Research On Mixed Fuzzy PID Based On Neural Network

In the actual industrial control ,PID control is still a major control method ,it is of greatsignificance how easy and effective implementation of the tuning PID parameters. With theincreasing complexity of control systems, the traditional PID control difficult to achievesatisfactory control effect. In advance fuzzy control does not need the exact mathematicalmodel of the object, but it is on the basis of human thought as well as productionexperience ,with language rules describes control process, and in accordance with the rules ofcontrol algorithm to adjust or control parameters.The work of this paper mainly consists of three parts:At first, combinate the mind of fuzzy reasoning in fuzzy control and conventional idea ofPID control, the error and error change as input, set up the initial fuzzy control rule base. Inthis paper, the concept of sequence pairs is proposed, The establishment of sequence pairspremises rules of fuzzy expert, composites he precise control rules table. The use of geneticalgorithm, optimizates the control rule table, and avoiding the initial fuzzy rules on theadjustment table.Second, research on the optimization method rule base of fuzzy control, made a newadjustment of fuzzy table methods. Precise the fuzzy rules, change the precise rule table intosequence pairs. Converse sequence pairs into vectors of fuzzy rules, using the geneticalgorithm , set up the objective function in accordance with kp, ki, kd, precise online settingthe error and error change, implementation of the three parameters of PID tuning on line.Based on sequence pairs, in this paper studies the use of neural network feed back to adjustthe rules of library online, the process of fuzzy table into vector, don't use fuzzification andanti-fuzzification operation. The use of neural network set up adjust precision of the fuzzyrule is conducive to the use of neural network rule base training study.Third, three-dimensional fuzzy control system is faster than second-dimensionalfuzzycontrol system.However,with the increase in dimension, it is difficulty to use the fuzzysystem and three-dimensional fuzzy system can not eliminate static error and limit cycleoscillations. In this paper, three-dimensional will be transformed into integral andtwo-dimensional fuzzy control of the controller in parallel. Analysis of the article the number of control rules, an analysis of the advantages of two-dimensional fuzzy control.