Study And Application Of Agent Rescue Decision Model Combined With DBN-CBR

All the time, a variety of coal mine disasters has brought huge economic lossesand casualties to the coal industry of China, caused great harm to the sustainabledevelopment of the coal industry and social harmony and stability of China. Thetraditional coal mine virtual simulation technology is to use virtual reality technology,model and simulate the environment of the coal mine,according to the variety ofdata of coal mine. In this way, we can minimize the incidence of coal mine disastersand improve the workers’ ability to respond to coal mine disasters through mineworker safety training by simulating variety of coal mine disasters demo in coalmines virtual scene.Case-based reasoning technology has the ability to reuse historical experience,which is a big guiding significance to the coal mine disaster, while Agent’sintelligent and dynamic is particularly applicable to the complex and volatile issues.Based on CBR technology and intelligent agent technology, considering the highincidence of duplication of coal mine disasters and the dynamic changes ofenvironments, specific studies of this paper are as follows:First of all, based on intelligent Agent and the analysis of coal mine disasters,research a task-oriented multi-Agent architecture, combine with the Agentarchitecture,i improve safety evaluation method LEC, for site evaluation of the minedisaster. The method provides reference data for risk rating for the subsequent rescuedecision-making, and speeds up the coal mine disasters rescue decision-making.Secondly, basing on traditional case-based reasoning, combined with dynamicincremental new expectation maximization Bayesian network parameter learningmethod, and using the advantages of Bayesian network, construct the DBN-CBRmodel. The model provides a good similarity evaluation function of the mine disasteraccident emergency rescue decision-making. On this basis, put forward the case baseorganization: K-D tree structure, improves the efficiency of case retrieval effectively.Finally, combined with the DBN-CBR model and parameters from the disastersite evaluation, design and realize virtual coal mine disaster accident emergencyrescue decision-making simulation system, and verify the effectiveness of theproposed model and the idea by an example of coal mine gas explosion.