Design of Gene-Based Fuzzy PID Controller

• Main Authors: Shao-Chuan Lo, 羅紹洤
• Other Authors: Lin Hong
• Format: Others
• Language: zh-TW
• Online Access: http://ndltd.ncl.edu.tw/handle/81462659631930147383

碩士 === 國立高雄應用科技大學 === 電機工程系博碩士班 === 96 === Conventional PID controllers generally do not work well for nonlinear and, higher order or time delayed linear systems, particularly for those uncertain systems which have no precise mathematical models. An alternate to get a better performance is to incorporate human intelligence into control algorithm by developing fuzzy logic controller. Nowadays, PID controllers incorporate fuzzy logic usually take a hierarchical structure. Because this can reduce the number of rule base. General fuzzy PID controller needs 3 variables. If there are 7 rules for each variable, then there will be 343 rules. That will make the system too complex. Therefore, most practical case use the hierarchical fuzzy PI+D or PD+I controller instead. On the other hand, the scaling factors of conventional fuzzy PID controllers must be adjusted according to personal experiences or using trial and error method. It does not have any certain rule to follow up. Novel fuzzy PID controllers such as in the cases of Escamilla-Ambrosio and Asim-Ali-Khan, though they use new method to get parameters, they need professional knowledge background and working experience. In this paper, we propose the fuzzy PID controller with the scaling factors modified by genetic algorithm (GA) method, which will reduce the complexity of the design. The parameters of the aforementioned controllers are determined by means of genetic algorithms. These methods have several advantages: (1) Systematic search technique can save much time and effort than that of conventional trial-and-error design method. (2) This technique does not need extra professional knowledge or mathematical analysis of system dynamics. (3) The system performance specifications can be achieved by selecting a proper fitness function.