基于BP神经网络和VSM的台风灾害经济损失评估

统计了对广东省造成直接经济损失的台风数据,包括致灾因子、孕灾环境和承灾体等评估因子,对数据进行无量纲化、归一化处理,作为验证模型的数据集。建立BP神经网络和空间向量模型(VSM)相结合的综合评估模型,利用BP神经网络进行初步预测,基于VSM对预测结果进行修正,从而构建台风灾害经济损失评估模型。将收集的历史台风经济损失数据分为训练和测试集,对模型进行训练和检验。经验证,采用BP神经网络和VSM相结合的台风灾害经济损失评估模型能够有效降低训练数据不足对评估结果的影响,平均误差率由30%降低到14%。

Economic Loss Assessment of Typhoon Based on BP Neural Network and VSM

We collecte the historical data of economic losses triggered by typhoon disasters into a typhoon data set, including disaster-causing factors, disaster-pregnant environment and hazard bearing body. The data set undertaken the process of dimensionless and normalization before being used to testify and verify the model. In order to evaluate the economic loss of typhoon, we establisha comprehensive evaluation model. It began with using BP neural network to evaluate economic losses of typhoon disaster. Then, the evaluation results are remedied via VSM model before the comprehensive evaluation model is built. The collected data is divided into two data sets, one for model training, and the other for testing. The experimental results showed that the comprehensive evaluation model is effective. More specifically, it could avoid the influence the shortage of training data brought on the evaluation results and the average error rate is reduced from 30% to 14%.