) is θ (s) 21.113
the voltage applied to the servo amplifier block. The G(s) = = , (51)
V(s) s(0.1s + 1)
reference angular velocity is selected as 1000 rpm.
Figure 3 shows (Ti − Td ) region with different val- where θ (t) is the motor shaft position and v(t) is the
ues of Kp . According to (17), the parameter Td should voltage applied to the servo amplifier block.
be in the range (0, 0.1]. But Ti is considered in the
arbitrary range (0, 10] due to drawing limitations. The shaft position set value is chosen as 60◦ .
Among all possible controllers, a controller satisfying The two-dimensional non-overshooting PD controller
AUTOMATIKA 407
Figure 5. The control signal for Example 1. Figure 7. The two-dimensional region for PD controller param-
eters in Example 2.
4. Conclusion
This paper presents a novel analytical tuning method
for the family of PID controllers. The design method
is based on adjusting the PID controller parameters
Figure 10. The Ti − Td region with different values of KP in
Example 3. to avoid overshoot in the closed-loop system step
response. This is accomplished by achieving a mono-
tonically decreasing closed-loop system frequency
response. Moreover, the loop gain phase is adjusted
to be flat around the desired gain crossover frequency
which makes the closed-loop system robust under gain
uncertainties. The simulation results on a second-order
plant and the experimental results on a laboratory DC
position and speed servomechanism demonstrate the
efficiency of the proposed non-overshooting PID con-
troller, as well.
Disclosure statement
No potential conflict of interest was reported by the author(s).
References
[1] Bennett S. The past of PID controllers. Annu Rev Con-
trol. 2001;25:43–53.
Figure 11. The effect of gain uncertainty in unit step response [2] Grimholt C, Skogestad S. Improved optimization-based
of Example 3. design of PID controllers using exact gradients. Comput
Aided Chem Eng. 2015;37:1751–1756.
[3] Fruehauf PS, Chien LL, Lauritsen MD. Simplified IMC-
is assumed in the arbitrary ranges (0, 35]. The range of PID tuning rules. ISA Trans. 1994;33(1):43–59.
Td is obtained according to (40c). With an appropriate [4] Vilanova R. IMC based robust PID design: tuning
guidelines and automatic tuning. J Process Control.
gain crossover frequency, the PID controller parame- 2008;18(1):61–70.
ters are selected according to relations (40), (47), (48) [5] Ho WK, Hang CC, Cao LS. Tuning of PID controllers
and (49). The obtained controller coefficients are given based on gain and phase margin specifications. Auto-
in Table 1. Figure 11 compares the nominal unit step matica. 1995;31(3):497–502.
response with those obtained with ±10% uncertainty [6] Mikhalevich SS, Baydali SA, Manenti F. Development
of a tunable method for PID controllers to achieve the
in parameter K. This uncertainty does not have any
desired phase margin. J Process Control. 2015;25:28–34.
effect on step response overshoot. The obtained rising [7] Astrom KJ, Hagglund T, Hang CC, Ho WK. Automatic
time and settling time for nominal response are given tuning and adaptation for PID controllers – a survey.
in Table 2. Figure 12 shows the control signal. Control Eng Pract. 1993;1(4):699–714.
AUTOMATIKA 409
[8] Phillips SF, Seborg DE. Conditions that guarantees no Proceedings of the 17th International Conference on
overshoot for linear systems. Int J Control. 1988;47(4): Automation and Computing; Huddersfield, September
1043–1059. 2011. p. 116–120.
[9] Moore KL, Bhattacharya SP. A technique for choos- [21] Chaoraingern J., Numsomran A., Suesut T., Trisuwan-
ing zero locations for minimal overshoot. IEEE Trans nawat T., Tipsuwanporn V. PID controller design using
Automat Contr. 1990;35(5):577–580. characteristic ratio assignment method for Coupled-
[10] Darbha S. On the synthesis of controllers for continuous tank process. Proceedings of the IEEE Conference
time LTI systems that achieve a non-negative impulse on Computational Intelligence for Modeling, Control
response. Automatica. 2003;39(1):159–165. and Automation, Vienna, Austria; November 2005. p.
[11] Darbha S, Bhattacharya SP. On the synthesis of con- 590–594.
trollers for a non-overshooting step response. IEEE [22] Chaoraingern J., Vaidee W., Trisuwannawat T., Tip-
Trans Automat Contr. 2003;48(5):797–800. suwanporn V., Numsomran A. The design of PID con-
[12] Bement M, Jayasuriya S. Use of state feedback to achieve troller for track following control of hard disk drive
a nonovershooting step response for a class of non- using Coefficient Diagram Method. Proceedings of the
minimum phase systems. J Dyn Syst Meas Control. SICE Annual Conference, Tokyo, Japan; September
2004;126(3):657–660. 2011. p. 2949–2954.
[13] Naslin P. Essentials of Optimal Control. Cambridge [23] Hajare V.D., Patre B.M. Design of PID controller
(MA): Boston Technical Publishers Inc.; 1969. based on reduced order model and Characteristic Ratio
[14] Manabe S. Coefficient diagram method. Proceedings of Assignment method. Proceedings of the IEEE Interna-
the 14th IFAC Symposium on Automatic Control in tional Conference on Control Applications, Hyderabad,
Aerospace, Seoul, Korea; 1998. p. 199–210. India; August 2013. p. 1270–1274.
[15] Kim YC, Keel LH, Bhattacharyya SP. Transient response [24] Pavković D, Polak S, Zorc D. PID controller auto-tuning
control via characteristic ratio assignment. IEEE Trans based on process step response and damping optimum
Automat Contr. 2003;48(12):2238–2244. criterion. ISA Trans. 2014;53(1):85–96.
[16] Rachid A, Scali C. Control of overshoot in the step [25] Vrančić D, Strmčnik S, Juričić Ð. A magnitude opti-
response of chemical processes. Comput Chem Eng. mum multiple integration tuning method for fil-
1999;23:S1003–S1006. tered PID controller. Automatica. 2001;37(9):1473–
[17] Lu YS, Cheng CM, Cheng CH. Non-overshooting 1479.
PI control of variable-speed motor drives with slid- [26] Vrančić D, Strmčnik S, Kocijan J, de Moura Oliveira PB.
ing perturbation observers. Mechatronics. 2005;15(9): Improving disturbance rejection of PID controllers by
1143–1158. means of the magnitude optimum method. ISA Trans.
[18] Bagis A. Tabu search algorithm based PID controller 2010;49(1):47–56.
tuning for desired system specifications. J Franklin Inst. [27] Papadopoulos KG, Tselepis ND, Margaris NI. Type-
2011;348(10):2795–2812. III closed loop control systems-digital PID con-
[19] Mohsenizadeh N., Darbha S., Bhattacharyya S.P. Syn- troller design. J Process Control. 2013;23(10):1401–
thesis of PID controllers with guaranteed non- 1414.
overshooting transient response. Proceedings of the [28] Papadopoulos KG, Papastefanaki EN, Margaris NI.
50th IEEE Conference on Decision and Control and Explicit analytical PID tuning rules for the design
European Control Conference, Orlando, FL; December of type-III control loops. IEEE Trans Ind Electr.
2011. p. 447–452. 2013;60(10):4650–4664.
[20] Saeed B.I., Mehrdadi B. Zero overshoot and fast [29] Kessler C. Ein Beitrag zur TheorieMehrschleifiger Reg-
transient response using a fuzzy logic controller. ulungen. Regelungst. 1960;8:261–266.