Study On Risk Situation Forecast And Early Warning In Chemical Industry Park Based On Systematic Risk Entropy

The expansion of chemical industry park is often accompanied by increase of hidden danger, also the complexity and uncertainty of the risk factors will increase. There are many deficiencies in traditional risk regulation means such as narrow coverage, information lag, subjective factors, and so on. The construction of integrated emergency management platform system is an important tool to strengthen daily supervision and emergency disposal of chemical industry park, implement monitoring coverage of environment comprehensive, and the massive real-time monitoring data from the field provides the underlying support for digging deeper risk situation information at the same time.The risk supervision trend of Chemical industry park is to turn passive disposition to active prevention in advance, so as to improve the dynamic risk control level. Therefore, how to use on-site real-time monitoring information efficiently to fully sensing risk situation is particularly critical. In this paper, from the perspective of overall situation, entropy theory representing the degree of system disorder was introduced, aiming at achieving the regional risk situation forecast warning, and research work was carried out as the following four aspects:(1) Evolution mechanism analysis of risk entropy. A solid risk hierarchical network model of chemical industry park was built on the basis of the accident hierarchical causing theory, and elaborated the risk characteristics of the system state transition of chemical industry park. The entropy theory was introduced to analyze risk situation from order to disorder, discussed accident internal entropy change power, and risk entropy cusp catastrophic model was built to analyze the internal mechanism and conditions of system state collapse and transition.(2) Establishment of risk entropy representation model. Risk situation awareness hierarchical structure model of chemical industry park was put forward, on the basis of monitoring data, parametric risk entropy model was proposed according to the fluctuate characteristics of monitoring data, so as to characterize integral fluctuation situation in regional scale; Projection pursuit function was proposed to solve the best projected vector, and systematic risk entropy model that objectively reflecting the multi-dimensional entropy feature was established to sense integral situation changes.(3) Study of risk situation forecast &early-warning method. Monitor abnormal time series of parametric risk entropy, and forecast time series of systematic risk entropy by support vector regression algorithm. Proposed “Support vector regression—Least squares fitting” to construct systematic risk entropy function of chemical industry park, a set of systematic risk state mutation analysis algorithms has been researched that sensing the collapse and transition process of system situation, and summarized two kinds of risk entropy early-warning mechanism——the parametric-level warning and systematic-level warning. Small testing platform for monitoring risk parameters was built.(4) Development of emergency information monitoring platform. Adopt the J2 EE application technology based on MVC architecture system, built a kind of mainstream SSH2 framework and developed B/S architecture platform by applying Java Web technology. The platform mainly includes two subsystem, emergency security and monitoring &early warning, and achieved connection of data from monitoring software of C/S architecture to the main platform through SQL Server database.