Abstract:
Type-2 fuzzy logic systems are proposed in the literature as an alternative to type-1 fuzzy logic systems because of their abilities to more e ectively model rule uncertainties. This thesis extends the idea of using the sliding mode control theory in the training of type-1 fuzzy neural networks to type-2 fuzzy neural networks. In the approach, instead of trying to minimize an error function, the weights of the network are tuned by the proposed algorithm in a way that the error is enforced to satisfy a stable equation. The parameter update rules are derived, and the convergence of the weights is proved by Lyapunov stability method. The performance of the proposed learning algorithms is tested for both type-1 and type-2 fuzzy neural networks on a real-time laboratory servo system. Simulation and experimental results indicate that the proposed type-2 fuzzy neural network with the proposed learning algorithm is more robust to uncertainties and computationally e ective than its type-1 fuzzy counterpart.