Design of PID Controller Based on an Expert System

Transcription

1 International Journal of Comuter, Consumer and Control (IJ3C), Vol. 3, No.1 (014) 31 Design of PID Controller Based on an Exert System Wei Li Abstract For the instability of traditional control systems, mathematical models were oversimlified. Adative PID controller design exert was used for the simulation study. Simulation results in this aer show that for various erformances, an exert PID control system is better than fixed arameters PID control system. The adjustment time of Exert control system is short and fast convergent, and the eak is small. Fixed arameters PID control system is difficult to determine the suitable mathematical model and the control system arameters. 1. Introduction Traditional control techniques are generally based on the roduction rocess, a mathematical model based on the controlled object. However, the mathematical model can not reach the actual control requirements. The intelligent control can be a good solution to the traditional control techniques, but they are facing roblems. The exert control technology is an imortant art of an intelligent control technology. Exert systems roosed will be alied to the PID controller. However, the research of PID self-tuning controllers is still in its infancy. Currently the develoment of self-tuning PID controller with a style exert erformance is almost entirely deendent *CorresondingAuthor:Wei Li ( @QQ.COM.) School of Electrical Engineering Jilin Engineering and Technical Teachers College Changchun,13005,China on the design and develoment of staff s access to exert exeriences. For the above roblems, in this aer, the PID controller uses exert systems combinedwith an adative PID controller. The controller inherits a simle PID controller structure, which has good stability and reliability, and is easy to adjust, etc., Turn-automatic control theory and artificial intelligence are combined to solve more comlex control roblems. By comaring the simulation exeriment, it is found that an exert adative PID control system withvarious erformances was significantly better than PID control system of fixed arameters. Regulation time of an exert system is short, and fast convergent, so the eak is small without overshoots.. PID Controller Exert System Princiles.1 Exert System An exert system is an intelligent comuter rogram system. Its interior contains a large number of exert levels of knowledge and exerience in a articular field, which enables the knowledge and methods to solve roblems of human exerts dealing with the roblem areas. In other words, the exert system is a lot of exertise and exerience of rogram systems. It uses artificial intelligence techniques and comuter technology, and has based on years of exerience and exertise in certain areas of one or more exerts, reasoning and judgment, analog thinking, and human exert decision-making rocess. In order to solve comlex roblems, it

2 International Journal of Comuter, Consumer and Control (IJ3C), Vol. 3, No.1 (014) 3 requires a human exert handling. For examle, there are many skillful doctors, and their resective areas of work have a wealth of ractical exerience. A secific area of medical exerience together with somehow reresenting atterns stored in the comuter is used to form the knowledge base. Then in the thinking rocess, exerts use this knowledge for diagnosis and treatment of diseases comiled by rogram of inference engine. Diagnosis and treatment of diseases can make your comuter think like a human exert, and then the system like an exert system.. Exert System Structure and the Princile of PID Controller The rincile of automatic control system architecture with an exert-tye self-tuning PID controller is shown in Figure 1. You can see that the system consists of a curve recognizer, some comarators, exert systems, PID controllers and disturbed, the curve of the time characteristics of the system recognizer shows error e identified, identifying each of the lurality of rocess arameters of the resonse curve of the amount e i, i = 1,,..., m (Eg overshoot, daming ratio, damed oscillation eriod, rise time, etc.). The measured values of the characteristic arameters of the comarator are comaredwith the value of the characteristic arameter. Its deviation is fed into the exert system, so inferred line exert system is used to eliminate the correction amount of deviation of the feature quantity arameters of the controller should T i, which are fed into a conventional PID controller to fix the arameters of PID controller. At the same time, the controller and the oeration of the system according to the error e, and tuning arameters, u outut control signals to the controlled object, and rocess until the resonse curve characteristic arameters to meet the requirements of a user s exectations. other accessories. If the closed-loo system is Figure 1: exert-tye self-tuning PID controllers automated system structure diagram

3 International Journal of Comuter, Consumer and Control (IJ3C), Vol. 3, No.1 (014) Exert PID Control Algorithm to Achieve ey Technical Systems 3.1 Exerts PID Controller Algorithm We know that the unit ste resonse error curve of a tyical second-order system is shown in Figure. In a tyical second-order system ste resonse, for examle, timely change analysis of PID arameters imacts on the dynamic rocess at all stages. Figure : A tyical ste resonse of second-order system 1). when e (t) emax,system error is large, In Zone III, No matter what stage movement, it ensures that the system has fast resonse by using the Bang-Bang control, That controller by ositive or negative maximum outut with the fastest seed is used to reduce systematic errors. ). When emid e (t) <emax, System error is still large, In Zone II, control of imlementation is still strong, so absolute error decreases raidly. However, due to the large angular velocity of the area into the initial threshold, deceleration control strategy is adoted to make the system smooth into region I. 3). when emin e (t) <emid,absolute value of the systematic error haens unlikely in Zone I. Segmented variable arameters PID control algorithm is used toimrove the control recision of the system. A cycle system resonse by error and error rate of change e and eare divided into four rocessing sections (see Figure 3). In the AB hase, when e> 0, e <0, system error decreases. To ensure that the system has fast resonse, roortional gain arameter should be greater in the initial segment, However, to reduce overshoots, it is desirable that when the error e is gradually reduced, it also decreases roortional gain, This makes the system seed decrease, so it will not have a big overshoot. Small differential gain arameter should be increased gradually, and this revents the system from affecting the seed of resonse inhibition overshoot. In the BC hase, e <0, e <0, I.e., when the direction of increasing the error e changes, the roortional gain arameter gradually increases;continuing to increase the differential gain arameter will therefore increase the reverse control action andreduce the overshoot. At this oint the system oututs a ositive dearture from the desired value. It should always strengthen the integral action.

4 International Journal of Comuter, Consumer and Control (IJ3C), Vol. 3, No.1 (014) 34 In the CD hase, e <0, e > 0, i.e., when the error e is reduced to change the direction, it will gradually reduce the roortional gain arameter, The system will be as soon as ossible back to the steady-state region ( e (t) <emin area threshold), At this oint the system outut is tending to the desired value, soit should be aroriate to introduce integral action and derivative action. In the DE stage, e> 0, e > 0, i.e., when the direction of increasing the error e changes, it will gradually increase the roortional gain arameter; At this oint the system outut deviates from the desired value, so it should always strengthen an integral role in the introduction of aroriate derivative action. In short, thinking exert PID control is in error, and error rate of change of e e constraint on the PID arameters at different stages of intelligent tuning adats to the dynamic erformance, and steady recision control system is high. 3. Resonse Curve Characteristic of Automatic Identification Resonse curve is characteristic of the automatic identification system through second-order error - time characteristic curve analysis.to get the system overshoot, daming ratio, rise time, oscillation eriod, and the time arameters are adjusted. As a real-time control system, the arametersare different from system arameters of unchanged PID control system. They refer to the arameters of the transient rocess. These transient arameters can be obtained from the relationshi between the individual and the individual eaks. Therefore, the resonse characteristic curve becomes the automatic recognition of the error - time curve to identify and record the eak time. For discrete systems, if the resence of disturbances and calculation errors, identifying and recording the eak time is very simle; for examle, for each samle time k, aaaa error is comared with the revious samling time error bbbb,and records are denoted by e( k 1) < 0 e(k) time - time curve eaks.. When e( k) < 0,,the error may be determined as the

5 International Journal of Comuter, Consumer and Control (IJ3C), Vol. 3, No.1 (014) 35 Figure 3: Automatic identification block diagram 3.3 Tuning Rules Established Automatic identification rocedures found error - time characteristic curve eaks, so we need to classify the waveform, and inference reasoning judgment under secific circumstances to determine with which subscritionsthe rule is tuning. Our exerts talk about self-tuning regulator waveform divided into four grous in the study: no eak waveform, a single eak or double eak waveform, and divergent oscillation daming waveform shae. As the self-tuning exerts now do not have a mature theory, and various waveforms tuning rules are not yet mature, we can only use the rules of self-tuning PID exert advice by consulting or tuning asects of the theory under the guidance of a large number of PID tuning arameters to get the test. Moreover, the rules so obtained also have some scoes. To the extent, according to the selected erformance indicators, by binding assay system to change the set value, the outut value and the control amount, and recycling tuning exerience of the oerator, PID arameter change should change direction, and gradually be adjusted.

6 International Journal of Comuter, Consumer and Control (IJ3C), Vol. 3, No.1 (014) 36 The following are tuning exerience we found and some PID data access given by exerts, the attenuation ratio of 4:1 corresonding erformance indicators. 1).Withthe ratio of about 1/ of the critical value, it changes direction deending on the attenuation ratio and overshoot. When overshoot is greater than the redetermined value, we reduce or increase the roortion of degrees. ).The integration time constant is close to 1/ of the critical eriod of the system. 3).The derivative time constant is equal to the integration time of 1/3 to 1/4. 4). Integration time and the ratio of the role of mutual comensation are alied. A lot of testing under the guidance of the theory of PID arameter tuning, and combined with the above exertise,we got the exert tuning rules wanted. 4. Exert System PID Controller Simulation Analysis 4.1 Simulation Environment objectis Let the transfer function of the controlled G s) = s s s ( 3 Simulation exeriments were conducted in Simulink environment, and Discrete S-function controller with Simulink module was imlemented as a combination. Simulink system rogram is shown in Figures 4 and, 5. Figure 4: PID controller simulation exert system structure diagram Figure 5: exert system PID controller

7 International Journal of Comuter, Consumer and Control (IJ3C), Vol. 3, No.1 (014) The Simulation Process and Results Analysis First rocess control system regulator arameter is set to ure roortional action ( i = 0, d = 0 ). The system gets into oeration, and then the roortional gain is gradually adjusted from small to large. Incase i = 0, d = 0, = 0.,0.4,0.8,1.0,1.5,.0, simulation results are under Figure 6(d) =1. the following circumstances: Figure 6(e) =.0 Figure 6(a) =0. Figure 6(b) =0.4 Figure 6(f) =1.56 The curves can be obtained from the above; the oscillation amlitude curves corresonding between 1.5 and.0 should be adjusted. The curve amlitude oscillations occur. In this case the corresonding is 1.56, and at this oint T k = Figure 6(c) = s. Through a critical roortional method, the PID arameters can be obtained as follows: =0.978, =., = i d

8 International Journal of Comuter, Consumer and Control (IJ3C), Vol. 3, No.1 (014) 38 Figure 7 shows the corresonding curves. All asects of the simulation to adjust the gain coefficient can be drawn with conventional PID, controller otimal control arameters and resonding erformance indicators, as shown in Table 1 Figure 7: The outut resonse curve Table 1: in resonse to control arameters and erformance J1 = e dt J = e t dt J 3 = e dt J 4 = e t dt The curve basically meets the requirements, and it can be used as the initial value of the controller arameters. An exert tuning PID controller is used to control the object, and increase disturbance in.5s time; the simulation results are shown in Figure 8: change rocess is shown in Figure 9: Figure 9: curve changes It can be seen from Figures 8 and 9 that when the system arameters change, the system can automatically select the PID controller based on the Figure 8: system outut resonse curve measured curve tuning arameters until a satisfactory resonse curve.

9 International Journal of Comuter, Consumer and Control (IJ3C), Vol. 3, No.1 (014) 39 Table : exert tuning PID control system erformance corresonding resonse J1 = e dt J = e t dt J 3 = e dt J 4 = e t dt The four functions are the minimum erformance index as the control system with the best control erformance criteria. Comaring Table 1 and Table, it can be seen that the exert system control erforms better. 5. Conclusion By exert control system simulation in MATLAB, We get the following conclusions: when the system characteristics changesfor various erformances, an exert PID control system is better than fixed arameters PID control system. The adjustment time of an exert control system is short and fast convergent, and the eak is small. Under the system not with dramatic oscillations, the exert PID control system enables the system to maintain long-term stable oeration. References [1]. LSusanto-Lee, R. ; Fernando, T. ; Sreeram, V..Simulation of fuzzy-modified exert PID algorithms for blood glucose control. Control, Automation, Robotics and Vision, 008, : []. TongjuanLiu ; Xiangguo Ma ; Nengqiang Jin. Exert PID control study of Hybrid maglev systems. Mechatronics and Automation, 009,: [3]. Sheng-boQi ; Cheng-rui Zhang ; Hong-linXie. Using exert PID control in hydraulic-based AMT system to reduce shift vibration. Mechatronics and Embedded Systems and Alications (MESA), 010, : [4]. RuiLin ;Zhenhua Wang ; Rongchuan Sun ; Lining Sun. Vision-based mobile robot localization and maing using the PLOT features.mechatronics and Automation (ICMA), 01, Page(s): [5]. Tajjudin, M. ; Rahiman, M.H.F. ; Ishak, N. ; Adnan, R. ;Ismail, H. Comarison between otimally-tuned PID with self-tuning PID for steam temerature regulation. Intelligent and Advanced Systems (ICIAS), 01,PP: [6]. Xiaoing Wang ;Yunliang Zhao ; Yuanqing Liao.Dynamic erformance analysis of PID controller with one memristor. Information Science and Technology (ICIST), 011, PP: [7]. Atef Saleh Othman Al-Mashakbeh," Proortional Integral and Derivative Control of Brushless DC Motor", Euroean Journal of Scientific Research 6-8 July 009, vol. 35, g [8]. JianfuGao,ChongzhaoHan,YangwangFang.The ory and alication of nonlinear control System.Xi'anJiaoTong University ress.001.4

10 International Journal of Comuter, Consumer and Control (IJ3C), Vol. 3, No.1 (014) 40 [9]. Ying H.(1998b),An analytical study on structure, Stability and design of general nonlinear Takagi-Sugeno fuzzy control systems Automatica, 34(1): Wei Li received the B.S.degrees in Automation from Dalian Maritime University in 003, M.S. degree in control engineering from Dalian University of Technology in 007.In 010,he joined the Jilin Teachers Institute of Engineering&Technology in 003,where he is cuttentlya Associate Professor and associate dean in electric work academy.her current research interests are Comuter control technology, Electrical Control Technology, intelligent controland their alications.