In the traditional loan credit risk assessment of enterprise, enterprise credit and loan risk is conflicting. That is, when trying to change the needs of either party, the credit risk will be increased necessarily. Therefore, we proposed an enterprise credit risk assessment method based on the improved genetic algorithm to establish the performance architecture of the latitude-longitude grid performance architecture, the performance of the architecture latitude grid, improving the the generation of initial population, which improved the derivative rate of genetic algorithm and reduced the number of iterations and operations. The speed of operation was improved. For the enterprise credit credit, the fitness function was proposed, and the adaptability of the improved algorithm to the credit risk of the enterprise was satisfied. The experimental results show that the improved and optimized latitude-longitude grid task genetic algorithm improved the enterprise credit risk assessment method, which is better than the traditional strategy, and can be widely used.