利用Elman神经网络模型评估广东热带气旋灾情

Tropical Cyclone Disaster Evaluation in Guangdong Province Based on Elman Neural Network model

热带气旋灾情评估是防台减灾工作的重要环节．介绍的E1man神经网络的基本原理，并将Elman神经网络应用于广东省热带气旋灾害经济损失评估中．选取14个评估因子，选用1998-2008年影响广东的36个台风作为数据样本，再运用主成分分析法降维处理，构建了基于Elman神经网络和基于BP神经网络的评估模型，分析结果显示，基于Elman神经网络模型的评估结果均方误差为0．953，平均误差率为17—76％，优于BP神经网络．研究的模型可实际应用于广东省实际热带气旋灾害经济损失评估，为防台减灾工作提供决策信息．

Estimating of tropical cyclone disaster loss is an important part of the preventing and mitigating disaster work. The paper introduces the basic principle of Elman neural net- work and the applications of Elman neural network in estimation of tropical cyclone disaster loss in Guangdong Province. In total 14 evaluation factors and 36 typhoons data of Guang- dong were chosen to process the reduction of dimensionality, using the method of principal component analysis. The study establishes two kinds of tropical cyclone disaster loss estimat- ing models, including the Elman neural network model and the BP neural network model. The comparative analysis shows that, the testing result of Elman neural network model, the mean square error of which is 0.953 and the mean error rate of which is 17.76%, is better than that of BP neural network. This estimating model can be used in estimation of tropical cyclone disaster loss in Guangdong Province, and provide important information to typhoon disaster reduction and decision support work.