Research On Risk Control Of Insurance Enterprises Based On Neural Network

China’s insurance industry has experienced rapid development for nearly 40 years and has become an emerging market with the most development potential in the world economy.However,with the rapid development of the insurance industry,the problems and risks of the insurance industry have become increasingly prominent.The solvency adequacy ratio is the most important core indicator of the insurance industry.With the establishment of a risk-oriented solvency system(hereinafter referred to as the "second generation of solvency"),insurance companies are required to establish effective restraint mechanisms to improve insurance risks and market Risk,credit risk,liquidity risk,operational risk,strategic risk,reputation risk,these seven categories of risks are highly valued.In order for insurance companies to achieve business goals and be willing to take certain risks,they also need to have an effective restraint mechanism.The risk-oriented solvency supervision system formulated by the China Banking and Insurance Regulatory Commission covers seven categories of risks in a scientific and reasonable manner.According to the new regulations of the China Insurance Regulatory Commission,insurance companies should adopt a combination of quantitative and qualitative methods according to their own business development direction and actual risk status,formulate their risk appetite,and realize the risk of risk appetite to risk tolerance and risk tolerance to risk limit.Transmission mechanism,effective early warning of seven types of risks.In the context of the insurance-industry’s promotion of a risk-oriented solvency system,this article uses the neural network machine learning theoretical knowledge from the perspective of the second generation of compensation supervision to establish a three-level early warning system to achieve dynamic early warning.This article states the risk preference framework,risk preference theory,neural network theory,and uses neural network to realize the decomposition and early warning of risk preference.This article covers a wide range of risk indicators,and for the first time the neural network is applied to the insurance industry’s second-generation phase-two project.The neural network algorithm is a qualitative breakthrough compared with the traditional multiple regression method used in the insurance industry for risk appetite conduction.It simulates the human brain for effective risk warning.This paper clarifies the input layer,hidden layer,and output layer of the neural network,clarifies the setting of network learning parameters,and selects the data of insurance companies operating in China as the training set,from capital,value,profit and growth,liquidity,etc.In terms of selecting risk indicators,using neural network algorithms for training,designing a dynamic early warning system,giving out the methods and implementation steps of risk preference transmission and risk monitoring,it can provide a reference for the insurance industry.