Research On Causation Analysis And Management Control Of Gas Accident Of Coal Mine Based On Probabilistic Inference

China's coal mining is a high-risk industry,with serious accidents and disasters,which have brought huge loss of life and property to the nation and the people.Among all kinds of accidents in coal mines,gas accident is the most serious one,which has been regarded as the "number one killer" of coal production.As a complex social and technical system,there are many internal and external factors that affect the occurrence of gas accidents in coal mines.The uncertainty of accident causes and conditions brings difficulties to the management and control of gas accidents.By analysis of the uncertain factors,conditions and probability changes that lead to the occurrence of coal mine gas accidents from the perspective of probability reasoning,and mechanism of coal mine gas accidents,the probability graph model and scenario analysis are applied to find the control countermeasures of coal mine gas accidents under the new situation.Firstly,the general situation of coal mine accidents is analyzed in China from a historical point of view to reveal the occurrence characteristics,change trends and causes of coal mine accidents in various stages since the founding of the People's Republic of China.The statistics focuses on the analysis of the current situation of gas accidents in horizontal and vertical directions,and points out the characteristics of coal mine gas accidents in the attributes of accident type,mine type,occurrence region,occurrence time,etc.Combined with the current situation and trend of coal mine situation,the problems in environment,personnel,equipment and management of coal mine gas accidents are pointed out.From the analysis results,coal mine gas accident has the characteristics of uncertainty,such as the severity of disaster consequences,the universality of regional distribution,and the randomness of occurrence time.The dynamic change and uncertainty of factors bring difficulties to coal mine safety management.In the case of limited input and management resources of coal mine safety,it is necessary to make full use of data information to study the cause of gas accident and the uncertainty of conditions,so as to improve the traditional safety management mode and the pertinence of gas accident control.Secondly,the uncertainty of gas accident in the production process of coal mine system is taken as the research object according to the physical mechanism of gas accident,combined with the classification model of accident cause.First of all,using the fault tree analysis to explore the internal and external factors and the logical conditions between them in terms of personnel,machine,environment,management from a large number of latest gas accident cases.The condition dependent model of gas accident cause and the main factors of accident are stressed.Secondly,the collected case data are used to build the Bayesian network model with the characteristics of coal mine gas accident by combining machine learning and expert experience followed by the verification and validation of the effectiveness of the model.Finally,based on the Bayesian network of gas accident,the accident reasoning is fulfilled to find all kinds of measures to reduce the risk by fixing maximum cause chain and control conditions of the accident.By determining the posterior probability of accident nodes under the influence of different factors,the probability of coal mine gas accident can be effectively determined.According to the maximum cause chain of coal mine gas accident,the cause and effect chain of gas accident can be fixed quickly.It can effectively reduce the probability of accidents to manage and control the sensitive factors in the causal chain.The results show that the randomness of gas accidents in coal mines can be studied from the perspective of probability.Compared with the traditional accident analysis method,Bayesian network has advantages in the expression and reasoning of complex uncertainty problems.The application of Bayesian network to the study of gas accident uncertainty can effectively integrate the characteristics of gas accidents.The prior knowledge and current information of accidents can realize the risk prediction and cause analysis of coal mine gas accidents based on probability reasoning,which makes clear the key points and ways for the prevention and control of coal mine gas accidents.Thirdly,in order to apply the Bayesian network model of coal mine gas accident to accident analysis and prevention,scenarios are constructed based on the change of conditions and the possible future situation of coal mine.Combined with the feature of gas accident,a coal mine gas accident situation network model(CFSE)is constructed based on "coal mine characteristic factors-influencing factors-state-event",and probability situation analysis is carried out.In order to determine the gas accident control process differently from the traditional way,the coal mine gas accident management countermeasures with Bayesian thought from the decision-making level,management level and operation level are put forward.The results show that the analysis of the possible scenarios of multiple systems is helpful to manage and control the complex systems with uncertainties in the future.The construction of gas accident scenario network model can determine each scenario corresponding to the accident expectation.In any scenario,the risk of gas accident can be inferred by Bayesian network,and the state of factors can be changed by events,so as to reduce the risk of gas accident in coal mines.In the coal mine gas control strategy,the coal mine administrators should make full use of the inference and information update mechanism of the Bayesian network of gas accident to establish the gas accident control mode based on probability inference and scenario analysis,to establish the early warning platform of the probabilistic inference of coal mine,to refine the feasibility measurement of the risk sources of coal mine gas accident.After this,the perception of the changes of the system factors and conditions,and each department should form a grid type management,from the overall perspective of decision-making and judgment and then take targeted measures.In Summary,the research is driven by data,based on Bayesian network and scenario analysis theory,through probability reasoning method to quantitatively study the uncertainty of coal mine gas accident.Management and control suggestions of coal mine gas accident systematically are put forward under different scenarios,so as to improve the pertinence and effectiveness of coal mine gas accident management and control and to minimize the occurrence of coal mine gas accident in China.There are 63 figues,32 tables and 201 references in this dissertation.