Research On Intelligent Prediction Of Mine Gas Based On Big Data

Gas accident is one of the main disasters during coal mine production in China.Frequent occurrence of gas accident not only brings enormous economic loss for coal enterprise but also seriously threatens to the life safety of coal mine workers.The effective analysis and deep mining for the huge amounts of monitored gas information and seeking for the implicit regularity in the gas concentration can significantly predict gas accident.Realizing gas concentration intelligent prediction is an important issue for gas disaster prevention.The theory analysis,mathematical modeling,software development and engineering application were used in this thesis.The topology framework of gas concentration intelligent prediction system was proposed.The system overall frame with four layers structure,data acquisition,data storage,data analysis,data application was built up,and the detailed design for each layer was carried on,and then the system overall design was completed.On the basis of the data acquisition and data storage,the influencing factors of gas concentration were analyzed,and the characteristics of dynamic nonlinear and temporal correlation were revealed.After preprocessing of the monitored gas concentration data,the spatio-temporal correlation analysis model for gas concentration was established on the basis of the modified dynamic clustering method,and the multivariate phase space reconstruction model based on the temporal and spatial correlation method was also built.The parallelism of BA algorithm is studied based on Mapreduce procedure and the BA-ELM single-step prediction model for gas concentration was proposed according to this algorithm.The real-time error compensation model was established using the predicted results and the multi-step intelligent gas concentration prediction method was put forward.According to the big data processing platform of Hadoop and the WebGIS platform architecture,the intelligent prediction system for gas concentration was designed and developed on the basis of data analysis methods.This system realized some useful functions such as the automated acquisition and storage for monitored gas data,the intelligent analysis and forecasting,visualization,sharing and publishing.The coal mine gas intelligent prediction system based on big data was applied to Chengzhuang coal mine in Jincheng coal group.Results verify good applicability and reliability of the models and system,which provide a new research approach for coal mine gas accident prevention and control.