Abstract: The diagnosis method based on CBR is focused in fault diagnosis recently. Radial basis function (RBF) is a kind of neuron functions based-on similarity. In allusion to the problem of similarity based case retrieval, the RBF is introduced into design of case retrieval and a RBF based model is proposed in this paper. By means of improving the model network, the retrieval model becomes more suitable for practical diagnosis. Simulation and practical application show that the RBF based retrieval is better and more effective than BP based one.

Key words: Fault diagnosis, Knowledge system, Neural network, Radial basis function, Retrieval