A novel evolutionary algorithm (EA) approach is developed for theimplementation of rule based controllers, which increases the diversityof individuals to reduce the opportunities of falling into local optima.The rule based controller is simply composed of a look up table whosedimensions are determined by the numbers of the subranges of errorvariables required for operation. Many experiments are successfully madeon the evolution of the rule based controllers by the novel EA for threeillustrative examples: DC servomotor control, ball and beam control, andcart pendulum control. The computer simulations show that the proposedEA is an effective approach to generating control rules which manipulatenonlinearly dynamical systems exceedingly well. It reveals that the EA'slearning behavior is very like that of a human expert
展开▼