Neural network optimisation using genetic algorithm: A hierarchical fuzzy method

In this paper, fusion of neural networks (NNs), genetic algorithms (GAs) and fuzzy logic (FL) is considered by taking account of the advantages of each. In this process neural networks are used for universal approximation, genetic algorithms are used for optimisation of the structure and weights of the neural network, and fuzzy logic is used to give directional and priority approach to genetic evolution. Hierarchical fuzzy approach can simultaneously provide a priority and direction of search to a chromosome so as to achieve an optimal or near optimal set of solutions. The proposed approach dynamically adopts the chromosome and maintains uniformity and diversity in the population to simultaneously provide local and global search. Modelling of flexible manipulator is used to demonstrate the performance of the proposed approach. Copyright (C) 2000 IFAC.