神经网络集成在洪灾损失评估中应用的的研究.pdfVIP

Abstract Every year, floods caused severe economic losses to China, to some extent, affected Chinas rapid economic development and construction, to assess the losses caused by the floods, flood control work, the theoretical basis for the development of flood control decision-making provided an important basis for the first thing to scientific, accurate and efficient assessment of losses caused by floods. There are many problems existed in current flood assessment,such as low assessaccuracy,heavy workload and complex operation,this paper raised a flood damage assessment model which is based on neural networks ensemble.First of all,this paper introduced the formation mechanism of flood,the types of losses caused by flood damage and flood damage assessment principles.In order to study the Poyang Lake region Duchang county,used the analytic hierarchy process method to analyze and extracted flood assessment indicators which can affect the study area flood losses scientifically,rationally,objectively, and divided their weights ；And then,this paper highlighted and discussed the flood damage assessment modelwhich is based on neural networks ensemble,after analysised and researched the commonly used flood damage assessment model and algorithmic method ；Once more,focused on expatiating on how to produce individual neural networks in ensemble and how to combine their output results,setting correlation parameters in individual neural networks,and built a program which can fast construct a neural networks ensemble with C# program language and AForge.NET open source frame,and proved the ease of use, reliability and stability of the neural networks ensemble by a experiment ； Finally,the flood damage assessment model which is based on neural networks ensembl