Fuzzy-logical Control Models of Nonlinear Dynamic Objects

The article considers the task of developing a fuzzy-logical PID-type controller for a nonlinear dynamic system. A feature of the structure is presented, which consists in simplifying its controller by decomposition. In the simplest version, three fuzzy controllers are used with one input and one output and separate rule bases. Parameters of fuzzy controllers are optimized using a genetic algorithm. A two-step controller tuning scheme for a nonlinear dynamic object is proposed. At the first step, the genetic algorithm is used to tune the linear PID controller; it is shown that the obtained coefficients are used at the output of each channel of the fuzzy PID controller. At the second step, using a genetic algorithm, a nonlinear transforming function is formed for each channel, implemented on the basis of an artificial neural network. The control algorithm is debugged and tested using the MatLab system. The results show a significant improvement in the characteristics of the transient process compared to traditional controllers.

Перевод пока недоступен