Structural simplification of a fuzzy-neural network model

The proposed Fuzzy Neural Network(FNN) model uses feedforward neural network to perform fuzzy systems with the product-sum fuzzy inference mechanism. So the number of fuzzy rules directly determines the complexity and efficiency of the FNN model. Based on the Neural Network Self-configuring Learning(NNSCL) algorithm, the modified NNSCL algorithm is obtained by solving a system of linear equations with the Conjugate Gradient Preconditioned Normal Equation (CGPCNE) algorithm the unknows of which are the adjusting factors of the remaining weights. The modified NNSCL algorithm is applied in the rule-reasoning layer to simplify the structure of the FNN model, aiming at minimizing its complexity and preserving a good level of accuracy without retraining after pruning. The simulation results demonstrate the effectiveness and the feasibility of the algorithm.