The Reas Arch On Application Of Time Series Mining Algorithm In Production Safety Accident

Annual economic losses caused by security problems account for about 6% of the total GDP, to the country and the people has brought great losses, so the prediction of production safety accidents is very important.The traditional analysis of production safety accidents include statistical analysis, regression model, gray model, etc., is not conducive to take measures to prevent accidents. In the research of production safety accident data, this paper combines the univariate time series theory and the theory of multiple time series, and has some innovations in the theory application.The time series forecasting model is applied to the prediction of production safety, especially the multiple time series The method of vector moving average autoregression for binary time series is applied to the production safety accidents. Based on the data of nearly 10 years, the trend and influencing factors of accidents are analyzed from various aspects. The suggestions and suggestions are given. Production safety accidents. In this paper, several aspects of the work started:1. Obtain the data from the official website of the State Administration of Work Safety, and obtain the production safety accident and its brief information through the regular expression. Through data pretreatment of the accident, through the visual display of its data characteristics.2. The time series ARIMA model and the cubic exponential smoothing model are compared and compared with the results of the 15-year production safety accident sequence data. The residuals, the number of accidents and the trend of development are compared, and the residuals of the univariate time series prediction models are obtained. -21.92, the relative error of 0.266 than the three exponential smoothing have higher accuracy and reliability.3. The theory of multiple time series is studied, and the multiple time series is modeled and applied to the prediction of production safety accidents. Compared with the previous models, more robust prediction results and accuracy are obtained. The correlation between the number of deaths and the number of accidents was analyzed, and the growth rate was modeled by time series analysis, and the values were not predicted for 12 months.4. This paper combines qualitative analysis with quantitative forecasting to predict production safety accidents. Qualitative analysis: the accidents are divided into different types of accidents, including traffic accident type accounted for the most accidents; for the accident analysis of the year, which in 2005 as a turning point, before 2005, the accident was on the rise after 2005 was year after year The trend of the accident occurred in the region mainly to the accident which is more in the southwest mountainous areas, geological instability area. Quantitative analysis: The accident sequence is forecasted by establishing the exponential smoothing,one-dimensional time series, the multiple time series model, adjusting the model parameters and selecting the best model for the production safety accident sequence. Get the next year the trend of production safety incidents. Through the trend in a timely manner to prevent and reduce the occurrence of countermeasures measures for the country to provide support for macro-decision-making.