An evolutionary tuning technique for type-2 fuzzy logic controller in a non-linear system under uncertainty

Uncertainty is an inherent part in control systems used in real world applications. Various instruments used in such systems produce uncertainty in their measurements and thus influence the integrity of the data collection. Type-1 fuzzy sets used in conventional fuzzy systems cannot fully handle the uncertainties present but type-2 fuzzy sets that are used in type-2 fuzzy systems can handle such uncertainties in a better way because they provide more parameters and more design degrees of freedom. There are membership functions which can be parameterised by a few variables and when optimized, the membership optimization problem can be reduced to a parameter optimization problem. This paper deals with the parameter optimization of the type-2 fuzzy membership functions using a new proposed reinforcement learning algorithm in a nonlinear system. The results of the proposed method referred to as Extended Discrete Action Reinforcement Learning Automata algorithm are compared to the results obtained by the Discrete Action Reinforcement Learning Automata and Continuous Action Reinforcement Learning Automata algorithms. The Performance of the proposed method on initial error reduction and error convergence issues are also investigated by computer simulations.