| The reality of "rich coal","poor oil" and "little gas" in China determines that coal is still in an important strategic position in China's fossil energy for a period of time.With the gradual exhaustion of high quality shallow buried coal seams,the central and eastern provinces will gradually enter the stage of deep coal mining.Due to the combined action of soft coal,low permeability of coal rock,high in-situ stress,high gas content and high gas pressure,a large amount of gas energy accumulated in underground coal seam bursts out instantly in the process of mining and driving in deep coal mine,resulting in coal and gas outburst accidents.The facilities inside the mine are damaged,the supporting structure of the mine is disintegrated,the casualties are caused and the safety production is adversely affected.At present,the research on coal and gas outburst in China is at the leading level in the world,but the research is limited to the risk assessment of gas outburst in shallow coal mines,and there are few researches on the risk assessment of disaster in deep coal mines,and the research on the safety risk assessment methods of deep coal mines is just emerging.Therefore,innovative research is carried out on the basis of algorithm fusion,and a set of safety evaluation method system for gas outburst closely combined with deep coal mine scenarios is proposed by using cutting-edge and applicable theories and methods to realize the integration,integration and sharing of deep coal mine gas safety risk data.It is of great practical significance to improve the safety control ability of coal mines and guarantee the safety production of deep mines by changing the passive control of safety risk to the comprehensive control,risk pre-control and source prevention of safety risk.This paper mainly focuses on the evaluation of coal and gas outburst in deep coal mines,and follows the general idea of embedding the evaluation model into the target mine:Combining theoretical analysis and empirical research,with scientific management,safety science and mining engineering theory as the foundation,combining with the intelligent information technology,mathematical statistics and multidisciplinary technology such as system engineering method,the data processing,and evaluation criteria and algorithm optimization,etc,to carry out innovative research,fully considering the deep coal mine,comprehensive research carried out in different stages of the combination of objective evaluation,The advanced and systematic safety evaluation method of gas outburst risk in deep coal mine is constructed.Firstly,the research status,hot spots and cutting-edge methods of coal and gas outburst at home and abroad are visualized.The knowledge map of coal gas outburst is drawn by using the combination of Citespace and Vosviewer,and the research context,trend and hot frontiers of this field are summarized.The research status of coal and gas outburst is visually and quantitatively presented.Secondly,a set of high-dimensional data reduction method for deep coal and gas outburst is proposed.By systematically sorting out the mechanism and influencing factors of coal and gas outburst in deep coal mines,14 index systems for initial evaluation are determined and data collection is carried out according to the characteristics of target coal mines.At the same time for deep coal mine coal and gas outburst is complex,high dimension,objective characteristics of sample data,considering the entropy weight method and gray system in the treatment of the objective data and unknown the unique advantage of uncertain information,first using the improved grey relation analysis based on entropy weight data dimension reduction process,and then by rough set,the improved genetic algorithm to simulate the development process of the evolution of species Data simplification and approximate pattern classification are carried out on the premise of retaining useful information in accordance with the selection method of competitive and merit-based improvement.The second dimension reduction of target coal mine data is realized.After EWM-GRA and GA-RS dimensional-reduction,four main controlling factors,namely mining depth,original gas pressure,original gas content and analyzable gas content,were selected as key evaluation indexes from 14 influencing factors.Finally,a series of intelligent combination evaluation is carried out for the complex nonlinear characteristics of deep coal and gas outburst.The coal and gas outburst in deep coal mine have typical nonlinear characteristics.The premise of accurate evaluation of deep coal and gas outburst is to use appropriate methods to capture the nonlinear change law of deep coal and gas outburst.Based on this,in this paper,intelligent algorithm and deep learning are effectively integrated on the basis of complex system theory.Through appropriate transformation,coal and gas outburst evaluation of deep coal mine is transformed into a model suitable for a specific intelligent optimization algorithm.Through the design of the relationship between coal and gas outburst in deep coal mine,various adaptive performance indexes are determined to describe the process of gas outburst in deep coal mine,and the dialectical relationship between global search and local search is solved by combining algorithm design in further deep learning and model training.To comprehensively improve the intelligent level of data processing,improve the evaluation accuracy and generalization ability of the evaluation model,and achieve the research purpose of effective prevention and control of deep coal mine gas outburst.Therefore,based on the unique advantages of genetic algorithm,particle swarm optimization algorithm,neural network,support vector machine and other machine learning algorithms to solve nonlinear problems,this paper constructed a deep coal and gas outburst evaluation model based on GA-BP on the basis of determining the key evaluation index system of deep coal mine gas outburst risk.Aiming at the deficiency of GA-BP,the PSO-ELM of the training model was improved,and the IQPSO-ELM model based on the characteristics of small samples was designed in combination with the characteristics of the data.A set of intelligent evaluation method was proposed,which could not only deal with the uncertain information but also give the quantitative risk probability value.The transformation from "prediction-response" type to "scenario-response" type of deep coal mine coal and gas outburst evaluation is realized,which effectively solves the problem of deep coal mine gas outburst risk evaluation.In a word,this paper aims at the complex objective reality of high temperature,high pressure,high gas and low coal permeability in deep coal mines,Taking the safety evaluation and application of deep coal mine gas outburst as the research object,based on complex system theory,innovative research has been carried out in sample data processing,data normalization method improvement,complex system evaluation criteria and intelligent evaluation algorithm optimization.According to the characters of deep coal mine gas outburst risk evaluation system data,from two aspects of neural network and support vector machine(SVM)for dimension reduction of deep coal and gas outburst data for model training and evaluation application,improved the traditional processing method of data standardization,broadened the research scope,evaluation theory and method of using the scientific and reasonable system design,The literature review,data dimension reduction,machine learning and intelligent algorithm effectively combine together,to explore and put forward a set of both can handle the unascertained information and gives quantitative risk probability value of the intelligent evaluation method,not only for the scientific evaluation of deep coal and gas outburst risk,promote the deep coal mine safety management provides a new way of thinking,It also provides important reference for scientific evaluation in other high-dimensional and nonlinear fields.Figure [54] Table [21] Reference [251]... |
PDF Full Text Request