Research On Safety Risk Level Of Chemical Industry Park Based On BP Neural Network

Benefit from the good international, domestic development environment and policy support, the development of chemical industry is rapid, large and small industrial parks was build up all over the country. But in recent years, due to the lack of scientific co-ordination, planning, design, the problems in the industrial structure, the internal layout, environmental pollution and other aspects were gradually revealed in many chemical park, and the security situation is not optimistic. In terms of the entire Chemical Industry Park, it was an extremely complex system, and all the companies, people, the natural environment and public property were inevitably threatened by various risk factors in the park. If the risk factors were not effectively identified and controlled, it was difficult to maintain the park operating sustainably, and the security of the resources, environment and personal property are vulnerable. Therefore, the establishment of security risk assessment system in the chemical park was important to improve the management level of the park.In this study, first of all the concepts of risk management theory, risk analysis, and evaluation methods was studied, and the superiority of the BP neural network was introduced emphatically. Secondly, Combined with accident causation theories, four factor model, large number of documents and questionnaires, the security risk assessment system in the chemical park was established. And the AHP was used to sort the importance of each indicator for the distribution of security risk in Chemical Industry Park being showed visually. Thirdly, combined with questionnaires and optimized BP neural network algorithm, the BP neural network model of safety risk assessment was established. Finally, the BP neural network model proved to be feasible, and it could be used to evaluate the security risks in the Chemical Industry Park. In this paper, the conclusions were showed that the subjective factors of questionnaires are effectively avoided by the BP neural network model, and the calculation process was simplified effectively. This model was useful for the security assessment of the Chemical Industry Park.