A Study for Learning Method of Modified PID Controller with On-line Hybrid Genetic Algorithm

摘要

The PID control system is one of the most famous controllers in the industrial products because it doesn't need theoretical knowledge to adjust unknown variable in the controller. In former study, a GA (genetic algorithm) was used to tune the parameters of PID. These studies attempt to find PID one and their results is pretty good. But these trials work on off line because it still needs too much calculating time. This study is concerned about learning method of on-line GA. At the first, the limitations of variables of PID are selected by relay feedback. At the second, the time range for object function also has limited by the time constant in the Ziegler-Nichols Method. At the final, parameters of PID are optimized by on-line GA that is hybrid GA based on micro GA.