System Forecast Technology Of Safety Incidents On Mining And Its Application

Mining industry is one of the industries of the production industrial which has been most serious, and security issue is restricting the development of the social economic. Then predicting the development tendency of the accidents scientifically and accurately is significant to the security management of mining system.In this paper, the common forecasting methods of mining is summarized systematically, and the related factors and system characteristics of mining accidents are analyzed comprehensively by the statistics of the historical data of security accidents in mining. On this basis, system forecasting technology for safety accidents is applied to the mining industry in view of the major application fields. The main studies include there aspects: the absolute index—the death toll, the relative index—the wounded rate per thousand, and the nonlinear systems—the accidents by roof falling and rib spalling in metal mining. After analysing those system characteristics, the suitable models of system forecast technology of safety incidents is built on mining. The sequence of absolute index of accidents—death toll in national mining accidents is fluctuant and periodic from 2005 to 2007. Then a model of system safety forecating is established by using the theory of seasonal periodicity forecasting. We can analyze the degree of fitting and the trend of changes of the forecasting results through the model. The results show that the seasonal periodicity forecasting model which not only forecasts the overall trend of the sequence of the death toll well, but also reflects the volatility and periodicity. The sequence of relative index—wounded rate per thousand in a mining of Huangshi from 1998 to 2007 declines and it is fluctuant. Because gray forecasting method is insensitive to volatile sequence, we can correct residual error of grey forecasting to compensate for it by exponential smoothing method . The results show that the gray-exponentially smoothing model is proved to have a higher accuracy rate. Cope with the nonlinear characteristic of the roof falling and rib spalling, a forecasting and evaluation model is used to be built based on the theory of Bayesian Network in this paper. The results show that the seasonal index prediction model which not only resolve the problem about the irregular length of the period in the series of incident indicator, but also forecast the trend well. It is of great significance to make a decision and prevent safty accidents in the management of mining.