Study On Casualty Assessment Method Of Destructive Earthquake

China is an earthquake-prone country.The earthquake caused great casualties to the people in our country,as well as great losses to the economy.After an earthquake,casualties can be quickly assessed through seismic data.According to the results of earthquake casualty assessment and expert opinions,the emergency rescue of earthquake can be guided,the economic losses can be reduced,the number of rescued people can be increased,and the waste of relief materials can be reduced.Based on the assessment data of earthquake disaster losses in China from 1966 to2021,the basic data of earthquake type is completed.According to the relevant earthquake cases,the formula of earthquake casualties based on the type of earthquake is fitted by empirical regression method.An earthquake casualty assessment method based on random forest algorithm is established,by machine learning method and big data.The main content of this article are as follows:1.The current method for evaluating earthquake casualties is introduced.The existing research methods of earthquake casualty assessment are elaborated from five aspects,including epicenter intensity,building vulnerability,population,multiple factors and machine learning.For each method,the corresponding principle is introduced,the corresponding formula is listed,the advantages and disadvantages of each method are compared,and the application scope of different methods is discussed.2.The source and processing of the data are explained.The effects of earthquake magnitude,intensity,earthquake type,focal depth and population density on earthquake casualties are analyzed.When other factors are the same or similar,the scatter diagram of earthquake magnitude factor and earthquake casualties is drawn.At the same time,the scatter diagram of earthquake type factor and earthquake casualties,the scatter diagram of focal depth factor and earthquake casualties,and the scatter diagram of population density factor and earthquake casualties are drawn.Different function models are selected for scatter fitting according to the image trend.The fitting effects of different function models were compared to determine the function models for different factors.3.The evaluation method of earthquake casualties based on different types of earthquakes is studied.By collecting and summarizing the data of predecessors,the basic data of earthquake types in earthquake cases in China are completed.According to the division of fault types,the earthquake types are divided into four types: reverse fault,strike-slip fault,strike-slip normal fault and strike-slip reverse fault.According to the case data of each type of earthquake,two empirical fitting formulas based on earthquake magnitude and intensity are regressed.Then the two formulas are compared,and the formula with good regression effect is selected as the empirical fitting formula of each type.Finally,the reliability and accuracy of the method are analyzed by comparing the estimated number of earthquake casualties fitted by the formula with the real number of earthquake casualties.4.Different types of machine learning algorithms are introduced,and the principle and important parameters of random forest are explained.According to earthquake examples,the evaluation model of earthquake casualties based on random forest algorithm is established,and the weight of each factor is calculated.Each factor is analyzed,and then redundant factors and unimportant factors in the model are removed.The evaluation value of the model is compared with the result of real earthquake casualties,and analyze the reliability and accuracy of the model.