Research On PLS-based BP Neural Network Forewarning Model

With the rapid development of China's capital market,the securities market is becoming increasingly active,and the number of listed companies is increasing.At the same time,manufacturing companies also occupy a very important proportion in listed companies.It can be said that the healthy development of listed companies in manufacturing is directly related to the healthy and sustainable development of China's national economy.However,in recent years,in response to China's macroeconomic environment,since 2012,China's GDP growth rate has begun to decline,China's macro economy continues to slow down,economic downward pressure continues to increase,and the stage of economic growth has undergone fundamental changes.Under the severe macroeconomic situation,the risk of financial crisis will increase greatly,and the consequences of financial crisis will be more serious.Therefore,the prediction of the financial crisis of listed companies in the manufacturing industry for the development of the national economy,company managers,investors And government regulators are of great significance.At present,there are various research methods and models for financial crisis early warning.The methods mainly include multiple linear regression,principal component analysis methods,etc.The model also gradually transitions from the traditional measurement model to the artificial intelligence model stage.There are a lot of research results,but there are also some areas that need improvement: On the one hand,it is necessary to establish early warning models for enterprises in different industries,instead of targeting all enterprises in a big way;on the other hand,it needs more predictability.The emergence of a high financial crisis warning model.Under this circumstance,this paper proposes innovative research ideas for the aspects that can be improved.For the financial crisis early warning model of China's listed manufacturing enterprises,the variable screening method adopts multiple regression,principal component analysis and canonical correlation analysis.The partial least squares method is based on artificial intelligence algorithm BP neural network to establish an early warning model.Through the combination of the two methods,it is possible to establish a high-accuracy financial crisis early warning model.This paper first selects 56 samples from the three-year interval of Shanghai and Shenzhen in 2014-2016.Each sample selects 22 economic indicators for research,and divides the research object into training samples and test samples,and then uses partial least squares method.Extracting the main content that can contain 22 economic indicators,and then taking the extracted components as the input of the BP neural network model,whether it is the output of the financial crisis company as a model,through the training of training samples,initially established for A model for early warning of the financial crisis of listed companies in the manufacturing industry,and then test the sample into the model to test the accuracy of the model.Through the above ideas and methods,this paper finally establishes a financial crisis early warning model suitable for manufacturing listed companies in different years.The accuracy of the model is above 80%.In order to make the model fall into the field,this paper applies the model to Tsingtao Brewery Co.,predicting whether it will have a financial crisis in 2018-2020,and finally concludes that the company will not have a financial crisis during this period,but the company The financial situation began to conclude with unstable factors..