Designing Power Supply using FUZZY controllers

Transcription

1 Volume 1, Number 1, MAY 2014 ARTIFICIAL INTELLIGENCE AND APPLICATIONS Designing Power Supply using FUZZY controllers Baridoon Shabaninia* and Mahdi Kazemi School of Electrical Engineering and Computer Science, Shiraz University, Iran. *Corresponding author: Abstract: This paper presents a method to design a kind of power supply (PS) using FUZZY controllers in two structures, FUZZY control and FUZZY PID tuning control. Here we adopt the control strategy of FUZZY algorithm to have a better response and output comparing with other control strategies like PID controllers. At first we have a fast review about kinds of Power Supplies and then have a bit introduction about methods of FUZZY control which has been used here. Then MATLAB simulation of PID control and FUZZY control and FUZZY PID tuning control and the system without any controllers are compared in this paper. The simulation results show the FUZZY-based systems have a better and faster dynamic response, high control precision, no overshoot and steady state error and also more stability. Their adaptability and robustness are strong. Implementation of these kinds of systems in PS causes to high quality voltage outputs for systems which need safe voltage input. At the end of this paper we ll present the hardware of PID mode based on AVR microcontrollers. Absolutely this schematic circuit will be useful for FUZZY type as we ll explain there. Keywords: None-linear power supplies; FUZZY control; PID controller; Membership functions; Simulation 1. INTRODUCTION 1.1 Power Supply One of the most important parts of every system is their power supplies part. In some system the power will be provided from battery; in other ones, it will be provided from AC signals. AC power supplies are divided into two main kinds; linear power supplies and non-linear power supplies or switching power supplies Linear power supplies These kinds of PSs are not too effective especially in higher powers, in other words the ratio of power to the volume of circuit and its board size is too low. These PSs have a controller element that works in its active region. Generally, it can be series bipolar junction transistor (BJT) with load, as it is shown Figure 44

2 Designing Power Supply using FUZZY controllers Figure 1. Linear power supply Figure 2. None-linear Power Supply 1. Linear power supplies have a simple design. However it has some disadvantages including low efficiency in high powers and large-size circuit, because of the need of the larger heat sinks in high powers, suitability for low powers and etc None-linear power upplies The general schematic of these PSs has been indicated in Figure 2. They work as following: 1. AC home voltage will be rectified and filtered. 2. The signal from previous step will be applied to a switching element. So, a square wave will be produced. 3. The square wave will be applied to load after transforming by a transformer and filtering. These PSs can generate more power in high efficiency and they can achieve to efficiency more than 50% rather than linear PSs. The transformer is designed in square wave frequency, so it is smaller than a 50 HZ or 60 HZ transformer. They can be compacted and they are suitable for systems like portable devices. However, they have some defect such as electrical and magnetic radiation from its transformer. There are three kinds of none-linear power supplies: Fly back or buck-boost, buck and buck-drive. All of them have two modes: voltage and current mode. 45

3 Here we use none-linear power supply using power MOSFET which high power efficiency and hi-speed switching is seemed in their properties [1]. The schematic works as below: When put the 50 HZ AC home voltage at the input of the system this signal is rectified through Diode Bridge while the MOSFETs are on. The gates of MOSFETs are controlled with control part of system to manage the time of firing. According to our desired voltage this time is different. After that we have a Capacitance for filtering voltage passing from bridge. 1.2 FUZZY Control With the continuous improvement for the requirement of system control, PID control system cannot meet the control requirements, and it has exposed its limitations, such as the parameters cannot be set online, dead zone is too large, the positioning accuracy is not high. For nonlinear system and time-varying system, the effect of PID control is inefficient. With the development of control theory, a more advance control strategies, FUZZY PID, was gradually adopted in the control system. Ding Hai-shan [2] using unknown nonlinear function of the fuzzy tree identification method in off-line identification system, improves the control performance. TAN Guan-heng [3] proposed a new optimal FUZZY PID controller that has the best control effect to a given object. This paper will present the simulation of FUZZY PID tuning and FUZZY control in MATLAB and will explain how to implement them on microcontrollers. 2. PID CONTROL SIMULATION DESIGN The output u (t) of the controller is determined not only by proportional to the error signal e(t), but also proportional to the integral of e(t) and the derivative of e(t). So that it is called proportional and integral and derivative controller. It is short for PID controller. Time-domain equations of PID controller is shown in equation U(t) = K p[e(t) + 1 Ti e(t)dt + de(t) ] (1) dt Where Kp is proportional coefficient, Ti is the integral time constant and Td is the differential time constant. When Kp becomes larger, the response of system will be more rapid and the steady-state error of system will be less. But the oscillation is more intense, the adjustment time will be longer. When Kp is too large, the system will tend to be unstable. When Kp is too small, the system response will be tardy. When Ti is small, integral role of the system will be strengthened, which will make system tend to be unstable, but it can eliminate the steady-state error and improve control precision. When Td is large, the overshoot will be larger, but adjustment time will be short. When Td is too small, overshoot will also be large, and adjustment time will be long. Only the parameters Kp, Ti, Td of PID control is set properly, can the performance of control system be more superior. Proportion, integral and differential can make system more stable, dynamic response faster and overshoot less. The PID control principle block diagram is shown in Figure 3 [4]. It can be seen from the figure, the error signal e(t)=r(t)-c(t) adjust by proportion, integral, differential becomes u(t) and u(t) controls the object to decrease deviation signal. SIMULINK is a visualization Control system imitation integrated environment in MATLAB. It is an assistant in modeling and simulation for complex control system. In SIMULINK condition, users can use the mouse and keyboard to complete all the process of the diagram system simulation. And the result can be more intuitive, fast and accurate. Here is PID simulation structure of our system shown in Figure 4.

4 Designing Power Supply using FUZZY controllers Figure 3. Using PID in closed loop Figure 4. PID type SIMULINK in MATLAB The different parts of system could be seen here, such as switches, Diode-Bridge, PID control part and etc. we can input our desired voltage in constant part and according to error signal PID controller send suitable control signal to our plant. 3. SIMULATION DESIGN OF FUZZY PID TUNING CONTROL The FUZZY PID control has a more obvious superiority compared with the PID control, FUZZY PID control is often selected in control nonlinear system and time-varying system. The FUZZY PID control is used to control system that requires high control precision, upper sensitivity, fast response speed. FUZZY PID control utilizes expert adjusting knowledge to establish adjustment rules if-then model, and uses FUZZY logic reasoning to adjust PID parameters in real time, which makes control object to achieve ideal control state. FUZZY controller is the core of FUZZY PID control system. FUZZY controller is made up of fuzzed, rule base, FUZZY reasoning, de-fuzzy. The role of each part is separately as follows: 47

5 Figure 5. Using FUZZY PID tuning in closed loop Figure 6. Inputs and Outputs of FUZZY block the fuzzed transforms accurate quantity of input to FUZZY quantity; rule base is control rule sets that are obtained by experience or other methods; FUZZY reasoning explains and utilizes expert experience of rule base to achieve optimum control; de-fuzzy outputs accurate amount that is converted by FUZZY reasoning conclusion. The structure chart of FUZZY PID control is shown in Figure 5. Above figure shows the PID control system with a FUZZY gain scheduler. The approach taken here is to exploit FUZZY rules and reasoning to generate controller parameters. It is assumed that Kp, Kd and Ki are in prescribed ranges. Appropriate ranges are determined experimentally. In the proposed scheme, PID parameters are determined based on the current error e (k) and its first difference as you see in Figure 6. The changes of the three PID parameters are the outputs of fuzzy block. Based on the previously established rules of the fuzzy control, we can make fuzzy reasoning and modify the value of PID parameters on-line, so as to achieve the PID parameters self-tuning. So that control object has a good dynamic and static performance. The fuzzy PID control simulation is shown in Figure 7. You can see the fuzzy PID controller in the above figure. Firstly, fuzzy PID control method finds the relations among the three parameters Kp, Ki, Kd of PID controller and the value of deviation and the value of deviation change rate. These are continuously detected in the control process, and then the three parameters are modified on-line according the fuzzy 48

6 Designing Power Supply using FUZZY controllers Figure 7. FUZZY PID tuning type SIMULINK in MATLAB control rules to meet the control requirements. In this paper, the domain of fuzzy set of error and error change rate when desired voltage is 10 V is [-10 10] and the domain of fuzzy sets of Kp, Ki, Kd is [-1 1].the word set is[nb,nm,ns,zo,ps,pm,pb] for error, Kp, Ki and Kd. The words set for error change rate is [NB, NS, ZO, PS, PB]. The paper summarizes the parameter setting rules: 1. When E is large, to achieve a stronger dynamic response performance, we can set larger Kp and smaller Kd. To avoid the overshoot of system response being too large, the integral role should be limited. 2. When E is medium size, in order to achieve a smaller overshoot of the system response, Kp can be set smaller. At this time, the value of Kp has greater impact for the system response. The value of Ki should be moderate. 3. When E is small; to make the system stable, the value of Kp and Ki should be set larger. At this point, in order to avoid the system oscillation near the set point, the value of Kd should refer to value of Ec (error change rate).when Ec is small, the value of Kd can be larger [2, 5]. According to the setting principles of Kp, Ki, Kd the fuzzy rules table can be obtained as in Table SIMULATION DESIGN OF FUZZY CONTROL The structure chart of FUZZY controlis in Figure 8. In this part we have error and change rate of error signals as inputs like previous part, but control signals in output. In fact we ve turned the FUZZY LOGIC CONTROLLER Block to PID controller, and each part of this controller play a role of PID parameters. For example the membership function of error 49

7 Table 1. FUZZY rule of Kp, Ki, Kd Figure 8. FUZZY control in closed loop system behave like Kp, and the membership function error change rate behave like Kd, and the defuzzification part like Ki (when we use center of average mode),because of its integral behavior. This type is a bit faster than previous type because of its simplicity and less rules. This simulation is shown in Figure 9. Here the domain of fuzzy sets of error and change rate error are [-10 10], and the domain of fuzzy set of control signal is [-1 1].this simulation is designed when desired voltage is 10V. If we want another voltage at first we input it in constant part and then we should add a scale factor in the way of error signal. The fuzzy rules for this type are shown in Table SIMULATION RESULT 50 The simulated results for all types are shown in Figure PID Controller: FUZZY PID controller: FUZZY control:

8 Designing Power Supply using FUZZY controllers Figure 9. FUZZY control type SIMULINK in MATLAB Table 2. FUZZY rules for second structure E Ec U PB ZO PB PM ZO PM PS ZO PS ZO NB NB ZO NM NM ZO NS NS NB ZO NB NM ZO NM NS ZO NS ZO PB PB ZO PM PM ZO PS PS ZO ZO ZO PB NS PM PS NB NM NB PS NM NS PB PM PS NS ZO NS PS ZO 51

9 Figure 10. PID Controller Figure 11. FUZZY PID controller Figure 12. FUZZY control And here there are simulations in Figure 13 when we didn t use any controller. 52

10 Designing Power Supply using FUZZY controllers Figure 13. No use of any controller Figure 14. Hardware design for PID type 6. SUGGESTION We can implement this system with AVR Microcontrollers. For PID controller type you can see [6].but if you want to implement FUZZY types you should use RS-232 protocol. By using this method the AVR will act as an actuator and MATLAB does the FUZZY works! Here is the hardware design for PID type (Figure 14 ) [6]: Where JP0 is MM74C922, JP9 is LM324 and JP16 is LCD 16*2 [6]. 7. CONCLUSION The design of power supplies using FUZZY controllers has been presented in this paper comparing to the usual kinds of power supplies. We showed this designed power supply has better voltage output and also better quality for devices which need smooth and safe input voltage. Also we showed that we can 53

11 implement these methods on AVR microcontrollers. But we can use microcontrollers which work based on FUZZY logic too. However it is a bit complicated for a power supply but the results will satisfy us to use these methods. References [1] A. S. Adib, Switching power supply, tech. rep., Information university of science and technology publication, [2] H. Ding, J. Mao, and Y. Lin, Indirect adaptive fuzzy control based on fuzzy tree model, Acta Automatica Sinica, vol. 34, no. 6, pp , [3] G.-z. Tan, H.-f. Xiao, and Y.-c. Wang, Optimal fuzzy pid controller with incomplete derivation and its simulation research on application of intelligent artificial legs, Control Theory and Applications, vol. 19, no. 3, pp , [4] N. Munro, The systematic design of pid controllers, pp , [5] Z.-Y. Zhao, M. Tomizuka, and S. Isaka, Fuzzy gain scheduling of pid controllers, in Control Applications, 1992., First IEEE Conference on, pp , IEEE, [6] T. Mohamadi, Designing power supply (ps) using digital pid based on avr microcontrollers, in Design & Test Symposium (EWDTS), th East-West, pp. 1 5, IEEE,