The 6th International Conference on Cyber and IT Service Management (CITSM 2018)

Inna Parapat Hotel – Medan, August 7-9, 2018

Performance of Manual and Auto-Tuning PID

Controller for Unstable Plant - Nano Satellite
Attitude Control System
Hani Hazza.A1, Mashor, M.Y2, Mohammed Chessab Mahdi3
1,2 School of Mechatronic Engineering, Universiti Malaysia Perlis, Pauh Putra Campus, 02600 Pauh Putra, Perlis, Malaysia
3 Al-Furat Al-Awsat Technical University, Iraq
[email protected]

and derivative elements. The PID control technique has been

Abstract- Over the last decades, control system and its design heavily used as the main control technique in many systems,
have been advancing rapidly to respond to the competition and it soon has become the lead control system since its debut.
demand of the industrial world, which are the key element Two PID controller techniques were more famous than others
in defining new successes. Therefore, pilot plants were created
to explore and examine the effects of changing the conditions
including the step reaction curve experiment, and the closed-
of a process, especially to control unstable systems. To execute a loop “cycling” experiment that execute their tests around the
similar process, this work is directed towards the design, nominal operating point [2].
development and implementation of an auto-tuning The earliest attempt at a full PID-type controller, was
Proportional Integral Derivative (PID) Controller for introduced by Elmer Sperry in 1911. His theories around PID
Innovative Satellite InnoSAT control attitude system. Auto- controllers was made known by Nicolas Minorsky who
tuning PID controller was designed using MATLAB, Simulink published Sperry’s analysis in 1922. The findings were based
as well as the optimization of PID controller without excessive on his efforts to design an automatic steering systems for the
mathematic calculations. This technique had faced few U.S. Navy. In addition, a number of PID setting formulae
stumbling blocks in cases where the analysis and design were
gained popularity like the Chien Hrones– Reswick formula,
monotonous since the resolution involves trial and error. To
overcome this issue, it was necessary to offer access to Cohen– Coon formula, Ziegler– Nichols tuning, Wang–
computers and programmable calculators to approach the Juang–Chan formula and Zhuang–Atherton optimum PID [3].
tedious mathematical calculations using MATLAB software. The Ziegler– Nichols tuning formula in particular has methods
The biggest asset from any developed program is its ability to recognize the equivalent first-order plus dead time model, a
to showcase the compensated and uncompensated relationship key element in the algorithms used for several PID controller
as well as absorb how these relationships work directly from designs.
the date showed. The calculated results would then be This paper argues that contrary to the common believe
confirmed by using manual, MATLAB and Simulink’s manual tuning method has proven to yield better results in the
calculations. The paper argues that human intervention during
the trial and error process provides better control over unstable
situations for unstable plants, where achieving the required
systems when compared with auto-tuning. In this paper, the stability for auto and adaptive tuning can be challenging.
value of Kp, Ti and Td will be calculated using manual calculation Taking into considering that Ziegler-Nicholas settings resulted
while step response graph for each cases will be solved using in good responses for the integrating process but poor
Simulink. performance coupled with delay, in addition to Ricera et al in
1986 experience with SIMple Control (SIMC) that resulted in
Keywords — PID Controller; Auto tuning; Manual tuning; poor disturbance responses for integral process[4], this paper
InnoSAT; Controlled System; shows that manual control can provide a maneuver to change
PID values without upsetting the output.
I. INTRODUCTION In the past, the manual calculation associated with the
tuning process had led to a time-consuming process that was
A control system comprises of subsystems and process
additionally limited by the availability of specific criteria. The
connected in order to achieve a specified output and
process of auto-tuning provides a minimal influence of the
performance when given a quantified input. To measure its
human factor when selecting the setting, running, or
performance, two measures are needed namely the transient
supervising the process. A specific algorithm is utilized for
response and the steady-state response. In addition, the system
self-tuning process, in particular during the subsequent
can be configured with additional controllers to enhance the
iterative cycles[5]. The developments in this field and the
stability when responding to the specified conditions [1]. This
increasing use of self-tuning algorithms have led to faster
additional step is known as Proportional-Integral-Derivative
implementation of industrial undertakings. A driver processes
(PID) control technique that involved the proportional, integral
 The 6th International Conference on Cyber and IT Service Management (CITSM 2018)
Inna Parapat Hotel – Medan, August 7-9, 2018

the data speedily to provide results in a fraction of time, which differentiation of the error. The equations of InnoSAT [7]
contributed to the advancement of its industrial uses. Since plant can be shown as
PID controllers gained this popularity, it is important to
discuss the three values (PID controller gains) used to
‫ ݏ‬ଶ ൅ ͲǤ͵Ͳͷͳ‫ ݏ‬൅ ͲǤʹͲͶͲ
calculate the output of PID controllers [6]. Alternatively, the ߶ሺ௦ሻ ൌ
world witnessed the development of the Internal Model Control ‫ ݏ‬ସ ൅ ͳǤͳͲͷͲ‫ ݏ‬ଶ ൅ ͲǤͳ͸ͷͲ
ͳ
(IMC), also known as Lambda tuning, and alternatively the ߠሺ௦ሻ ൌ ଶ
‫ ݏ‬െ ͹Ǥͳͳ͵ͺ ൈ ͳͲିଷ (2)
most recent SIMC rules that formed the commonly used
‫ ݏ‬ଶ െ ͲǤ͵Ͳʹ͵‫ ݏ‬൅ ͲǤͺͲͺͺ
algorithms of control systems. Terry and Willy explored the ߰ሺ௦ሻ ൌ ସ
‫ ݏ‬൅ ͳǤͳͲͷͲ‫ ݏ‬ଶ ൅ ͲǤͳ͸ͷͲ
strengths of these model based approaches, and highlighted its
ability to affect the performance and sturdiness through the use
of tuning parameters. Tuning parameter are used to establish a
III. METHODOLOGY
trade-off between performance and robustness, while
coordinating responses to follow each loop to reach control A. PID Design
objectives. Manual Tuning gives quick responding system
As discussed, the PID controller and its feedback
often offers more adequate responses when compared with the
results recorded by auto-tuning the same system, this is mechanism has been the most used controller in industrial
particularly true when we try to control an unstable system. applications [8]. Although it has straight forward control
algorithms, it still provides a solution in the absence of
II. PID CONTROLLER specific knowledge around the implemented process. The
typical structure of the PID control system is shown in Fig.1,
The following PID algorithm illustrate this relation and
identifies the output as u(t): where Kp is the proportional gain ,Kd is the derivative gain
, and Ki is the integral gain. The desired output can be reached
by continuously adjusting these gains. Shown in Equation (1)
(1)
is the mathematical description of the PID controller, the
and the tuning parameters are: results from the error signal e(t) are assessed and calculated to
1. Kp: Proportional gain – larger Kp typically means faster form the control signal u(t) applied to the plant model. In the
response since the larger the error, the larger the equation, u(t) is the input signal to the plant model , the error
larger the Proportional term compensation. However, an signal e(t) is defined as e(t) = r(t) − y(t) , and r(t) is the
excessively large proportional gain will lead to process reference input signal.
instability and oscillation. The proportional action is calculated and executed in relation
2. Ki: Integral gain – larger Ki implies steady state errors are to the detected error and processed to decrease the rise time.
eliminated quicker. The trade-off is larger overshoot as The integral action focuses on the past error and is used to
negative error integrated during transient response must be reduce the steady state error [9]. Lastly, the derivative action
integrated away by positive error before we reach steady detects future behavior of error and increases the stability by
state. reducing the overshoot and improving the transient response.
The tuning process is implemented by assessing the
parameters Kp, Ki , and Kd, that give an appropriate closed-
loop response. To reach the desired response, other factors
need to be considered including the measures of settling time,
oscillation period, and overshoot, to name a few.

B. PID Tuning Method

Tuning is the process of calculating the relevant PID
parameter values to achieve an optimal performance from a
process. Tuning is the most important part in closed loop
Fig.1. PID Controller for InnoSAT control attitude system
control systems, and has a number of approached that can
enhance and accelerate the performance. However, all
approaches still involve conducting an experimental
3. Kd: Derivative gain – larger Kd decreases overshoot and determination of the characteristics of the control loop. Using
these characteristics, the tuning parameters are defined and
speed up transient response. However, it may lead to
can consequently result in achieving the goal performance [10].
instability due to signal noise amplification in the
The 6th International Conference on Cyber and IT Service Management (CITSM 2018)
Inna Parapat Hotel – Medan, August 7-9, 2018

Kp, Ki , and Kd are essential to the controller process if these D. Manual Tuning
parameters were chosen incorrectly, instability might The PID controller depends on the simple process of trial
dominate process input resulting in disorders of outputs, and error. This process is relatively easy to understand as soon
regardless of oscillation, restricted only by saturation or as the PID parameters are set [12].
mechanical breakage. Therefore, the process of loop tuning is The integral and derivative values should be set to zero, and
the adjustment of these parameters and their gains and then the proportional gain is increased to reach the point
optimizing their value to achieve the required control reaction, where the output of the control loop equivocates at a constant
more specifically, gain/proportional band, integral gain/reset, rate. If the increases of proportional gain are not affecting the
derivative gain/rate stability of the system, the responses of the system will
accelerate.
C. Auto-Tuning When the proportional response reaches the appropriate
PID tuning approaches has been developed over time and a speed, the integral value is then set. The equivocation
number has been introduced including the renowned Ziegler resulting from first process will start to witness a reduction.
Nichols tuning method, all the way to the recent Simple At this point, the integral value should be changed to reduce
Internal Model Control (SIMC) by Skogestad (2003) [11]. the steady state error. However, this process might also
increase overshoot.
The SIMC method for PID controller tuning will attempt to
After setting the proportional and integral values to achieve
tune the gains of PID controller automatically in Simulink a minimal steady state error, the overshoot of the system
environment. The following block diagram shows the Simulink response will diminish as a result of increasing the derivative
set up for InnoSAT attitude control using PID controller system value.
with auto tuning as shown in Fig.2. Fig.3 shows the block diagram of the Simulink set up for
InnoSAT attitude control using PID controller system

Fig. 2. PID Controller for InnoSAT control attitude Fig.3. PID Controller for InnoSAT control attitude system with manual
system with Auto Tuning. tuning

Also, the gain co-efficient parameters for PID controller Also, the gain co-efficient parameters for PID controller with
with auto-tuning is shown in Table I. manual tuning is shown in Table II.

TABLE I TABLE II
Gain Co-Efficient Parameters For PID Controller With Auto-Tuning Gain Co-Efficient Parameters For PID Controller With Manual-Tuning
Controller Controller
Parameter Parameter

Roll 16.979 6.152 8.164 Roll 23 0.56 6

Pitch 0.1515 0.0063 0.8334 Pitch 18.5 0.04 4.35

Yaw 0 0.0293 0 Yaw 20 0.5 5

 The 6th International Conference on Cyber and IT Service Management (CITSM 2018)
Inna Parapat Hotel – Medan, August 7-9, 2018

IV. SIMULATION RESULTS AND DISCUSSION controller shown in TABLE II. The stimulation
demonstrates that after several steps of manual tuning
To consider making a decision on the appropriate process the system achieves a correct control process with
methodology to control an unstable plant, we make the less overshoot in a relatively short time of regulation.
selection on the bases of the new PID-parameters. As shown Autotuning method as shown in Fig.4. provided
in TABLE II, the manual tuning method provides fast step unaggressive responses with higher overshoot, the angle
responses with small overshoot, as illustrated by Fig.5. On ߰ሺ௦ሻ in Fig.4. were unable to reach a steady state level
the other hand, the parameters from the auto tuning method providing unacceptable result for InnoSAT attitude
in TABLE I showed a comparatively poor responses. To control system. In addition, the method produces zero values
clarify further, Fig. 4. demonstrates the response produced for Kp and Kd (TABLE I) which contributes to inaccurate
by PID Controller for InnoSAT control attitude system with response as shown in Fig. 4 . Comparatively, the angle ߰ሺ௦ሻ
Auto-tuning and Fig. 5 shows the step response produced provided more steadiness proving that the Manual tuning
by PID Controller for InnoSAT control attitude system provides a method with limited complexity that offers
with manual-tuning. The reference signal was introduced as smaller margins for oscillation during the tuning process
a step signal to the input, the resulting response shown in the in comparison with the auto tuning method as
process monitor were analyzed to determine the quality demonstrated in Fig.5. It can also prompt the system to
indicators. identify optimum PI settings as the ones shown in
Chosen simulation results are shown below in Fig. 4 and TABLE II.
Fig. 5, both show the process of self-tuning for PID
controller shown in TABLE I and manual tuning for PID
Response

Time(s)

Fig. 4. Step response obtain by PID Controller for InnoSAT control attitude system with Auto- tuning
Response

Time(s)

Fig. 5.Step response obtain by PID Controller for InnoSAT control attitude system with manual tuning.
The 6th International Conference on Cyber and IT Service Management (CITSM 2018)
Inna Parapat Hotel – Medan, August 7-9, 2018

V. CONCLUSION

The human intervention during the process of trial and error

to control an unstable system, provides efficient and
optimized control over the process when compared with the
auto-tune approach. During the setting of the manual-tune
parameters, it is noticed that the use of a trial and error
method can result in increased stability in the system with
less overshoots and noticeable acceptable settling time as
compared with the auto- tune methods. The results
demonstrated in the paper indicates that human intervention
can in fact optimize the control over the system of the
InnoSAT control attitude system.

REFERENCES

[1] M. W. Foley, R. H. Julien, and B. R. Copeland, “A Comparison of PID

Controller Tuning Methods,” Can. J. Chem. Eng., vol. 83, no. 4, pp. 712–
722, 2008
[2] F. Padula and A. Visioli, Tuning Rules for Optimal PID and Fractional-
Order PID Controllers, vol. 21. 2011.
[3] P. Chang, J. Lin, and K. Cho, “Ziegler-Nichols-Based Intelligent Fuzzy
PID Controller Design for Antenna Tracking System,” Proc. Int. Multi
Conf. Eng. Comput. Sci. Vol II, vol. II, no. 1, pp. 1–6, 2011
[4] R. Vilanova, IMC based Robust PID design: Tuning guidelines and
automatic tuning, vol. 18, no. 1. 2008.
[5] T. L. Blevins, “PID advances in industrial control,” IFAC Proc. Vol., vol.
2, no. PART 1, pp. 23–28, 2012.
[6] V. Chopra, S. K. Singla, and L. Dewan, “Comparative Analysis of
Tuning a PID Controller using Intelligent Methods,” vol. 11, no. 8, pp.
235–249, 2014.
[7] F. A. Bakri, M. Y. Mashor, S. M. Sharun, S. N. Bibi Sarpinah, and Z. A.
Bakar, “Adaptive fuzzy logic controller with direct action type structures
for InnoSAT attitude control system,” IOP Conf. Ser. Mater. Sci. Eng.,
vol. 152, no. 1, 2016
[8] M. C. Mahdi, A. Shehab, and M. J. F. Al Bermani, “Direct Fuzzy Logic
Controller for Nano-Satellite,” vol. 4, no. July, pp. 210–219, 2014.
[9] A. K. Suked and J. Arora, “Auto Tuning of PID Controller,” Int. Conf.
Ind. Instrum. Control, no. Icic, pp. 1459–1462, 2015.
[10] M. C. Mahdi, “Fuzzy PID Controller for Nano-Satellite Attitude
Control,” J. Sci. Arts, vol. 4, no. 4, pp. 407–416, 2016.
[11] S. Skogestad, “Single analytic rules for model reduction and PID
controller tuning,” J. Process Control, vol. 13, no. 4, pp. 291–309,
2003.
[12] T. L. T. Mohamed, R. H. A. Mohamed, and Z. Mohamed, “Development
of Auto Tuning PID Controller Using Graphical User Interface (GUI),”
2010 Second Int. Conf. Comput. Eng. Appl., pp. 491–495, 2010.