A memetic GSA with niching selection for training fuzzy wavelet neural network

This paper proposes an effective memetic Gravitational Search Algorithm (GSA) that utilizes Solis and Wets’ (SW) algorithm as local search. GSA has good exploration ability and SW helps to improve the exploitation ability of the memetic algorithm. Furthermore, a selection strategy is proposed to select suitable individuals for local refinement that is based on subtractive clustering. Proposed memetic algorithm is employed for tuning fuzzy wavelet neural network parameters. We evaluate the performance of the proposed memetic algorithm on two system identification problems. Computational results confirm the performance of the proposed memetic algorithm and selection strategy on training fuzzy wavelet neural network parameters.

کلیدواژه ها:

Fuzzy wavelet neural network ، Memetic algorithm ، GSA ، Subtractive clustering