Research And Application Of Intelligent Perception And Identification Method Of Coal Mine Gas Risk Based On Big Data Technology

Under the background of Carbon Neutralization,coal would still serve as an important part of domestic energy consumption,promoting national economic development and energy security.With the popularization of big data,artificial intelligence and other technologies in coal mines,it provides technical and data support for the perception and judgment of coal mine safety production risks and the transformation of the Internet + coal mine supervision mode.Based on massive gas perception data,this dissertation uses big data and artificial intelligence technology to build a massive data file storage system,and proposes a method for gas over-limit abnormal data classification and coal and gas outburst risk identification methods to realize coal mine gas risk classification and judgment regulatory monitoring provides support for decision-making.A new file storage system for massive gas monitoring data was established.Using Kafka,Flink and HDFS big data component technologies,a KFH file storage system for massive gas data across the country is built to quickly write small files for massive gas monitoring.It has been verified by the on-site environment and supports 3.5 billion daily and 40,000 per second perception data write throughput.Based on the file structure and measuring point characteristics of gas monitoring data,a high-efficiency compression algorithm based on second-order difference was developed by using the merging optimization method of massive small files.The field environment verification showed that the overall file compression rate reached 12.64 times,effectively reducing the time stamp and numerical value.Data storage space overhead.A gas anomaly data classification method based on multi-trend feature Shapelet analysis of key turning points was proposed.Using the multi-trend feature shapelet discovery method of key turning points,based on the information gain,the shapelets with the most representative features are selected to form a classifier.Through the weighted voting mechanism of Bayesian probability,the gas over-limit alarm time series can be quickly classified,and the shape features of five types of alarms such as sensor calibration,sensor displacement,sensor wrapping,sensor failure and coal and gas outburst can be quickly and accurately identified.A risk identification method of coal and gas outburst based on FAHP was constructed.Taking the perception data of gas emission as the characterization factor of coal and gas outburst,using statistics and deep learning methods,establishes moving average,deviation rate,dispersion rate,volatility,root mean square error(RMSE),mean absolute percentage error(MAPE)as the main body of coal and gas outburst risk identification index system.A gas risk identification method based on FAHP is constructed to realize intelligent identification and early warning of coal and gas outburst risks.A coal mine gas risk assessment model based on the PSO-AHP coupled cloud model was established.Using the knowledge graph theory and method,a knowledge graph of coal mine gas risk is established.A comprehensive assessment system of coal mine gas risk is proposed,which takes gas over-limit alarm and coal and gas outburst identification results as key indicators,and takes personnel quality factors,facilities and equipment factors,production environment factors and safety management factors as evaluation indicators.A method for determining the weight of indicators based on PSO-AHP is proposed,and the numerical distribution of evaluation indicators is analyzed with cloud model theory,and a coal mine gas risk assessment model based on the PSOAHP coupled cloud model is constructed to achieve accurate identification of coal mine gas risks.The engineering application of the gas risk assessment method based on big data and artificial intelligence was realized.Based on the national coal mine safety production risk monitoring and early warning system,the massive gas data storage technology,the gas anomaly perception data classification method,the coal and gas outburst risk identification method,and the coal mine gas risk comprehensive assessment method are applied to the national coal mine gas risk assessment and early warning.Technology and data support is provided for remote monitoring and precise supervision of coal mines.The dissertation has 68 figures,25 tables,and 172 references.