| Summary: | 碩士 === 國立臺北科技大學 === 電機工程系研究所 === 95 === PID controller, which is simple in structure and easy to use and whose parameters possess clear physical meanings, has long been favored by the industry. PID controller parameters have traditionally been adjusted by trial and error, by the experience of the operator, or by experiments. The adjustment consumes time and manpower vastly. Therefore, in recent decades, literature about utilizing the genetic algorithms to design PID controller has increased gradually.
Due to the fact that conventional genetic algorithms lack wide variety in initial generations and the global optimal solution is difficult to find, the mutation rate has decisive impact on gene evolutions. In addition, the policy and method of mutation provide crucial influences to gene evolutions as well. To further improve the mutation behavior, this thesis proposes a modified genetic algorithm that incorporates the adaptation of mutation and the transient response specifications. By limiting the range of gene mutation, the cooperation between gene mating and mutation is improved and the proposed algorithm tends to converge and reach the global optimal solution rapidly. Different from the traditional optimization methods, this thesis introduces specifications of transient response into the fitness function and converts the original optimization problem into unconstrained optimization problem. When the chromosomes can not meet requirements, proper punishment function is added to the fitness function. This modification enhances the adaptability and possibility of survival for the chromosomes, increases the variety among groups and decreases the computation time of genetic algorithm. As a result, the settling time and overshoot are reduced and the dynamic response of the controlled system is improved.
Finally, the simulation results indeed indicate the effectiveness and feasibility of the design method of PID controller proposed by this thesis.
|
|---|
