基于GIS的BP神经网络洪涝灾害评估模型研究

A Study on the Neural Networks Assessment Model of Flood and Waterlog Disaster Based on GIS

洪涝灾害损失的大小不仅与洪灾的自然属性有关,还受地形、天气气候、人口、社会经济状况及其分布等因素影响。基于GIS技术及其背景数据,实现GIS空间信息单元格点上淹没水深的模拟和空间社会经济数据的展布,并针对影响洪涝灾害评估的复杂因子对各空间单元格点损失评估的不确定性和复杂性,提出了BP神经网络计算方法,运用matlab神经网络工具箱实现区域洪涝灾害的快速评估,建立了基于GIS的BP神经网络洪涝灾害评估模型。运用此方法,通过少量的样本资料,对鄱阳县洪涝灾害经济损失个例进行评估,评估结果误差为12%。

Floods across China have very different characteristics. There are so many types of flood, depending on geography, chimate/weather characteristics, and human population, social and economic situations, etc.Thus, assessment of flood damage is a complex task. Base on the GIS technology and its database, the paper manages to simulate the flood depth and the distribution of economy on each GIS grid. And to solve the uncertainty of the influences of various complex factors on the economic loss assessment precision, Back Propagation Neural Networks are applied by the madab neural network tool - boxes. Thus, a model of flood damage assessment is constructed. In this way, with the help of a few sample datas, the certain Flood and Waterlog Disaster Assessment can be completed more quickly and precisely in Poyang, and the margin of the error is 12 percentage points.