Abstract: A fuzzy RBF neural network learning algorithm based on genetic algorithm is proposed.This algorithm adopts genetic algorithm to optimize the fuzzy RBF neural network’s structure parameters,adapts the optimization result as the initial value of the fuzzy RBF neural network’s structure parameters,and dynamicly adjusts the parameters using the gradient decent methods.Finally  the algorithm is applied to classify the data and function approximation.The simulation shows that the fuzzy RBF neural network has  higher precision and robustness.

Key words: genetic algorithm(GA), fuzzy RBF, neural network, function approximation