Evolving fuzzy controllers through evolutionary programming

Fuzzy logic controllers have been proven to be an effective means of solving real world control issues. One of the difficulties in the construction of fuzzy controllers is the design of the rule base under which they operate. This paper investigates the application of evolutionary programming as an iterative learning process for the fuzzy rule base. This approach is applied to the problem of an elevator control system. The system is optimized for efficiency and smoothness by encouraging higher velocities with minimal changes in acceleration, and by discouraging violations of the design parameters for the system. The performance of the evolved system compares favorably to that of fuzzy controllers designed using traditional methods.