Research On Emergency Instruction System Of Long Highway Tunnel Based On Human-machine Hybrid

At present,the emergency decision-making of long highway tunnels mainly depends on the pre-made plans.Sometimes,there are ambiguous mission instruction expressions in the emergency response process,which leads to deviations in the execution of instructions,threatening public life and property safety.Aiming at the traffic accidents of the long highway tunnel,this paper combines the resources of human-machine hybrid to study the emergency command system from three aspects: the model of the knowledge ontology construction in tunnel emergency domain,the mode of emergency command construction and the safety evaluation model of emergency command execution.First,for the problem of "reuse" and "sharing" of knowledge in the traffic accident emergency field of long highway tunnel,by analyzing the highway tunnel traffic accident and its emergency rescue status,combined with historical accidents,relevant laws and regulations,emergency plans,etc,the study establishes the knowledge ontology of the tunnel emergency domain based on the framework of "highway tunnel-traffic accident-emergency instruction".The formal specification description of the ontology sharing conceptual model improves the accuracy of the knowledge representation in the long tunnel emergency domain.In addition,aiming at the situation that the computer can not find the corresponding solution strategy or knowledge of the solution,this study establishes a graph database of man-machine emergency command based on graph database theory.Secondly,it focuses on the problem of emergency commands construction based on human-machine hybrid.The construction is composed of static data and dynamic data.It includes personnel instruction templates for road administration,traffic police,firefighting,medical,obstacle clearance and maintenance,and machine instruction templates including jet fans,lighting,information boards and lane indicators.Then the dynamic data is obtained through four channels: emergency database,resource scheduling,mathematical model and event reporting.Based on JAVA language,the emergency instruction program composed of instruction template module,data retrieval module,mathematical model module and interfacemodule is designed to efficiently construct a complete and professional man-machine emergency instruction.Finally,for the safety problem of emergency instructions in the process of execution,AHP is used to evaluate the execution safety of the generated emergency instructions.Through analysis and expert investigation,the evaluation index system including "tunnel environment-traffic environment-road condition" is optimized.AHP is used to determine the weight of evaluation indicators,and the membership of each indicator is determined according to the relevant standards and regulations to achieve real-time monitoring instruction execution security.Then,based on the GRU neural network,an emergency instruction security prediction model is built,which greatly improves the safety guarantee of rescuers through real-time monitoring and predicting security.Taking the collision accident of Zhongnanshan Tunnel as an application example,the man-machine emergency command of the accident is generated based on the emergency command system of this paper and the execution safety of the command is evaluated.The results show that the theoretical model proposed in this paper is feasible and the command standards constructed are of good practicability,which can effectively deal with traffic accidents in long highway tunnels.