Study Of Meteorological Disaster Assessment Based On Hybrid Optimization Neural Network

The earth’s meteorological disasters have become increasing frequently, and the torrential rain caused by flooding is the disaster which has large influence and fast growth. Long-term since, flood control engineering measures have been the dominant, along with the development of society, people came to realize that the development of non-engineering measurers against floods is very important and necessary. The flood disaster assessment is one of the important links in the entire non-engineering flood controlling measures, its evaluation results provide important basis for disaster management decision.This paper is based on the results of the evaluation system of flood disaster at home and abroad, and to explore the theories and methods of flood disaster intelligence assessment model. For the nonlinear relationship between the evaluation factors and the flood disaster with a high degree of uncertainty, we use the network theory to evaluate the flood disaster. Select the BP neural network which has an extensive application of artificial neural network, and for the shortcomings of its initial weights sensitive, easy to fall into the local optimal, this paper puts forward a new hybrid optimization-based BP neural network for the flood disaster assessment model. This study mainly includes the following aspects:(1) The analysis of flood disaster assessment method. In this paper, we introduced the flood prevention measures of engineering and non-engineering measures, summarized the flood disaster assessment methods, and analyzed the advantages and disadvantages of the existing methods.(2) To solve the Problems of BP neural network learning rate slow and easily falling into local optimal, with the thought to optimize the parameters of BP neural network, we choose the particle swarm optimization algorithm of the intelligence algorithms, and put forward an improved PSO-GA employing crossover operation(IPGC), and then use IPGC to optimize the parameters of storm intensity formula of different repeat periods in Beijing suburban, the application results show that the newly improved algorithm both practical and efficient.(3) The study of the flood disaster assessment model based on hybrid optimization BP neural network. In the design of assessment model, we select three aspects from the evaluation factors, disaster degree rank and the structure of the model to research in-depth, discuss the construction method of the flood disaster assessment model based on the hybrid optimization BP neural network(IPGC-BP), and compared IPGC-BP with the flood disaster assessment model based on the traditional particle swarm optimization algorithm BP neural network model (PSO-BP).The experimental results show that IPGC-BP can reflect the relationship between the multiple evaluation indexes and the disaster level, it can be used for flood disaster assessment, and shows out a good performance on the speed of response and the precision of evaluation.