Research On Credit Rating Of Small And Medium-sized Enterprises Based On Fuzzy Neural Network

The economic structure and industrial structure of China is in a stage of deep adjustment and optimization, and the endogenous growth of economy demands for more human capitals and the participation of technological innovation in social and economic division of labor, and ultimately changes into economic growth output. Small and medium-sized enterprises will play a more important role in the process of adjustment and upgrading of economic structure of China. However, currently in the period of constricting the supply of the credit capital, there is asymmetric information between bands and enterprises, which affects small and medium-sized enterprises to obtain loans. With the continuous refinement of loan business, various banks have launched their own review programs before the credit loan of small and medium-sized enterprises. Meanwhile, it also has put forward higher requirements with regard to the methods of credit rating as well as the effectiveness of the methods. The credit rating link happens in front of the selection process of borrowers, which is an important part of the review process before the loans, and it occupies an important position in the feed-forward control system of risk control of the loan business of banks.The following terms require special attention while conducting study on credit rating of small and medium-sized enterprises:first of all, understand the theory of credit rating of Small and medium-sized enterprises, and establish proper evaluation index system; secondly, pay attention to the following three points while selecting the right methods of credit rating:the first one is the combination of subjective and objective, emphasizing the long-term accumulated experiences of experts, while being able to fully reflect the data as well as the objectivity of its changes during the analysis process of rating; the second point is the combination of qualitative and quantitative, the rating methods not only can quantitatively process the bulk data, but also can adjust the qualitative classification of data during the processing process; the third point is the combination of process analysis and interpretation of results. The rating process should be able to propose reasonable analysis procedure and calculation process for the samples being analyzed, as well as can explain and illustrate the results of analysis, strengthening the cognition of operator on the analysis results. In this context, this paper takes the Fuzzy Neural Network (FNN) model as the research method of credit rating of small and medium-sized enterprises.Based on the description of the background, research significance and the trends of relevant studies, this paper introduces the credit rating theory of small and medium-sized enterprises and the fuzzy neural network theory, and then it explains the research indexes and sample data required by the modeling, as well as describing the establishment of modeling, simulation experiment and text. The main conclusions of the study are as follows:1. The overall accuracy of FNN model on small and medium-sized enterprises as well as the test sample of credit rating of small enterprises is separately 100.00% and 96.67%, and the error rate of type Ⅱ are 0.00%, FNN model is an effective means of credit rating.2. In comparison of models, the overall accuracy of BP model on the above two tests is separately 93.34% and 90.00%, and the error rate of type II is separately 50.00 % and100.00%, that is, it is more likely to prejudge bad loans as healthy loans in the future. FNN has better testing performance than the BP neural network. Generally speaking, the establishing of fuzzy neural network model in this paper has the suitability and robustness in the studies on credit rating of small and medium-sized enterprises.