engleski

Evolutionary Algorithm Aided Design of Multi-switch Controller

In this paper is investigated the possibility to adjust parameters of multi-switch controller of complex process employing evolution strategies (ES). Investigations performed and results obtained indicate that ES can be efficiently and practically used as methods for adjusting controller parameters. It should be noted that the implementation of ES in adjusting controller parameters does not require familiarity with special methods used for their adjustment. A computer simulation is made using a mathematical model of the process. Software tools MATLAB and Simulink are used in simulating the process. ES is the method used to adjust parameters of controller. Experiments were performed using ES with different parameters (number of parents, offspring, step size, generations) to find optimal ES parameters. The controller parameters thus adjusted are compared with empirically adjusted ones and obtained results achieve noticeable improvement in process behaviour.

controller; evolution strategy; MATLAB; optimization; process; Simulink

nije evidentirano