Research On The Evolutionary Relationship Modeling Between Emergency Scenarios And Reasoning Method

In recent years,emergency management in China has developed rapidly.The Ministry of Emergency Management has been established by the institutional reform of the State Council,and public attention has been continuously raised.Among them emergency decision-making is one of the important research directions of emergency management.With the progress of science and technology,a large number of emergency cases and data have been accumulated in various fields,recording the development process and characteristics of emergency events.Exploring the empirical rules in these historical events will help people acquire and accumulate relevant knowledge,which can be applied to new problems in the future and promote the development of human cognition under the fourth paradigm.At present,the "scenario-response" model has been widely recognized by the academia and become the mainstream emergency decision-making paradigm.Therefore,from the perspective of scenarios,it is an important scientific research problem,faced by the decision-makers when making scientific and effective emergency plans,that how to apply the evolution law of historical events to the reasoning of the current emergency evolution.In order to better assist emergency decision-making,identify the evolution patterns of historical events and solve the uncertainty problems in the process of emergency scenario modeling and reasoning,this paper proposes a method for the evolutionary relationship modeling between emergency scenarios and scenario reasoning to realize the efficient reasoning process based on real-time scenarios.First,from dimensions of disaster-affected factors,disaster-inducing factors,disaster-pregnant environments and response activity,the common characteristics of elements and their complex interaction relations within the event system was described from the perspective of cognitive science.Then,this paper proposed the Emergency Scenario Model based on the knowledge element and analyzed the common characteristics and mechanism of the evolutionary process of emergency scenarios.After data discretization and normalization the topological structure of BP neural network was constructed based on two emergency scenarios.Then the Levenberg-Marquardt algorithm was used to train the network model to obtain the optimal parameters,and the modeling process of complex nonlinear interactions between emergency scenarios was realized.In the face of a real?time scenario,a belief rule base was constructed through expert's experience and BP neural network.This paper introduced RIMER to realize the emergency scenario reasoning process.After determining the antecedent attribute weight through the correlation coefficient between attributes,the real-time scenario information was transformed into the input based on the fuzzy theory,and the target scenario was inferred through the belief rule activation and combination.Finally,taking the forest fire data in the National Fire Incident Reporting System as an example,the forest fire scenario model was established,and the implementation process of the proposed method was described.Compared with other algorithms,BP neural network has higher accuracy in evolutionary relationship modeling of historical events.In the whole process,various uncertainties in scenario modeling and reasoning are solved by making full use of experts' experience and evolutionary rules.By comparing the reasoning results and the actual values,the feasibility and effectiveness of the proposed method are proved,which provides theoretical support for quick response and scientific decision-making.