Considering that the problems of low accuracy,heavy workload and complex operations in traditional flood loss assessment,this paper puts forward a new method of rapid flood loss assessment based on neural network ensemble.Firstly,the author analyzes the current flood loss assessment method and puts forward rapid flood loss assessment based on neural network ensemble.Secondly,taking the country in Poyang lake area as the research object,using analytic hierarchy process to analyze and extract the index of the flood loss evaluation which impacts the area,the weights of impact factors are assigned.Then,doing the study on the method of individual generation and conclusions of the neural network ensemble model,and using the C＃ programming language and the neural network class library in Aforge.NET open source framework to build a neural network ensemble model rapidly.Finally,testing this method,and comparing the results with the actual flood loss value to prove the feasibility of the assessment model,thus providing a new approach for flood loss assessment.