Research On Intelligent Early Warning Of Information Technology Enterprise’s Financing Mismatch Risk Based On Convolutional Neural Network

Looking at the development strategies of countries around the world,almost all innovative countries have made information technology a strategic development area.China’s information technology industry has developed rapidly during the 12 th and13th Five-Year Plans and has become one of the pillar industries.Modern technologies such as the Internet of Things,big data and cloud computing are all closely related to information technology.The information technology industry has become one of the most important components of the future economy.However,compared to the traditional manufacturing industry,the information technology industry requires more capital and core technology,more intangible assets and R&D investment,and needs to continuously carry out financing activities to inject new vitality into itself,and is more likely to face financing problems due to the uncertainty of future earnings.As a result,the risk of mismatch in financing has become an issue worthy of in-depth study by many experts and scholars.The outbreak of financing mismatch risk is also a long-term accumulation process.Establishing an effective intelligent early warning model for financing mismatch risk to capture the company’s financing status and predict future financing mismatch risk will help enterprises to perceive financing mismatch risk early and carry out targeted risk prevention and control,so as to nip the financing mismatch risk in the bud and reduce unnecessary losses.This thesis takes information technology enterprises as an example,analyses their financing mismatch characteristics and current situation from the micro level,explores the financing mismatch risk of information technology enterprises from the dual perspective of financing maturity mismatch and financing structure mismatch,and conducts intelligent early warning on financing mismatch risk through risk intelligent identification,factor intelligent analysis,construction of intelligent models and empirical analysis.(1)Designing an early warning indicator system for financing mismatch risk of information technology enterprises.The risk factors of financing mismatch and financing structure mismatch of information technology enterprises were constructed through web crawler technology,text analysis method and root analysis method,and the preliminary early warning indicators were designed based on these,and then the early warning indicators of financing mismatch risk were selected through the steps of "indicator pre-processing→normality test→difference test" to establish The system of early warning indicators for financing mismatch risk was established.(2)Constructing an intelligent early warning model for the financing mismatch risk of information technology enterprises.Firstly,the intelligent early warning model of financing mismatch risk of information technology enterprises is constructed based on convolutional neural network,and then the model is optimised through parameter training to finally determine the early warning model used to achieve prediction,so as to carry out dynamic early warning of financing mismatch risk of information technology enterprises.(3)Designing an early warning positioning matrix for financing mismatch risk of information technology enterprises.The variance coefficient method is used to synthesize the individual indicators of financing mismatch risk warning into the financing maturity mismatch risk warning index and financing structure mismatch risk warning index,and determine the risk threshold,and then construct the financing mismatch risk positioning matrix to position the financing mismatch risk warning degree of information technology enterprises.(4)Empirical analysis.This thesis takes the ST-companies and their paired enterprises among information technology enterprises from 2001 to 2020 as the research sample,and conducts empirical research in three stages: training stage,detection stage and prediction stage.Firstly,the sample of the three stages is measured by an early warning model,then an early warning index is constructed to locate the financing mismatch risk,then convolutional neural network is used to forecast the financing mismatch risk of information technology enterprises in the next three years,and finally a financing mismatch risk control path is designed in order to achieve effective prevention of financing mismatch risk.The innovations of this thesis are:(1)Intelligent identification of financing mismatch risk elements.This thesis obtains literature related to financing mismatch risk of information technology enterprises through web crawler technology,then analyses and identifies the obtained literature through text analysis methods,and finally identifies and analyses the elements of financing maturity mismatch and financing structure mismatch risk based on rooting analysis methods.(2)Convolutional neural network method was introduced in the construction of an intelligent early warning model for financing mismatch risk of information technology enterprises.This thesis mainly uses convolutional neural network method to find the mapping relationship between financing mismatch risk alert degree and risk characteristics,and determines the intelligent early warning model of information technology enterprise financing mismatch risk based on convolutional neural network through continuous training and tuning.