With the continuous change of the global economy and the increasingly fierce competition in the market, the financial risks faced by enterprises are increasingly complex and diverse. This paper constructs a LSTM neural network financial risk early warning model for enterprises based on recursive neural network. 130 listed companies were selected as the research object, and 28 financial indicators and 2 non-financial indicators were obtained as the research samples. Then the 30 financial early warning indicators were downscaled using factor analysis to extract the principal component factors. The principal component factor scores are input to the LSTM network, the relevant parameters of the network are set, and the network is trained to complete the construction of the enterprise financial risk early warning model. The training results of this paper’s model show that the model tends to be balanced after three hundred iterations, and the fit of the model is better, and the loss value is only 0.12. The empirical results of the model’s financial risk prediction show that this paper’s model has a better performance than the traditional prediction models, such as Random Forest, in terms of the prediction of corporate financial risk. The application of LSTM neural network to the financial warning of enterprises has obvious advantages.
