An improved human-simulated intelligent control

algorithm for bidirectional DC-DC converter

Shanlu Zhang Lei Li*

College of Automation College of Automation
Nanjing University of Science and Technology Nanjing University of Science and Technology
Nanjing, China Nanjing, China
Email: [email protected] Email: [email protected]

Abstract—There are some limitations in the conventional bus voltage [4]. From the early 1980s,Eidirectional DC-DC
algorithm of Human Simulated Intelligent Control (HSIC). It converter for the first time applied to the battery charger, so
was neither accurately simulate the human control thought, different types of topology has been raised, the control
nor adequately reflected the predictability of the system method continues to improve and has to intelligent, digital
response characteristics, with under the behavior and function control era. The performance of the converter not only can
of multi-modes control as well as multi-objectives decision-
make the storage system to obtain better voltage and current
making. On this basis , this paper proposed an improved
waveform, but also provide protection and stable, so that
Human Simulated Intelligent Control algorithm, according to
the size of the error to choose the gain coefficient, it can
without voltage shock and current impact in the energy
improve the controller to predict the dynamic response of the conversion process.
system and to predict the trend so that the system more This paper is organized as follows. In Section II,
precision and fastness. Finally, a 3kW prototype is built and introduces the human intelligent control theory and
the simulation results and experimental results is conducted to prototype basic algorithm. In Section III, then proposes an
verify the effectiveness of the proposed improved algorithm. improved human simulated intelligent control algorithm. In
Section IV, successfully builds the simulation platform in
Keywords—Human Simulated Intelligent Control algorithm;
Matlab /Simulink; Bidirectional DC-DC converter
Matlab / Simulink environment, present simulation results
and experimental results, respectively. Finally, In Section V
concluded this paper. A 3kW prototype is built and both
I. INTRODUCTION the simulation and experimental results is conducted to
With the development of power electronic technology, the verify the feasibility of the proposed improved algorithm.
controlled object tends to be complicated, showing a high II. THE THEORY OF HUMAN-SIMULATED INTELLIGENT
degree of non-linearity and uncertainty. These complex
CONTROL
objects are difficult to describe with precise mathematical
models and conventional PID control based on models is Human simulated Intelligent Control was proposed by
also difficult to deal with the needs of increasingly complex Zhou Qijian the of Chongqing University, in the early 80s of
control objects. On this basis, many improved control the 20th century. The basic idea is from the behavioral
algorithms are proposed, such as artificial neural network, function and structural function imitation. Human Simulated
fuzzy control, expert system and genetic algorithm and so on. Intelligent Control has developed decades of development
However, the complexity of its control increases the and absorbed the essence of other control theories[5].It has
difficulty of system design. Therefore, this paper proposed elucidated the feature model, feature recognition, feature
an improved human intelligent control method to solve the memory, stable monitoring and so on from the humanoid of
complex control problems. The prototype algorithm of view, which has formed a relatively complete theoretical
human simulated intelligent controller was presented by system and system controller design methods, and in practice
Professor Zhou Qijian of Chongqing University in 1979, and has been widely application.
was officially published in 1983[1]. With the support of the The Human Simulated Intelligent Control theory
National Natural Science Foundation of China, more than 20 combine artificial Intelligent with computer technology. The
years of efforts have been made by Li Zushu and others to human behavior function is studied on the basis of the
construct the theoretical system and the system controller macroscopic simulation of human control structure [6]. The
design method of human simulated intelligent control, and biggest characteristic is the starting from the lowest level of
have been successful in many practical applications [2]-[3]. hierarchical intelligent control system. The artificial
Bidirectional DC-DC converter in the distributed intelligence technology is not only used for high levels, but
household photovoltaic power generation system can provide also for running control level. The dominant idea of the
users with a stable and reliable power supply, the existence human simulated intelligent control theory is to further study
of the battery can also be photovoltaic components of the the function of physical - dynamic intelligence control
power generation peak load pad and smooth direct current behavior on the basis of macro - simulation of human control
*
Corresponding Author, Email: [email protected]. Supported by structure and to simulate it. This intelligent control algorithm
Natural Science Foundation of Jiangsu Province (BK20161499). is based on the intelligent imitation of observation, memory,

978-1-7281-5169-4/20/$31.00 2020
c IEEE 789

Authorized licensed use limited to: Saveetha Engineering College. Downloaded on November 06,2024 at 05:10:10 UTC from IEEE Xplore. Restrictions apply.
 decision, and so on. It determines the control strategy e
according to the error and derivative of error. Human- em1
Simulated Intelligent Control not only has the characteristics
of parallel, logical control and language control, such as em3 t
fuzzy control, expert control and so on, but also has the
characteristics of quantitative control of traditional control em2
based on mathematical model [7]-[9]. Human simulated u
intelligent control in the structure and behavior should uo1 uo2 uo3
follow the four basic principles: (1) Hierarchical information t
processing and decision making; (2) Online feature
identification and feature memory; (3) Combined Control of
Open-Closed Loop and Combination of Qualitative Decision (a)The suppression coefficient κ is bigger
and Quantitative Control; (4) Heuristic and intuitive
e
em1
reasoning applications [10]. em3
em2 t
III. THE PROTOTYPE ALGORITHM OF HUMAN-SIMULATED
INTELLIGENT CONTROL
The basic algorithm of human simulated intelligent u
control is based on the observation and decision making of uo1
uo2
the skilled operator, and the control strategy is determined t
according to the adjusted error and the trend of change.
Therefore, it is close to the way of thinking [11]. When the
control error of the control system tends to increase, the (b)The suppression coefficient κ is smaller
Fig. 2. The dynamic characteristics of humanoid controller prototype
controller emits a strong control effect and suppresses the
increase of the error. When the error returns to zero, the Figure 2. is the dynamic characteristics of the Human
controller reduces the control effect and waits for the simulated intelligent prototype. It shows that the current
observation system Change. At the same time, the controller system is affected by the step interference, the output
constantly record the extreme value of the error, correct the response deviation from the given value, the humanoid
control point of the controller to meet the changing intelligent controller in the error input under the action of the
requirements. The prototype algorithm of Human-Simulated control output and the corresponding response to the error
Intelligent Control: response.
­ n −1 When the suppression coefficient κ is made too big, as
° K p e + κK p iΣ=1 e m ,i (e ⋅ e > 0 * e = 0  e ≠ 0) shown in Figure 2. (a), the dynamic response of the error
u=® n −1
(1) will exhibit an attenuation oscillation of zero. The
°κK Σ e (e ⋅ e < 0 * e = 0) corresponding output of the controller exhibits an inverted
¯ p i =1 m ,i
jagged shape centered on the value uoi. When the error
Where, u is the control output, Kp is the scale factor, κ is changes to the direction of the deviation from the given
the suppression coefficient (0 ӊ κ ӊ 1), e and ơ are the error value, the controller output will superimpose a control
and error rate of change, em, i is the error ith peak. increment proportional to the error on the hold value uoi, u =
uoi + Kp·e. When the error changes to a given value, the
Figure 1. shows the characteristics of the error phase plane controller outputs hold the value uoi. As the error curve
and the corresponding control mode and static characteristics. decays, the hold value uoi will tend to a stable value. When
When the system error is in the first and third quadrants of the suppression coefficient k is made small, as shown in (b),
the plane of the error phase, e· ơ > 0 or e = 0 and ơ  0 . And the error curve will appear as an attenuated oscillation form
then the human simulated intelligent controller is operated in of zero. The controller output exhibits a zigzag curve on the
the proportional control mode; and when the error is in the side of the hold value uoi. Similarly, when the error changes
second and fourth quadrants of the error phase plane, that is, from the direction of the given value, the controller outputs a
e ·ơ < 0 or ơ = 0, the human simulated intelligent controller is control increment Kp·e proportional to the error on the hold
operated to keep the control mode state. value uoi, that is, u = uoi + Kp·e. Similarly, when the error
changes to a given value, the controller outputs the hold
ơ value uoi. As the error curve decays, the hold value uoi will
tend to a stable value. Therefore, when the suppression
factor κ is obtained, the error curve will not oscillate and
return to zero once.
e·ơη0
Figure 3. is diagrams of static and dynamic characteristics,
shows that (a)static characteristics; and (b)dynamic of the
Human simulated intelligent control and compared with the
traditional PID control characteristics respectively, the
e·ơζ0
following for its analysis:
(1) In the oa , when the system given a step change, the
system output is less than the given value, that e > 0, ơ <0,
e · ơ <0, the human simulated intelligent control adopts the
Fig.1. Diagrams of characteristics of the plane error phase

790 2020 15th IEEE Conference on Industrial Electronics and Applications (ICIEA)

Authorized licensed use limited to: Saveetha Engineering College. Downloaded on November 06,2024 at 05:10:10 UTC from IEEE Xplore. Restrictions apply.
 not enough to reflect the predictability of the system
u response characteristics and the behavior and function of
a multi-modes control and multi-objectives decision-making
[11]-[14]. Therefore, this paper proposes an improved
human simulated intelligent control algorithm, which is a
discrete control algorithm:
c b
ef ­ ­°K1 K p en (| en | ≥ M , en ⋅ Δe > 0 * Δe = 0)
g °u0 (n − 1) + ®
° °̄K p en (0 <| en |< M , en ⋅ Δe > 0 * Δe = 0)
u(n) = ®
° ­°K1 K 2 K p em,n (| en | ≥ M , en ⋅ Δe < 0)
d °u0 (n) = ®°̄K K e (0 <| en |< M , en ⋅ Δe < 0 * Δe = 0)
t ¯ 2 p m, n
em2 o em1 ˄2˅
(a) Static characteristics of the humanoid intelligent controller
 Where (2)ˈen is the current value of the error; en-1 is the
u
previous period of the error; em, n is the nth extreme; Ǽe =
uPID en-en-1; Kp is the proportional gain; K1 is the amplification
factor (K1 ˚1); K2 is the suppression coefficient; M is the
uHSIC absolute error amplitude limit; n is the natural number,
indicating the control cycle number; u (n-1) is the holding
1
value of the output of the previous control cycle.
e t
The main improvement of the algorithm for the basic
em1
control algorithm is to divide the proportional gain
segmentation, that is, the absolute value of the error | en | set
a threshold value M, when | en | <M, only Kp as a
proportional gain, When K1Kp> Kp, so that the error is large,
o a b c d e the system has a stronger control effect, to accelerate the
t
tracking response of the system. The result of this
(b) Compared with the traditional PID control segmentation allows the system to take into account the high
quality and speed of the system at a higher level. The human
Fig.3. Diagrams of static and dynamic characteristics simulated intelligent control flow chart is shown in Figure 4.
control mode, and the u > u human simulated intelligent
HSIC PID
A. Simulation analysis
control has faster response speed.
Under the action of unit step input, the unit step
(2) In the ab , when the overshoot occurs, that is, e < 0, ơ simulation of the second-order inertial plus pure lagged
< 0, e · ơ > 0, the human simulated intelligent control uses object in industrial engineering is carried out, and then
the proportional control mode, because HSIC allows a much compared with the traditional PID controller and the
larger proportion than PID coefficient, so u to a very fast HSIC prototype human intelligent controller, the improved human
rate down to a smaller value than the u , which effectivelyPID simulated intelligent control the effectiveness of the control
suppress the increase in overshoot. algorithm.
(3) In the bc, when the system output reaches the
overshoot limit em1 (ơ = 0), u is not expected to be too large
in the negative direction, causing a large callback or even
multiple oscillations. The HSIC uses the hold mode and
maintains the u with an appropriate value by the effect of
HSIC

the suppression factor, and the change rate of the control

error e is not too large, so that there is a smaller oscillation
than the PID control. This retention has the characteristic of
integral, but since this memory only works for extreme
points, it eliminates errors and does not cause the controller
to be too saturated. So this proportional control mode and
hold control mode alternately, making the human simulated
intelligent control faster than the PID control response speed
and smaller overshoot, to solve the traditional control theory
of stability, accuracy and speed ability cannot be balanced.
IV. AN IMPROVED HUMAN-SIMULATED INTELLIGENT
CONTROL
 The prototype algorithm of human simulated intelligent
control  adopts proportional control and maintain control
mode, which reflects the basic control thought of human.
However, from the point of view of improving performance, Fig.4. Block diagrams of the human simulated intelligent control
the prototype algorithm still has some limitations, which is

2020 15th IEEE Conference on Industrial Electronics and Applications (ICIEA) 791

Authorized licensed use limited to: Saveetha Engineering College. Downloaded on November 06,2024 at 05:10:10 UTC from IEEE Xplore. Restrictions apply.
  Based on the simulation results, this paper will use the
improved human simulated intelligent control algorithm to
control the DC-DC bidirectional converter, and compare the
PID control and prototype control under the same
conditions to further verify the effectiveness of the
improved algorithm in the photovoltaic power generation
system. Figure 7 is a block diagram of a bidirectional DC-
DC converter control architecture based on improved
human intelligence.

Fig. 5. Simulation of human intelligent control platform


Fig.8. Bidirectional DC-DC converter prototype experimental platform.

Figure 8. for the development of a power of 3kW

bidirectional DC-DC converter prototype experimental
platform. Circuit parameters are as follows: P = 3kW, L =
 3.6mH, C1 = 1100ȝF, C2 = 1600ȝF, IGBT is K40T120,
switching frequency is 10kHz. Figure 9. (a), (b),
Fig .6. Waveforms of simulation results
(c) ,(d) ,(e)for the experimental waveform.
Assume that the following model is selected for the Figure 9. (a) based on PID control under the load
controlled object transfer function (sampling time d = 0.1s): switching process, the output voltage and current changes,
e −0.5 s visible voltage and current have a greater fluctuation; and
G p ( s) = (3)
(5s + 1)(2s + 1)
Switch point2 Switch point2
According to the human simulated intelligent controller u
u
u (50V/Grid)

u (50V/Grid)
i (1A/Grid)

i (1A/Grid)
flow chart, the step signal as input, according to the
Switch point1
controlled object and the required parameters, the human Switch point1
i i
simulated intelligent control platform shown in Figure 5.
Simulation result shows in Figure 6.
 As we can see the simulation results show that the t/ (500ms/Grid) t/ (500ms/ Grid)
improved human simulated intelligent controller has the    㸦a㸧PID    㸦b㸧Prototype algorithm HSIC
characteristics of fast response, overshoot and control
stability compared with the ordinary PID controller.
Switch point2
u
u (50V/Grid)

B. Experiment analysis
i (1A/Grid)

Switch point1
i

t/ (500ms/Grid)
㸦c㸧Improved of HSIC
current

current
u (200V/Grid)

u (200V/Grid)
i (5A/Grid)

i (5A/Grid)

uH=630V uH=630V
uL=500V uL=500V

t/ (2.50us/Grid) t/ (2.50us/Grid)
㸦d㸧 Start the process (e) Switching the process
Fig.7. block diagram of Bi-directional converter control architecture
based on improved human intelligence. Fig.8. Waveforms of the experimental .

792 2020 15th IEEE Conference on Industrial Electronics and Applications (ICIEA)

Authorized licensed use limited to: Saveetha Engineering College. Downloaded on November 06,2024 at 05:10:10 UTC from IEEE Xplore. Restrictions apply.
 figure 9. (b) based on the prototype HSIC controller [12] LI Zhonghao Huang Yijun Zhang Peichao et al. “Research on
experimental waveform, in the same test figure 9. (c) is to automatic generation control strategy of large scale air combustion
coordinated equivalent power plant”, Power system protection and
improve the experimental waveform of the HSIC controller, control 2016 4ऄpp.40-50
and there is no fluctuation. Figure 9. (d) No surge and [13] Guo Qian, Lin Liaoyuan, Wu Hongyan, et al. “Distributed power
oscillation during start-up; figure 9. (e) No impact during control strategy of micro grid considering adaptive virtual
bidirectional switching. The experiment proves that the impedance”, Automation of electric power system 2016, 19ऄpp. 23-29.
improved humanoid intelligent control algorithm has better [14] LI Zhonghao‫ ޖ‬Huang Yijun‫ ޖ‬Zhang Peichao et al. “Research on
control effect in the control process of bidirectional DC-DC automatic generation control strategy of large scale air combustion
converter. coordinated equivalent power plant”, Power system protection and
control 2016,4‫ޣ‬pp.40-50.
V. CONCLUSIONS
The paper mainly introduces the human simulated
intelligent control theory and the prototype basic algorithm.
Hence, an improved algorithm is presented. Compared with
the conventional algorithm, simulation results demonstrated
the effectiveness of the proposed improved algorithm. At the
same time, through the bidirectional DC-DC converter
control test, further verify the algorithm has advantages of
respond quickly and the smaller overshoot. The next step
will focus on the behavior and function of multi-level
hierarchical structure, online feature recognition, memory
and learning, perceptual reasoning, multi-modes control and
multi-objectives decision-making from the perspective of
humanoid and imitation.
REFERENCES

[1] Zhou Qijian, Bai Jiangguo. “An Intelligent Controller of Novel

Design.Proceedings of a Multinational Instrument Conference”.
ShanhaiˈChina: China Instrument and Control Societyˈ1983,137-
149
[2] WANG Pei-jin. “Implementation of human like intelligent controller
[J]. Journal of Computer Engineering and Applica-tion”, 2002,
38(13): 45-48.
[3] LiZu-shu, Zhang Hua, Wen Yong-ling , Wang Guiping. “Human
simulated intelligent control based on sensory-motor intelligent
schemas”. Hang zhou, China. Fifth World Congress on Intelligent
Control and Automation ( IEEE Cat .No . 04EX788). 2004:2423-7.
[4] WANG jing, LIU Ting-ting. “An analysis of the strategies for human
simulated intelligent control system in complex process”.2017 29th
Chinese Control And Decision Conference (CCDC) .pp.7062-
7065,2017.
[5] Aihui Wang, Dongyun Wang and Shengjun Wen. “Robust Nonlinear
Multivariable Tracking Control Design for A Human-Simulated
Robot Arm using Operator Approach”, Proceeding of the 11th World
Congress on Intelligent Control and Automation Shenyang, China,
June 29 - July 4 ,pp.2590-2595,2014
[6] Dong xie, Guorong Chen,Jie Li, “Research on the Human Simulated
Intelligent PID Control Method for Servo Feed Driver System”, 2018
IEEE 17th Int,I conf.on Cognitive Informatics & Cognitive
Computing .pp.460-464,2018.
[7] Rui Li, Hongli Zhou, Xi Li, Ze Zhang, and Xiaojie Wang. Human
Simulated Intelligent Control on Magneto-Rheological Vibration
Isolation System for Subway Floating Slab Track with Moving Load .
[8] R.S.Liu, Y.J. Gao, Y.L. Yang,et al. “Human Simulated Intelligent
Control with double-direction dead-zone compensation for join
motion control of a large-sized boom system”. Journal of Control
Science &Engineering,Vol.24,pp.1-7,2015.
[9] P.Xu,J.Xiao,S.Li and Y.Yang. “Human Simulated Intelligent Control
for permanent Magnet Synchronous Motor”, Applied Mechanics and
Materials,Vols.241-244,pp.1821-1827,2013.
[10] L.Zhang and S.Z.Hu, “The Research on the Human Simulated
Intelligent of the Inverted Pendulum Based on Feedback
Linearization”, Advanced Materials Research,Vol.710,pp.511-
514,2013.
[11] A. Wang, M. Deng, and D. Wang, “Operator-based robust control
design for a human arm-like manipulator with time-varying delay
measurements”, International Journal of Control, Automation and
Systems, vol. 11, no.6, pp. 1112-1121, 2013.

2020 15th IEEE Conference on Industrial Electronics and Applications (ICIEA) 793

Authorized licensed use limited to: Saveetha Engineering College. Downloaded on November 06,2024 at 05:10:10 UTC from IEEE Xplore. Restrictions apply.