Bank loans risk evaluation based on combination of self-adaptive RBF neural network and expert system (ID: 5-030)

In this paper, excellent evaluation index system is chosen, equal design method is adopted to design sample, and through multistage Gyre relation-fuzzy comprehensive performance evaluation, many samples are selected to do self-training, and bank loans risk evaluation system based on combination of self-adaptation RBF neural network and expert system is worked out. Experiment shows, every single index data which is produced by expert system equals to the evaluation of specially invited industry experts. At the same time, combining with these index data, RBF neural network can bring its self-adaptation ability and non-linearity approaching ability into play to evaluate loans risk. The result is ideal, and can greatly conquer the subjectivity and uncertainty of evaluation.