Financial Management Risk Prediction Algorithm of New Energy Enterprises Based on Improved Neural Network

In order to make new energy enterprises more flexible and stable in development, people must strengthen the control of financial management risks (FMRs for short here) and establish a more strict and standardized financial system. At present, most of the research is based on statistical models, and neural network (NN) has gradually become the research focus of FMRs. However, this method has both advantages and disadvantages and needs continuous improvement. Because there are certain limitations in using traditional methods in risk, this paper attempts to use Back Propagation (BP) NN to predict the risk of new energy enterprises, in order to provide a better FMRs assessment plan for managers and investors. In order to improve the accuracy of prediction, this paper used Particle Swarm Optimization (PSO) to optimize the BP NN. The experimental results showed that the prediction accuracy of BP and PSO-BP was 68.3% and 84.5% respectively under 300 training samples. Under 300 test samples, the prediction accuracy of BP and PSO-BP was 58.3% and 89.5% respectively. It can be found that the prediction accuracy of PSO-BP was higher than that of BP NN whether in training samples or test samples.