Research On Parameters Tuning Of PID Controller And Its Application

PID controllers have been widely used in industrial control systems as a classical control method. With the development of computer technology and artificial intelligence in automatic control field, many methods that about PID controller design have been appeared in recent years.This paper studies PID control theory deeply and designs PID controller with various methods. The paper includes two parts, single variable PID is studied in the first part which from the first chapter to the fourth chapter and multivariable PID is studied in the second part which from the fifth chapter to the sixth chapter.In the first part, the paper introduces the important position of PID in industrial control, illuminates the basic principle of PID controller and the status of PID parameter tuning. Crossover and mutation is improved in the new genetic algorithm in the second chapter, the designs PID controller with the improved genetic algorithm and the PID controller apply to motor control system, form the simulation, it brings to a satisfied control result. Based on BP neural network, the wavelet neural network can be configured when wavelet function substituted sigmoid function. Then, a wavelet neural network PID controller is designed with this algorithm and applies to a control plant, it shows perfect performances. Finally, the paper proposes an algorithm of a kind of multi-object robust PID controller. Regional pole placement, H 2performance and H∞performance are considered in this method, then designs multi-object robust PID controller based on this algorithm. On the other side, a algorithm of multi-object robust optimal H 2PID controller can be deduced from the method, this kind of PID controller can make H 2performance become optimal in closed-loop system at the condition of the given H∞performance, at the same time,the method can make the pole placement in artificial required region.In the second part, first, the paper summarizes various methods that about how to design multivariable PID. The research contents include some algorithms of multivariable PID that use genetic algorithm, neural network, particle swarm algorithm and robust control. Giving specific design flow and approach in each algorithm. Once again, the parameters tuning of multivariable PID base on quasi-diagonal recurrent neural network (QDRNN) is proposed in this paper, then, both the method and multivariable PID algorithm base on diagonal recurrent neural network (DRNN) applies to a two coupling plant, form the simulation, the method in this paper brings to a satisfied control result. At last, the paper summarizes the researched PID control theory and prospects more research directions and contents on the parameters tuning of PID controller.