Research And Implementation Of Assistant Decision-making Method For Railway Emergency Rescue

With the rapid development of the railway transport industry,higher requirements are put forward for the guarantee of safe transportation.At present,the concept of "one case,three systems" is regarded as the central idea of China’s railway emergency management system.Under the multiple guarantee of railway management system,operation mechanism and emergency legal system,based on the railway emergency disposal plan,combined with the situation of sudden accident scene and relevant rescue experience,emergency disposal decisions are made to implement rescue.In the process of making the disposal decision,the rescue experience of the decision maker directly affects the reasonable degree of the disposal decision.At the same time,the rescue experience is closely related to the rescue operation of historical unexpected accidents.Therefore,it is of great significance to assist decision-makers to complete the current emergency disposal plan through the rescue decision-making of similar historical accidents.In this thesis,the idea of case-based reasoning is integrated into the process of railway emergency disposal plan,and the auxiliary decision-making method of railway emergency rescue is taken as the research object.The purpose of this paper is to use the scientific and reasonable case retrieval method to obtain the disposal plan of similar historical accidents in a very short time to assist in the decision-making of the current emergency.Firstly,starting from the railway emergency itself,the characteristic attributes of railway emergencies are extracted and divided.The abstract representation model of emergency is established by analyzing the hierarchical structure of emergency cases.The case representation is realized based on XML language.Secondly,rough set theory is used to calculate the weight of case feature attributes.In the attribute similarity evaluation,the feature attributes are divided into different types,and different similarity measurement methods are set according to the type.In order to deal with the problem of missing attributes,case structure similarity is introduced.The overall similarity between cases is obtained by combining attribute and structure similarity.In order to match the case more accurately,the confidence theory of rescue assistant decision-making is introduced.Through the analysis and calculation of the potential indicators that affect the confidence coefficient,the confidence decision tree is established by combining the potential index data with C4.5 gain rate algorithm.A railway emergency rescue auxiliary decision matching method based on confidence coefficient mixed similarity is designed.Thirdly,in view of the lack of accident description information due to the low level of informatization or confusion at the accident scene,resulting in the reduction of case similarity.Through the analysis and calculation of railway accident types and the core attributes of various types,a double-layer Bayesian classification model for railway emergencies is constructed based on the idea of naive Bayesian classifier.A matching method of railway emergency rescue auxiliary decision based on double-layer Bayesian classification model is designed.A training method of double-layer Bayesian classification model is established according to the update of case base and case information.Finally,the design and implementation of the railway emergency rescue assistant decisionmaking system is completed.The functional requirements,software architecture,modeling and functional implementation of the system are analyzed systematically.The research content of this paper is verified by a simulation example combined with the main functions of the system.The verification results are in line with the expected results of the theoretical analysis of this paper,indicating that the railway emergency rescue assistant decision-making method proposed in this paper is feasible and effective in the field of railway emergency rescue.