Genetic algorithms applied to fuzzy sliding mode controller design

The drawbacks of sliding mode control in terms of high control gains and chattering are overcome by incorporating fuzzy control to the switching logic. This hybrid system increases the complexity in design and, at present, there exists no effective design tools due to the lack of analytical and numerical approaches. This paper develops an automated design approach to this design problem, using tournament and rank-based genetic algorithms to replace the trial-and-error designs. The method is illustrated through the design of a near-optimal fuzzy-sliding mode controller for a nonlinear liquid-level control system. The control strategy gives a relatively low overshoots with smooth control action and retains robustness of the sliding mode approach.