Small, medium and micro enterprises started late in China, with poor transparency of financial information and relatively weak stability of profitability and asset strength, which makes commercial banks need to bear more risks when providing loans to small, medium and micro enterprises than large enterprises. When there is no credit record of some small and medium-sized micro enterprises in commercial banks, it will increase the risk that banks need to bear when they lend to these small and medium-sized micro enterprises without credit record, and it will also increase the difficulty of small and medium-sized micro enterprises’ credit loans in commercial banks. It is a big problem that how to accurately predict and evaluate the small and medium-sized micro enterprises with unknown reputation, and then reduce the risk pressure of banks to provide loan services to these small and medium-sized micro enterprises with unknown reputation. The purpose of this project is to predict and evaluate the credit risk of enterprises with unknown reputation, and then evaluate the comprehensive strength of small, medium and micro enterprises that need loans according to the TOPSIS comprehensive evaluation model, so as to calculate the interest rate of bank loans to each type of enterprises, and provide a credit strategy for commercial banks when the total amount of bank credit is 100 million yuan. We use DNN intensive neural network algorithm to list the important parameters of enterprise capability, train a large number of raw data covered by these parameters, so as to evaluate the credit rating of enterprises, and then use TOPSIS comprehensive evaluation model to evaluate the strength of enterprises. Finally, we use k-means clustering algorithm to classify these 203 enterprises and make the best credit decision.