Original from: Puchalski B., Rutkowski T., Duzinkiewicz K.

: Implementation of the FOPID Algorithm in the PLC Controller - PWR

Thermal Power Control Case Study. 2018 23rd International Conference on Methods & Models in Automation & Robotics (MMAR)/ ed.
2018 : IEEE, 2018, p. 229-234. DOI: 10.1109/MMAR.2018.8485807 1
Under a license: CC BY-NC-ND 4.0

Implementation of the FOPID algorithm in the

PLC controller - PWR thermal power control
case study
Bartosz Puchalski, Tomasz A. Rutkowski, Kazimierz Duzinkiewicz Faculty of Electrical and Control
Engineering
Gdańsk University of Technology, G. Narutowicza Street 11/12, 80-233 Gdańsk, Poland
Email: [email protected], [email protected],
[email protected]

Abstract

In the paper authors describe proposition of design and verification procedures of the discrete Fractional Order PID
(FOPID) algorithm for control of the Pressurized Water Reactor (PWR) thermal power near its nominal operating
point. The FOPID algorithm synthesis consists of: off-line optimal tunning of its parameters in continuous time-
domain with LQ (Linear Quadratic) performance index and simplified models of nuclear reactor and control rods
drive; its transformation into equivalent integer order structure with Oustaloup filters; and finally its transformation
into equivalent discrete form. Discrete FOPID algorithm is further implemented in the PLC controller and verified by
real-time simulation in the Hardware In the Loop (HIL) structure with non-linear nuclear reactor model. Promising
simulation results were obtained, which confirm improved flexibility of the discrete FOPID algorithm in comparison
to its classical PID counterpart.

I. I NTRODUCTION

Theory of the Fractional Order Calculus (FOC) involving, non-integer derivatives and integrals, is a well known
field of study [1]. Methods based on FOC are willingly used in order to describe complex system dynamics and/or
to achieve better control quality in various control systems [2], [3]. The Nuclear Power Plants (NPP) with PWR
reactors are an example of such systems. There have been many recent publications presenting the usage of the
FOC in modeling of processes occurring in nuclear reactor’s cores [4], [5], [6] and for control purposes of such
complex plants [7], [8], [9].
Nowadays more than 95% of the control loops in industry process control is based on classic PID (Propor-
tional–Integral–Derivative) algorithm [10]. In most cases the integer order PID controllers are being used due to
simplicity in implementation on digital platforms and due to well-known tuning methods. In order to enhance
performance of the PID algorithm, in more complex and demanding applications, modifications in the structure of
the algorithm can be introduced. In addition various techniques of parameter adaptation can be also introduced to
the PID algorithm in order to control non-linear and non-stationary processes [11], [12], [13], [14], [15].
 2

Despite various improvements in PID algorithms structure and tuning methodologies, all the above solutions have
a subtle type of limitation, which is generally related to the integer order of integration and/or differentiation. A
solution to this problem is the Fractional Order PID (FOPID) or PIλ Dµ algorithm [2]. The FOPID algorithm comes
with two additional degrees of freedom related to integration (λ) and differentiation (µ) order. Various types of digital
and analog applications of the aforementioned FOPID controller with appropriate tuning and design methodologies
may be found in the literature [16], [17]. The FOPID algorithm has also been introduced to the field of NPP control,
for instance in the design of a fractional order phase shaper augmented with an optimal PID controller [18], robust
fractional order PID controller [4] or fuzzy multi-regional controller with local PIλ Dµ controllers [9]. It has been
shown that control algorithms which utilize fractional order operators prove to have better performance in terms
of selected evaluation criteria, for instance ISE (Integral Squared Error) or LQ, in comparison to the integer order
algorithms (PID) that were tuned with corresponding methods.
In the presented paper the authors focus on the optimal design, implementation and verification of the discrete
FOPID algorithm for control of the PWR thermal power near its nominal operating point. Performance of the
algorithm was verified by means of real-time simulation in the Hardware In the Loop (HIL) structure with non-
linear PWR reactor model simulated in Matlab – SDRT environment [19], which cooperated with PLC controller
where the designed discrete FOPID algorithm was implemented.

II. P ROBLEM STATEMENT

The NPPs are complex objects with non-linear processes dynamics with various time scales [20]. They are also
classified as critical infrastructure facilities, therefore control of its basic controlled quantity – thermal power, is not
a trivial task. Typically a classical SISO control system, presented on Figure 1 with conventional PID controller
is used for that purpose. Every effort related to the improvement of the control system efficiency, for instance by
introducing additional degrees of freedom in the form of integral and/or differential operators is greatly desired.
Downloaded from mostwiedzy.pl

In the paper, authors describe discrete version of the FOPID algorithm design procedure, for the PWR reactor
thermal power control purpose. It consists of four main steps presented on the Figure 2. At the beginning, the
continuous FOPID algorithm is subjected to an off-line optimization task in the time-domain according to the LQ
criterion. This task is performed using a simplified nuclear reactor model linearized in operation point that is related
to the nominal working conditions of the PWR reactor – 100% of generated thermal power. Then, fractional order
integral and differential operators, with parameters obtained from optimization are approximated with integer order

Fig. 1. Typical control system structure for reactor thermal power control where: rT H (t), pT H (t) - reference (SP) and realized (PV) reactor
thermal power trajectory, e(t) - error signal, u(t) - control signal (CV) in the form of control rods velocity, x(t) - control rods immersion into
the reactor core.
 3

Fig. 2. Main steps of discrete FOPID algorithm design procedure.

continuous Oustaloup filters [21] in frequency-domain. Suitable Oustalouop filter parametrization related to selection
of the appropriate order and frequency band, guarantee its adequate frequency characteristic which is compatible
with the identified process dynamics and fractional order operators. After that, the Triangle Hold discretization
method [22] is applied to the FOPID algorithm. As a result of this operation a discrete structure of the FOPID
algorithm is obtained. This structure is easily implemented in the PLC controller with a high level Structured Text
(ST) language [23] for instance with Simulink PLC Coder toolbox [24]. Finally, digital FOPID algorithm is verified
via simulation tests in the Hardware In the Loop (HIL) structure. For HIL simulation tests purposes the multi-nodal,
non-linear simulation model of PWR reactor was implemented in Simulink Desktop Real-Time (SDRT) toolbox
[19]. This toolbox allows to simulate wide range of virtual objects in real-time regime on Personal Computer (PC).
In HIL structure, the multi-nodal PWR reactor model, implemented on PC, cooperates via analog I/O acquisition
boards with designed FOPID algorithm implemented in industrial PLC controller.

III. R ESEARCH METHOD

A. Plant models

The mathematical models of the processes occurring in the PWR reactor may be divided into two main groups, the
first one is related to complex and accurate models, while the second one represents less accurate, reduced and simple
Downloaded from mostwiedzy.pl

models [20], [25], [26]. In this paper two such models of PWR nuclear reactor core are used for different purposes.
Firstly the simplified linearized PWR reactor model is used for proposed FOPID control algorithm synthesis stage
(ModelA). This model is a result of simplification of the multi-nodal model. Secondly the multi nodal, non-linear
complex PWR reactor model (ModelB) [25], [26] is used for the HIL simulations that are related to control algorithm
verification stage. The concept of the multi-nodal model is shown on Figure 3.
The multi-nodal PWR reactor model presented on Figure 3 is built based on the following components [25], [26]:
(i) point kinetics model of nuclear reactor core used to describe the time-depended average neutron density, including
six groups of delayed neutrons; (ii) heat transfer model with two coolant nodes (odd and even respectively) assigned
to each i−th distinguished fuel node; (iii) reactivity feedbacks due to changes in the fuel and coolant temperature;
(iv) reactivity feedback due to the control rod bank movements; (v) and the assumption that the thermal power
generated in the reactor core is proportional to the neutron flux and the average neutron density.
The multi-nodal reactor model (ModelB), consists of five fuel nodes and ten coolant nodes. Its simplified version
(ModelA) is characterized by, one average delayed neutron precursor group, one fuel node (i = 1), two coolant
nodes and thermal power distribution coefficient equal to 1 (DCi = 1).
 4

Fig. 3. Multi-nodal nuclear reactor core model structure, where: PT H (t) - reactor thermal power, αF - fuel reactivity coefficient, αC - coolant
reactivity coefficient, i - fuel node index, (2i − 1) - odd coolant node index, (2i) - even coolant node index, TF i (t) - i − th fuel node
temperature, DCi - thermal power distribution coefficient for the i − th fuel node, play role of of weight factors, which values are related to
the depth x of control bank immersion into the reactor core, TC(2i) (t) and TC(2i−1) (t) - even and odd coolant nodes temperatures, TCin (t)
Downloaded from mostwiedzy.pl

and TC(2i) (t) - coolant temperatures at inlet and outlet of reactor core.

In both models a non-linear control rods drive and control rods reactivity models are used
  
d∆ρext (t) ρb 2πx
= 1 − cos , (1)
dx H
e H
e

d∆ρext (t) d∆ρext (t) dx

= · , (2)
dt dx dt

dx
0 ≤ x ≤ H,
e −1, 9 cm/s ≤ ≤ 1, 9 cm/s (3)
dt
where ∆ρext describes the deviation of the external reactivity from its initial, critical value (∆ρext (0) = 0), ρb
denotes control rods reactivity worth when they are fully immersed into reactor core, H
e denotes the reactor core

height. Initial immersion of the control rods x(0) into reactor core is equal to 30% of H.
e Due to restrictions related

to the size of this conference paper the authors did not describe parameters and equations of ModelA and ModelB.
 5

All mentioned PWR reactor models (ModelA, ModelB) descriptions with parameters values may be found in the
following publications [25], [26].

B. FOPID algorithm design

In the field of automatic control the Grunvald-Letnikov definition [1], [2], [3] of the continuous integro-differential
operator is widely used
t−a
[Xh ]  
r −r m r
a Dt f (t) = lim h (−1) f (t − mh), (4)
h→0
m=0
m
where a and t are the limits of the operation, r denote the order of the operation. For r > 0 the operator describes
differentiation and for r < 0 integration. Based on this definition, the FOPID algorithm may be presented in the
continuous time-domain as follows

u(t) = KP · e(t) + KI · 0 D−λ µ

∞ e(t) + KD · 0 D∞ e(t) (5)

where λ and µ are nonnegative real numbers (λ, µ ≥ 0), KP denote proportional gain, KI and KD denotes the
integration and differentiation gains, respectively.
Intuitively, due to two additional parameters introduced in the FOPID controller comparing to the classical PID
equivalent there are more degrees of freedom in adjusting overall control system performance. However, finding
optimal set of values for those parameters, in order to meet the user specification for the given process, is not a
trivial task. In the literature may by fond various propositions for FOPID algorithm tuning strategies, starting from
empirical rules and ending with analytical techniques [3]. In the paper the optimal tuning methodology for FOPID
algorithm has been proposed. Optimal values of FOPID parameters are the solution of a optimization task defined
below in the continuous time-domain with LQ integral quality criterion
Downloaded from mostwiedzy.pl

Z t
e2 (x, τ ) + u2 (x, τ ) dτ,


min fLQ = st. (6)
x 0

x = [KP , KI , λ, KD , µ] and x ≥ 0, (7)

π
|arg(pj )| > α · . (8)
2
In the optimization problem summarized by formulas (6)-(8) symbols e and u represent the error and control
signal as shown on Figure 1. Optimization task utilizes linear models of both the plant (ModelA) and the control
rods drive – actuator. Therefore, the entire model of the control system (Figure 1) used in the optimization task
consisted of linear elements. Hence, in the optimization task a unit step reference thermal power trajectory was
used.
The stability of the controlled system is guaranteed in the optimization process by the inequality constraint de-
scribed by formula (8). Using this constraint, it is examined whether the poles pj of the corresponding commensurate-
order [3] closed-loop control system are in the appropriate region of the complex plane. The commensurate-order
transfer function of the corresponding fractional order system is defined as
 6

b1 δ m + b2 δ m−1 + ... + bm δ + bm+1

G(δ) = , (9)
a1 δ n + a2 δ n−1 + ... + an δ + an+1
where δ = sr , and r is chosen herein as equal to 0,001.
For comparison purposes, corresponding optimization task has been carried out for the classic version of the PID
controller with parameters KP , KI , KD . In this case the constraint (8) correspond to the classical version of the
stability of the LTI systems where all poles of the closed-loop control system must be located in complex LHP. The
off-line optimization results for FOPID and PID controllers are presented in Table I. Mesh Adaptive Direct Search
[27] algorithm with additional search step that utilized Genetic Algorithm [28] was used to solve the optimization
task (6)-(8).

TABLE I
O PTIMIZATION RESULTS

Controller KP KI λ KD µ fLQ
PID 1,284 0,036 n/a 0,000 n/a 1,400
FOPID 0,186 1,076 0,161 0,028 0,543 1,383

C. FOPID algorithm approximation with Oustaloup filter

The numerical procedures typically used to evaluate fractional order integral and differential operators are not
useful, due to their complexity and the need for infinite memory resources. Therefore implementation of the
FOPID controller on PLC platform is impossible using typical FOC numerical procedures. The authors propose to
approximate the fractional order operators D−λ and D−λ of the optimally designed continuous FOPID algorithm
µ
with integer order continuous Oustaloup filters G−λ
O (s) and GO (s) adjusted in the frequency-domain
Downloaded from mostwiedzy.pl

−λ,µ µ
CO (s) = KP + KI · G−λ
O (s) + KD · GO (s). (10)

The single Oustaloup filter [21] is defined as a frequency-band real non-integer differentiator, with fractional
order α and frequency-band in the range ω ∈ hωb , ωh i as follows
!α
1 + ωsb
 α
α ωu
GO (s) = , (11)
ωh 1 + ωsh
√
where ωu denote the unit gain frequency (ωu = ωb ωh ) and α ∈ R. Oustaloup [21] proposed the approximation
0
of non-integer order filter (11) with integer order filter, which real zeros ωk and poles ωk are recursively distributed
over the complex plane. Transfer function of the integer order Oustaloup filter takes the following form [3]
N 0
Y s + ωk
Gα α α
O (s) = lim GO,N (s) = lim ωh (12)
N →∞ N →∞ s + ωk
k=−N

where α is the order of the fractional operator.

Aforementioned Oustaloup filter definition avoids the restriction on ωb ωh = 1 so the frequency boundaries can
µ
be selected independently [3]. The parametrization of Oustaluop filters G−λ
O (s) and GO (s), must be done according
 7

to the appropriate filters orders (2N + 1) and frequency band ω ∈ hωb , ωh i. Appropriate parametrization guarantee
adequate frequency characteristics of the filters which takes into account dynamics of the considered plant and
fractional order operators. Frequency band of the plant was identified with respect to the lowest/highest real values
of the zeros and poles of the linearized ModelA and it is within the range of frequencies hωA , ωB i given in Table II.
To obtain a correct approximation of the fractional operators with the Oustaloup filters the frequency boundaries
of the filters hωb , ωh i must satisfy the following conditions [21]: ωb  ωA , ωh  ωB . The final parameters of the
Oustaloup filters have been listed in Table II.

TABLE II
O USTALOUP FILTERS PARAMETERS

ωb [rad/s] ωh [rad/s] ωA [rad/s] ωB [rad/s] N

0,0006095 36283,35 0,06095 362,8335 4

D. FOPID algorithm discrete representation

During implementation of any control law in the PLC controller it is necessary to take into consideration following
aspects: (i) controller sweep time, which should be constant and adjusted with respect to the fastest time constant
identified in the controlled process; (ii) control algorithm representation should be described in a way, which allows
direct implementation with one of the PLC programming languages; (iii) algorithm computational time should allow
its execution in finite and predetermined period. In the last case the sweep time of the PLC controller should be
considered as constraint during algorithm synthesis and implementation.
In the case of fractional order control algorithm implementation, the description presented previously in subsec-
tions III-B and III-C is still problematic for direct implementation in PLC controller. For that purpose the discrete
Downloaded from mostwiedzy.pl

control algorithm representation is needed. In the literature may be found several methods for fractional systems
direct discretization e.g. PSE (Power Series Expansions) or CFE (Continuous fractional Expansions) [3]. In the paper
the authors propose to use the Triangle Hold discretization method [22]. It is applied to the controller structure with
the integer order Oustaloup filters (10). Finally, the FOPID controller structure may be presented in the discrete
state space form. That form of control algorithm description allows its easy implementation in PLC controller with
e.g. structural text language (ST) [23] using typical data structures such as vectors, basic arithmetic operations such
as addition, subtraction, multiplication and division, basic repetition control structure in the form of FOR-loop and
repeating cycle nature of user program execution in PLC device
During algorithm verification stage the authors observed that proposed discretization method guarantee satisfactory
quality of continuous systems approximation in frequency-domain, and its discrete representation with a minimal
set of parameters in the form of rational numbers in compassion to other popular discretization methods.
According to the Table II, the largest time constant present in a nuclear reactor model is related to frequency
ωB = 362, 83 rad/s. Taking into account Nyquist’s theorem, the digital control algorithm should sample data at a
frequency minimally twice as high as the highest time constant in the model. Therefore, the PLC controller sweep
 8

time should be set to a minimum value of approx. 8 ms. The minimum sweep time that could be achieved on
the PLC controller used in this study was equal to 9 ms. Although the required sweep time was not achieved,
the implemented FOPID algorithm was capable of effective control of the thermal power of a nuclear reactor as
presented in section IV.

IV. R ESULTS

The discrete FOPID algorithm presented in the article has been subjected to simulation tests carried out with the
usage of ModelB. The simulations were carried out in two different environments. Firstly, overall control system
structure presented on Figure 1 was simulated in the Matlab/Simuling rapid prototyping environment – subsection
IV-A. Secondly, simulation tests in the HIL structure presented on Figure 4 were performed and described in
subsection IV-B. HIL structure is characterized by the separation of the control algorithm, in this case implemented
in the PLC controller, from the plant, which in this instance is implemented in the real-time environment. In both
cases the same stepwise trajectory was used which was supposed to cause the control system under consideration
to deviate from nominal point by ±5%.

A. Software simulation results

In this subsection results regarding comparison between different realizations of the FOPID algorithm in the
software simulation environment (Matlab/Simulink) are shown. During the simulation tests, the integrator windup
effect which reduces the quality of control was observed for the FOPID controllers. To overcome this effect
the appropriate Back-Calculation Anti-Windup strategy [10] was applied to the FOPID algorithm with feedback
gain KB = 1/Tt set to 2.61. The analogous Anti-Windup strategy was introduced to the PID/FOPID controller
implemented in the PLC controller which was tested in HIL structure. The KB gain for PID controller was set to
0.05.
Downloaded from mostwiedzy.pl

On Figure 5 responses of the considered in the paper control system with different realizations of FOPID algorithm
are shown. The blue dashed line represent given set point (SP) trajectory of the PWR reactor thermal power. The red
and the yellow lines represent responses of the control system with FOPID algorithm which has been implemented
using continuous Oustaloup filters and discretized Oustaloup filters, respectively. On this figure the response of the

Fig. 4. Hardware In the Loop (HIL) structure.

 9

Fig. 5. Comparison of stepwise trajectory tracking between FOPID algorithms with Anti-Windup strategy tested in software simulation
environment and HIL structure.
Downloaded from mostwiedzy.pl

Fig. 6. Comparison of control signals generated by FOPID algorithms with Anti-Windup strategy tested in software simulation environment
and HIL structure.

HIL control system with the discretized FOPID algorithm implemented on PLC controller has also been plotted
with violet line for comparison.
On Figure 6 the control signals from different realizations of FOPID algorithm are presented. The blue dashed
line represents control signal equal to 0 m/s. Other colors on that figure has similar meaning as on the Figure 5.
As expected, presented simulation tests results show that different implementations (continuous, digital) of the
FOPID algorithm yield the same results, which indicates the correct design and implementation of discrete FOPID
algorithm. The analysis of Figure 6 shows that the control signal from the controller operating in the HIL structure
 10

is noisy which was also to be expected because of the physical, analog connections between the elements present in
this structure. Despite the noise, the nature of this signal is similar to the reference signals generated by continuous
and discrete implementations of the FOPID controller.

B. Hardware In the Loop simulation results

In this subsection results regarding comparison between different types of discrete PID and FOPID algorithms
tested in the HIL structure are shown.
In this case, on Figure 7 responses of the considered control system structure (Figure 4) with various types of
discrete PID and FOPID algorithms implemented in the PLC controller are presented. As previously, the blue dashed
line represents given set point (SP) trajectory of the desired PWR reactor thermal power. The red and the yellow lines
represents responses of the control system with discrete PID algorithms which were provided by the manufacturer
of the PLC controller. Those algorithms were tuned by automatic tunning procedures provided by the manufacturer.
The red line corresponds to the PID algorithm tuned according to the simplified procedure (PreTuning), the yellow
line corresponds to the PID algorithm tuned using the more accurate tuning method (FineTuning). While the violet
and the green lines corresponds to the control system with discrete FOPID and PID algorithms, which were tuned
based on the optimization procedure presented in the subsection III-B. The PreTuning procedure provided by the
PLC manufacturer is based on the open-loop step response method [10]. On the other hand the FineTuning procedure
provided by the PLC manufacturer is based on the closed-loop Relay-Method [10]. On Figure 8 the control signals
from various types of PID and FOPID algorithms in HIL structure are shown. The blue dashed line represents
control signal equal to 0 m/s. The red and the yellow lines, as on previous figures, represents control signals form
PID algorithms, which were provided by the manufacturer of the PLC controller. As in previous figure the violet
and the green lines corresponds to control signals from FOPID and PID algorithms, respectively.
Downloaded from mostwiedzy.pl

Fig. 7. Comparison in stepwise trajectory tracking between different types of PID algorithms and FOPID algorithm working in HIL structure.
 11

Fig. 8. Comparison in control signals generated from different types of PID algorithms and FOPID algorithm working in HIL structure

Presented results show that PID algorithms provided by the manufacturer of the PLC controller are characterized
by the best tracking of the demanded thermal power trajectory. Despite this fact they cannot be considered as proper
due to the very high noise level of the generated control signal. On the other hand FOPID controllers with applied
Anti-Windup strategy which were tuned with the procedure presented in this article do not differ significantly from
the PID controllers provided by the manufacturer. Their advantage is clearly visible in the control signals which
are far less noisy. The classical PID controller was the worst of the controllers that participated in this study due
to poor reference trajectory tracking.

V. C ONCLUSIONS
Downloaded from mostwiedzy.pl

In the article the method of implementation of the FOPID algorithm on the PLC platform was presented.
Presented method of implementation was verified in the HIL environment using a complex PWR reactor model.
Verification phase proved that the implemented FOPID algorithm is comparable to the PID controller proposed
by the PLC manufacturer in trajectory tracking. It should be clearly stated that both algorithms were tuned with
completely different methods. It is also important to state that the presented FOPID algorithm outperformed the
PLC manufacturer PID controller in terms of less noisy control signal. The main drawback, identified by the
authors, is related to computational complexity of the FOPID controller. Presented approach allows to obtain a
very good representation of fractional order operators. Despite the effort spent on implementation, the use of the
FOPID controller result in an improvement in the control quality especially in comparison to the classical PID
approach tuned with the corresponding method. At present, there are very few articles related to the problem
of implementation of fractional order controllers on the basic digital control platform used in industry, namely
the PLC controller. The authors believe that this article is an important contribution related to research on the
possibilities of implementing control algorithms based on fractional order operators on modern digital platforms,
 12

i.e. PLC controllers. In particular, it shows a case in which a very complex plant, such as a PWR reactor, has been
taken into account.

ACKNOWLEDGEMENTS

The research work was done under grant Polish MNiSW 8902/E − 359/M/2017: Young Researcher Support Program. The authors wish to express their thanks for support.

R EFERENCES
[1] K. B. Oldham and J. Spanier, The fractional calculus: theory and applications of differentiation and integration to arbitrary order. Dover
Publications, 2006.
[2] I. Podlubny, Fractional differential equations. Elsevier, 1998, vol. 198.
[3] D. Xue, Fractional-order Control Systems: Fundamentals and Numerical Implementations. Walter de Gruyter GmbH & Co KG, 2017,
vol. 1.
[4] S. Das, S. Das, and A. Gupta, “Fractional order modeling of a PHWR under step-back condition and control of its global power with a
robust P I λ Dµ controller,” IEEE Transactions on Nuclear Science, vol. 58, no. 5, pp. 2431–2441, 2011.
[5] T. K. Nowak, K. Duzinkiewicz, and R. Piotrowski, “Fractional neutron point kinetics equations for nuclear reactor dynamics–numerical
solution investigations,” Annals of Nuclear Energy, vol. 73, pp. 317–329, 2014.
[6] ——, “Numerical solution analysis of fractional point kinetics and heat exchange in nuclear reactor,” Nuclear Engineering and Design,
vol. 281, pp. 121–130, 2015.
[7] S. Das, I. Pan, and S. Das, “Fractional order fuzzy control of nuclear reactor power with thermal-hydraulic effects in the presence of random
network induced delay and sensor noise having long range dependence,” Energy Conversion and Management, vol. 68, pp. 200–218, 2013.
[8] B. Puchalski, K. Duzinkiewicz, and T. Rutkowski, “Analiza sterowania ułamkowego P I λ Dµ moca˛ reaktora jadrowego,”
˛ Informatyka,
Automatyka, Pomiary w Gospodarce i Ochronie Środowiska, no. 4, pp. 63–68, 2013.
[9] ——, “Wieloobszarowa rozmyta regulacja P I λ Dµ mocy reaktora jadrowego,”
˛ in Aktualne Problemy Automatyki i Robotyki, K. Malinowski,
J. Józefczyk, and J. Światek,
˛ Eds. Akademicka Oficyna Wydawnicza EXIT, 2014, vol. 20, pp. 544–557.
[10] K. J. Åström and T. Hägglund, PID controllers: theory, design, and tuning. Instrument society of America Research Triangle Park, NC,
1995, vol. 2.
[11] Z.-Y. Zhao, M. Tomizuka, and S. Isaka, “Fuzzy gain scheduling of PID controllers,” IEEE transactions on systems, man, and cybernetics,
vol. 23, no. 5, pp. 1392–1398, 1993.
[12] B. Puchalski, K. Duzinkiewicz, and T. Rutkowski, “Multi-region fuzzy logic controller with local PID controllers for U-tube steam generator
in nuclear power plant,” Archives of Control Sciences, vol. 25, no. 4, pp. 429–444, 2015.
Downloaded from mostwiedzy.pl

[13] B. Puchalski, T. Rutkowski, J. Tarnawski, and K. Duzinkiewicz, “Comparison of tuning procedures based on evolutionary algorithm for
multi-region fuzzy-logi PID controller for non-linear plant,” in Methods and Models in Automation and Robotics (MMAR), 2015 20th
International Conference on. IEEE, 2015, pp. 897–902.
[14] P. Sokólski, T. A. Rutkowski, and K. Duzinkiewicz, “The excitation controller with gain scheduling mechanism for synchronous generator
control,” in Methods and Models in Automation and Robotics (MMAR), 2015 20th International Conference on. IEEE, 2015, pp. 23–28.
[15] P. Sokólski, K. Kulkowski, A. Kobylarz, K. Duzinkiewicz, T. Rutkowski, and M. Grochowski, “Advanced control structures of turbo
generator system of nuclear power plant,” Acta Energetica, vol. 3, no. 24, pp. 83–90, 2015.
[16] A. Soukkou, M. Belhour, and S. Leulmi, “Review, design, optimization and stability analysis of Fractional-Order PID controller,”
International Journal of Intelligent Systems and Applications, vol. 8, no. 7, p. 73, 2016.
[17] S. Das, I. Pan, S. Das, and A. Gupta, “A novel fractional order fuzzy PID controller and its optimal time domain tuning based on integral
performance indices,” Engineering Applications of Artificial Intelligence, vol. 25, no. 2, pp. 430–442, 2012.
[18] S. Saha, S. Das, R. Ghosh, B. Goswami, R. Balasubramanian, A. Chandra, S. Das, and A. Gupta, “Design of a fractional order phase shaper
for iso-damped control of a PHWR under step-back condition,” IEEE Transactions on Nuclear Science, vol. 57, no. 3, pp. 1602–1612,
2010.
[19] MathWorks, “Simulink Desktop Real-Time,” March 2018. [Online]. Available: https://www.mathworks.com/products/
simulink-desktop-real-time.html
[20] J. J. Duderstadt and L. J. Hamilton, Nuclear reactor analysis. Wiley New York, 1976, vol. 1.
[21] A. Oustaloup, F. Levron, B. Mathieu, and F. M. Nanot, “Frequency-band complex noninteger differentiator: characterization and synthesis,”
IEEE Transactions on Circuits and Systems I: Fundamental Theory and Applications, vol. 47, no. 1, pp. 25–39, 2000.
 13

[22] G. F. Franklin, J. D. Powell, and M. L. Workman, Digital control of dynamic systems. Addison-wesley Menlo Park, CA, 1998, vol. 3.
[23] K. H. John and M. Tiegelkamp, IEC 61131-3: Programming Industrial Automation Systems. Springer Science & Business Media, 2010.
[24] MathWorks, “Simulink PLC Coder,” March 2018. [Online]. Available: https://www.mathworks.com/products/sl-plc-coder.html
[25] B. Puchalski, T. A. Rutkowski, and K. Duzinkiewicz, “Multi-nodal PWR reactor model—methodology proposition for power distribution
coefficients calculation,” in Methods and Models in Automation and Robotics (MMAR), 2016 21st International Conference on. IEEE,
2016, pp. 385–390.
[26] ——, “Nodal models of pressurized water reactor core for control purposes–a comparison study,” Nuclear Engineering and Design, vol.
322, pp. 444–463, 2017.
[27] C. Audet and J. E. Dennis Jr, “Mesh adaptive direct search algorithms for constrained optimization,” SIAM Journal on optimization, vol. 17,
no. 1, pp. 188–217, 2006.
[28] D. E. Goldberg, Genetic Algorithms in Search, Optimization and Machine Learning, 1st ed. Boston, MA, USA: Addison-Wesley Longman
Publishing Co., Inc., 1989.
Downloaded from mostwiedzy.pl