Wildfire Tactic Pattern Recognition And Solution Generation Method Research

In recent years,the problem of global warming is becoming increasingly serious,and the wildfire disaster is becoming progressively more prominent.With the progress of human science and technology and the change of social ideological trend,how to scientifically and critically understand the impact of wildfire on human and nature has become an urgent scientific problem.It is very important to fully understand the positive and negative effects of the wildfire incident,and choose the appropriate tactics and schemes.In addition to the top-level design of response action covered by the emergency plan and the specific action scheme generated based on the three elements of fire scenario,disaster pregnant environment and disaster bearing body,the research of wildfire response action should also include the tactical decision between linking strategic decision and operating decision.Therefore,it is necessary to conduct in-depth research on the coping tactics of wildfire events,establish the pattern recognition model of wildfire coping tactics combined with massive real wildfire data,and analyze the selection of wildfire coping tactics from the perspective of data science.In the case of confirming the coping tactic mode,coping scheme generation can give full play to the advantages of "Tactic + Scheme" combination path,and improve the accuracy and efficiency of case retrieval.In this paper,based on the knowledge element model,the case of wildfire is formally described.Combined with the knowledge analysis of wildfire,four types of coping tactics are divided,including fully extinguished,controlled combustion,laissez faire combustion and planned fire.After data preprocessing,appropriate data features are selected through collinearity analysis,significance analysis and random forest feature importance,SMOTE algorithm is used to process unbalanced data,and RF,LGBM,LSTM and CNN algorithms are comprehensively used to generate a Stacking integrated learning model with four features and five levels of cross validation.After parameter adjustment and optimization,the accuracy of the model is greatly improved.This part of work can quickly and accurately make the choice of new wildfire response tactic mode,which simplify the case data and point out the decision-making path for the next generation method of emergency plan.In addition,considering the demand that highly similar cases are not unique in the process of emergency plan generation in the massive data environment,a scheme effectiveness evaluation based on cost benefit analysis is designed to generate the emergency schemes.The information entropy is used to calculate the similarity of case scenario,response action and event result,and then the fusion weight is calculated by means of data envelopment analysis,and finally the emergency scheme is generated.In the case of one to many similar cases matching,the specific emergency scheme is generated by case fusion,which makes up for the shortcomings of the original case based reasoning.Finally,taking the wildfire data of the American NFIRS fire system as an example,this paper expounds the actual generation process of wildfire response strategies and decisionmaking schemes,and verifies the applicability and feasibility of the model.