An Empirical Research on Financial Crisis Forecasting Based on BP Neural Network

The research on financial distress prediction is significant to the economy. Tak ing into full account the status quo of the accounting information supply in Chi na, this paper constructs a six-category warning index system in cluding solvency, assets and liabilities management, profitability, growth, cash flow and the disclosure quality of accounting information. It compares the effe cts of early warning using BP Neural Network before and after the rank sum test, sampling the listed companies in Shenzhen and Shanghai Security Markets. The re sults show that BP neural network model has good application value to the warnin g　for　financial　distress of China's listed companies. Especially, the accurac y of obtained classifying model is improved obviously after the rank sum test.