地震灾害生命年损失多模型评估方法研究

传统地震灾害损失评估方法没有统一的指标对地震灾害损失进行全面的评估,这造成一些应用上的困难.为了能够对震害损失进行全面评估并进行多模型对比研究,本文给出了基于生命年的地震灾害损失多模型评估方法.通过生命年损失理论的改进以及方法参数取值的分析,确定了计算方案,并以年份、地区分组对我国震例生命年损失进行了对比研究.研究了总生命年损失随各影响因素:受灾人口、震中烈度、震级、恢复时间的变化规律,结果表明:四个影响因素均与总生命年损失呈正相关关系,且受灾人口相关性最大,其余三项依次递减.在影响因素分析的基础之上建立了线性回归模型、神经网络分位数回归模型与基因表达式编程非线性模型,并进行了对比分析.采用多模型对地震灾害生命年损失进行评估预测,减少了数据依赖性并简化了评估流程.

Research on multi-model evaluation methods for earthquake disaster lifeyears loss

The traditional evaluation methods of earthquake disaster loss does not have a unified index to make a comprehensive assessment of the loss of earthquake disaster, which will cause some difficulties in the application. In order to comprehensively evaluate the loss of earthquake disaster and carry out multiple models comparison research, this paper presents multi-model methods for evaluating the loss of earthquake disaster based on lifeyears. Through the improvement of the lifeyears loss theory and the analysis of the value of the method parameters, the calculation scheme was determined. A comparative study was conducted on the lifeyears loss of earthquake in China grouped by years and regions. The relationship between the total lifeyears loss and its influence factors, such as the affected population, the epicentral intensity, the earthquake magnitude and recovery time was studied. The results showed that four influencing factors were all positively correlated with the total lifeyears loss. The affected population has the greatest correlation and the correlation of the other three factors is decreasing in turn. The linear regression model, the neural network quantile regression model and the nonlinear model of gene expression programming were established on the basis of the analysis of the influencing factors and the comparative analysis was carried out. Earthquake disaster loss was evaluated and predicted by using the evaluation model, which reduced data dependence and simplified the assessment process.