Study On A Credit Evaluation Model And Algorithm Base On Neural Network

Comprehensive evaluation of complex object has been studied and concerned widely. The evaluation of university student credit is typical comprehensive evaluation of complex object. It relates to a lot of complicated influence factors which are described in different way and where there are many fuzzy phenomena. Therefore, it is important to consider the influence of fuzzy factors when making comprehensive evaluation.Based on university student credit, this paper introduced interrelated theories and methods of comprehensive evaluation of complex object and analyzed status quo of comprehensive evaluation. It analyzed each characteristic of neural network and fuzzy comprehensive evaluation and compared each other. According to advantages and disadvantages of neural network and fuzzy system, it studied on the combination of them and present fuzzy evaluating neural network model to solve the fuzzy problem in the university student credit evaluation.Fuzzy evaluating neural network model is based on neural network and fuzzy comprehensive evaluation. The model presents the structure and arithmetic. It studies on membership function, weights and fuzzy synthesis operator of model. The mode integrates fuzzy theory into neural network. So it makes use of the fuzzy theory to deal with the fuzzy factors and self-learning of neural network to adjust the membership function and the weights which fuzzy evaluation handle fuzzy information with. The arithmetic is better than traditional arithmetic at convergence speed because the initialization of parameter is confirmed by fuzzy comprehensive evaluation.The paper made use of fuzzy evaluating neural network to evaluate the university student credit and researched the evaluating index system of university student credit. It confirmed relevant network structure for university student credit and utilized improved AHP to get weights. The membership function of each index and fuzzy synthesis operator were defined. It made use of MATLAB7.1 to build fuzzy evaluating neural network and do simulation. The simulation proved that the mode was reasonable and feasible. Finally the software of university student credit was worked out based on MATLAB to apply fuzzy evaluating neural network. The model provides useful reference to the right decision and a good way to use comprehensive evaluation of complex object in other fields.