Emergency Decision-making Methods For Extremely Major Natural Disasters Based On Evidence Theory

The frequent occurrence of extremely major natural disasters has caused huge economic and human losses,and has seriously affected the sustainable development of society.The unpredictability of natural disasters has brought huge challenge for decision makers,and it has become a research topic of great significance worldwide.Although some progress has been made in the emergency management of natural disasters,the ability to make emergency decisions in actual disasters is still quite limited.So the decision-making problems need to be focused in different natural disaster scenarios.Through literature analysis and background research,three key elements are identified in the emergency response of extremely major natural disasters: rescue plan,rescuers,and rescue supplies.Then we determine three research problems that need to be addressed immediately.The first is to choose an emergency plan,then the dispatch of rescuers and the forecast of emergency supplies demand are needs to ensure the smooth implementation of the plan.In order to effectively deal with the uncertainty in the environment of extreme natural disasters,evidence theory is combined with multi-attribute decision making method,optimization modeling method and case-based reasoning method,conducts research on the above three problems.Firstly,facing the uncertainty of the emergency decision-making environment for extremely major natural disasters,a multi-attribute risk decision-making method for emergency plan selection is proposed based on evidence theory.First,evidence theory and Pythagorean fuzzy set are combined to propose a natural disaster multi-attribute emergency plan selection method,which not only realizes the effective expression of fuzzy information for decision makers,and solves the basic problem of emergency plan selection;second,in order to solve the problem of interactive multi-attribute emergency plan selection,the non-additive measure determination method and the attitudinal nonlinear integral method are proposed to measure and calculate the correlation between attributes in different states.Furthermore,an emergency plan selection method is constructed;third,considering the important role of public participation in emergency response,the web crawler technology,TF-IDF algorithm and Word2 Vec method are used to mine public-level attribute information and calculate the weights.Then the expert judgments are aggregated to make the final decision.This chapter enricifies the theory of emergency plan selection.Secondly,aiming at the situation of multiple disaster areas caused by extremely major natural disasters,a model of rescue personnel dispatch under the situation of multiple rescue points and multiple disaster areas is established.First,in order to define more flexible and accurate information expression methods to represent the subjective judgments of decision makers,evidence theory and linguistic term sets are combined to propose the concept of evidential linguistic term sets,and the basic operation,aggregation operator,distance measure,entropy function and score function on it are defined,which not only lay a theoretical foundation for the establishment of a rescuer dispatch model,but also can be used as a way of expressing information in an uncertain linguistic environment in decision-making and evaluation fields;second,in order to determine the weight of the criteria for evaluating the ability of rescuers,evidence theory and best-worst method are combined to propose the evidential best-worst method,which makes up for the shortcomings of traditional weight calculation approaches in representing uncertainty;third,to address the issue of dispatching emergency rescuers for natural disasters with multiple rescue points and disaster areas,based on the above proposed methods and the optimization modeling method,the model is constructed with the goal of maximizing the competence of rescuers and satisfaction of the rescue time,and the optimal plan for the assignment of rescuers is determined by solving the model.Finally,the empirical analysis is done to illustrate the advantages of the proposed method in improving the scientificity of multitasking group decision-making.Finally,facing the forecast of emergency supplies for extremely major natural disasters,a two-stage forecasting method based on evidence theory and case-based reasoning is proposed.First,for the problems of feature value missing,feature heterogeneity and feature interaction in case-based reasoning,a case retrieval strategy is proposed based on evidence theory,which not only lays a theoretical foundation for the subsequent emergency supplies demand prediction,but also improves the deficiencies of case retrieval strategy;second,in order to predict the emergency supplies demand effectively in the absence of decision basis in the initial stage of natural disaster,a natural disaster scene matching method is proposed based on the proposed case retrieval strategy.This method can find the historical case most similar to the current disaster,and adjust the solution of the case according to the actual decision-making environment,so as to realize the reuse of the case to predict the loss of the current disaster;third,with the development of time,the impact of natural disasters will change constantly.Based on the predicted loss results,a dynamic decision model of emergency supplies demand is further constructed.This method not only provides a new way for emergency supplies demand decision-making for sudden natural disasters,but also lays the method and model foundation for emergency supplies storage and allocation.In summary,this thesis starts from the three key elements in the emergency response to natural disasters,namely rescue plans,rescuers,and rescue supplies,and constructs an emergency decision-making framework under different scenarios.This thesis uses evidence theory,optimized modeling methods,machine learning methods,case-based reasoning methods,and multi-attribute decision-making methods to develop emergency decision-making methods under different scenarios.It not only contributes outstanding theoretical value to emergency decision-making,but also provides important suggestions and guidance for the emergency management of natural disasters in reality.