| This dissertation presents a novel approach to combining wavelet networks and multi-layer feedforward network for fuzzy logic control systems. Most of the existing methods focus on implementing the Takagi-Sugano fuzzy reasoning model and have demonstrated its effectiveness. However, these methods fail to keep the knowledge structure, which is critical in interpreting the learning process and providing insights to the working mechanism of the underlying systems. It is our intention here to continue the previous research by the PARCS group in this area by utilizing individual subnets to implement decision-making process with the fuzzy logic control systems based on the Mamdani model. Center Average defuzzification has seen its implementation by a neural network so that a succinct network structure is obtained. More importantly, wavelet networks have been adopted to provide better locality capturing capability and therefore better performance in terms of learning speed and training time. Offline orthogonal least squares method is used for training the wavelet subnets and the overall systems is updated using the steepest descent algorithm. Simulation results have shown the efficacy of this new approach in applications including system modeling and time series prediction. |
