Big Data Prediction Based On Methane Concentration Time Series

In this paper,under the support of the massive monitoring data produced by the coal mine safety monitoring system,the big data technology is applied to the coal mine methane concentration prediction with the methane concentration prediction model of the predecessors.The paper mainly includes the following research results:A big data storage system for methane concentration time series is built.By analyzing the web page structure of the coal mine safety monitoring system and the structure of methane concentration data,a web crawler used to collect massive time series data of methane concentration and the HBase table structure for storing massive time series data of methane concentration are designed.After that,the methane concentration big data are collected and stored.After analyzing the feature of the big data of methane concentration time series,a scheme for the short term prediction of methane concentration using the big data calculation framework MapReduce is designed.MapReduce first collate the methane concentration time series at each monitoring point,then output to the reduce function,and use the time series prediction model ARIMA to predict the methane concentration.By this way,the maximum and the trend of methane concentration of each monitoring point in the next day can be obtained.Based on Markov theory,a scheme for real-time prediction of methane concentration using MapReduce is designed.In the past research on methane concentration prediction,Markov theory only existed in the role of auxiliary optimization.In this paper,Markov theory is used as the main prediction model for real-time prediction.The Markov one step state transfer probability matrix of the limited methane concentration state value space is calculated by MapReduce.Then,the real-time prediction of methane concentration is realized by the way of probability comparison.Finally,based on the one-step probability matrix,the k-step over-limit probability corresponding to each concentration value is calculated,which is used as the evaluation standard for the methane concentration over-limit risk of the K-Means cluster analysis.Through the cluster analysis,the methane concentration values were divided into three levels: “very dangerous”,“dangerous” and “safety”.Through the classification of dangerous levels,managers can formulate corresponding safety management measures for each level to ensure safe production.