The Research Of Coal Mine Production Safety Time Series Data Prediction Methods

China is the biggest country to produce coal.The consumption of coal in China takes 69 percent in disposable source.But people died from coal in China are three times more than other countries.The problem,security of manufacturing the coal,is to be resolved.In the process of manufacturing the coal,gas accident is the most threatening factor.Because the environment of mine is complicated.The consistence of gas is affected by temperature,wind speed,carbon monoxide and so on.A real-time monitoring for the wine is one of the most urgent matter in the coal’s manufacture.In the paper,there are some forecast methods both at home and abroad and some disadvantages of real-time monitoring.By the data fusion technology,single forecasting model and combined forecasting model research,in view of the single sensor prediction one sidedness in the presence of single data source and single data model for forecasting accuracy is low and not while taking into account the data in the linear and the nonlinear case is put forward based in data fusion technology and the combination forecast model of prediction model,based on data level fusion and feature level fusion and prediction based on RBF neural network combination forecasting model of multi-sensor mine time-series data,a variety of sensors on the underground,including the gas concentration,wind speed,temperature sensor fusion forecast is built.1、Taking the gas concentration,temperature and wind speed sensor time series data as the research object,the data fusion and feature level fusion of sequential objects are firstly studied by using data fusion technology;2、According to the linear time series data after data fusion,using single exponential smoothing prediction model,obtained the experimental results and error analysis.3、According to the existing nonlinear time series data after data fusion,a single prediction model using RBF neural network,the results of error analysis;4、The linear and nonlinear time series data are taken into account,using the combined forecasting model based on RBF neural network to predict the timing of data after data fusion;5、Research on prediction methods of three coal mine time-series data,simulation experiment,the three kinds of prediction results were compared with the,draw more applicable to the prediction methods of multi sensor data of coal mine underground,so it can provide a basis for the process of coal mine production safety further decision-making.