A new condition assessment Model and optimization by using hybrid GA

This paper proposes a new condition assessment model for generators. The model is based on radial basis function information neural network and fuzzy inference. The architecture and some training algorithm of the neural network are described. The input membership functions and fuzzy logical rules are selected in the light of the expert experience and knowledge, such as fault-operating years relation "bathtub curve". The membership functions are designed and optimized. The 12 parameters of the membership functions are optimized using hybrid genetic algorithm. The hybrid approach combines a genetic algorithm with a feature of simulated annealing. The features of the hybrid GA in this paper are discussed. The availability of the approach is examined by simulated test example.