Application of improved BP neural network model in bank financial accounting

•Combining factor analysis with BP neural network, the model is optimized by using L2 norm, and the learning rate of model training samples is adjusted to dynamic change to form an improved BP neural network model.•The improved BP neural network model is applied to the financial accounting field of commercial banks to evaluate the accounting performance.•Combining with commercial banks, a special type of enterprise, a targeted financial indicator system is designed and applied to the model. The research studies the enterprise group facing commercial banks, which are both market-oriented and socialized, and constructs an evaluation model based on the particularity of their daily financial accounting. This model can analyze the financial accounting performance in combination with the financial characteristics of commercial banks. The research will propose a BPNN model with the introduction of correlation factor analysis, select reasonable indicators according to the principle of indicator design, introduce factor analysis into the BPNN model, and optimize the model using the L2 norm. Finally, adjust the learning rate of the model training samples to dynamically change according to the increase in the number of iterations. The mean square error of the model in the training processes decreases and becomes very stable in the convergence process. The predicted value curve of the model almost coincides with the expected value curve, and the total error sum of the predicted results of the model is 0.003796, with higher accuracy and good generalization ability. The accuracy of the four models was analyzed. The accuracy of the model proposed in the study reached 95.6%, and the accuracy of the other three models were 83.1%, 86.8% and 91.2%, respectively. The results show that the model has higher accuracy, reliability and effectiveness in the financial accounting of commercial banks, reduces the losses in the financial accounting of commercial banks, provides a technical basis of the monitoring for financial accounting of micro banks, and provides theoretical guidance and support for the application of BPNN model with the introduction to relevant factor analysis in practice.