Research On The Method Of Coal And Gas Outburst Situation Awareness

Coal and Gas outburst is one of the typical forms of gas dynamic disaster in coal mine.The occurrence of gas outburst accident will cause huge economic loss and adverse social impact to enterprises.In order to discover the risk of gas outburst as early as possible and take scientific measures in time,this paper uses the basic ideas of situation awareness for reference and uses technical theories such as security risk management,compressed sensing,pattern recognition,information fusion,and machine learning.The combined research methods of theoretical analysis,numerical simulation and field test carry out in-depth research on gas outburst situation awareness from several aspects such as gas outburst situation awareness,situation understanding and situation prediction.The research content and results lay the theoretical foundation for the construction of an intelligent perception system for gas outburst situation and provide auxiliary decision-making for scientific management of gas outstanding disasters.Based on the analysis of gas outburst process and influencing factors,through theoretical analysis,field data analysis and numerical simulation experiment,the gas emission law,coal and rock fracture acoustic emission signal evolution characteristics during outburst process were analyzed.The results show that before the gas outburst,the gas emission has a precursory regularity;before the main destruction of coal and rock masses,the acoustic emission has obvious precursory characteristics.The task of gas outburst situation awareness are clarified,and a gas outburst situation awareness model that combines local situation awareness and global situation awareness is constructed.It is put forward that the situation elements of gas outburst should meet the principles of scientificity,precursory,real-time,operability,comprehensiveness and sensitivity.Taking Zhaogezhuang mine as an example,the real-time monitoring information of gas emission and acoustic emission are selected as the main gas outburst situation elements,and the drilling cuttings quantity,drilling cuttings desorption index,gas pressure and gas content are taken as auxiliary situation elements.The feasibility of selecting outburst situation elements is analyzed and demonstrated.An effective information extraction method of gas outburst situation elements based on compressed sensing is proposed.Taking the incomplete gas emission time series as the research object,the compressed sensing is used to achieve high-precision repair of the gas concentration time series.In order to extract the acoustic emission signal of coal and rock under the noise background,the combination of compressed sensing and wavelet denoising achieves the separation of strong noise signals and effective acoustic emission signals of coal and rock.Research on identification methods of gas prominent precursors.A trend feature extraction method for gas concentration time series based on the combination of five-point cubic smoothing and nonlinear segmentation is proposed.The mean value,trend slope and fluctuation rate of gas emission time series are taken as identification indexes of abnormal gas emission time series and the dynamic pattern matching distance is combined with hierarchical clustering to realize the identification of gas concentration anomaly time series including prominent catastrophes.The time domain,frequency domain and time-frequency domain characteristics of acoustic emission signal in coal and gas outburst process are studied.The energy characteristics of acoustic emission signal are extracted by wavelet packet energy spectrum and wavelet packet energy entropy.The results show that during the highlighting process,acoustic emission signals exhibit low-frequency and high-amplitude changes,energy is concentrated in dominant frequency bands,the energy entropy value of wavelet packets decreases,and the rate of change of the energy entropy value of acoustic emission signals can reflect a prominent trend.The evaluation index system of gas outburst situation is constructed,and a gas outburst assessment model based on information fusion is established.In order to solve the impact of uncertain factors such as randomness and ambiguity on gas outburst assessment,a gas outburst assessment method based on cloud model-improved evidence theory is proposed.The mass function of evidence body is constructed by using cloud model,and the combination weighted evidence theory is used to reduce the degree of conflict between evidences and improve the accuracy of gas outburst situation assessment.A method of gas outburst situation prediction based on machine learning is proposed.Using Beetle Swarm Optimization(BSO)to optimize the hyper-parameter combination of Long Short-term Memory(LSTM),a gas concentration prediction model of BSO-LSTM was established.The spatial and temporal correlation of gas concentration on heading face is analyzed,and the input of prediction model is optimized from the perspective of time and space.The results show that the prediction accuracy of the gas concentration prediction model based on BSO-LSTM coupled with space-time is improved.The cloud model-improved evidence theory is used to assess local prediction of outburst situation based on gas emission monitoring information.As for the global prediction of coal and gas outburst situation,the situation evaluation results are quantified as situation values,and a coal and gas outburst situation value prediction model based on generalized regression neural network(GRNN)optimized by chaos immune particle swarm optimization(CIPSO)is established,and the short-term prediction of coal and gas outburst situation is realized.The engineering test results show that the proposed method can accurately sense the coal and gas outburst threat faced by the heading face.Using gas pressure,gas content,drilling cuttings and other indicators to verify the use of gas emission,acoustic emission and other real-time monitoring information to perceive the situation of coal and gas outburst in the heading face,which further shows that the paper can improve the ability of coal mine to prevent and control coal and gas outburst disasters,and ensure the safety production of the mine.The paper has 91 figures,29 tables and 188 references.