基于QRNN模型的生命年损失概率密度预测

doi:10.16265/j.cnki.issn1003-3033.2019.05.007

摘要/Abstract

摘要： 为全面评估预测震害损失,提出一种概率密度预测方法。首先,通过改进的生命年损失计算法,获取生命年损失值;其次,采用基于Akaike信息量准则(AIC)的逐步回归分析法,辨识生命年损失的强关联因素,在此基础上构建神经网络分位数回归(QRNN)模型;然后,得到生命年损失预测值与强相关因素的非线性关系,输出不同分位点下生命年损失预测值,运用高斯核函数预测生命年损失概率密度;最后,选取我国1996—2014年的189条地震灾害损失数据作为训练样本,预测2015年10例地震的生命年损失,并与B样条分位数回归(QRBS)模型及3种线性模型作对比。研究表明:基于QRNN模型的震害损失评估概率密度预测,降低了数据依赖性,提高了评估效率;预测值平均绝对误差不超过7.5%,便于震害评估。

关键词: 生命年损失, 逐步回归, 神经网络分位数回归(QRNN), 关联因素辨识, 概率密度函数

Abstract: In order to comprehensively assess and predict earthquake damage losses, a method for probability density prediction is proposed. First, the life-year loss was obtained through the improved lifeyears loss calculation method. Secondly, by use of stepwise regression analysis which is based on Akaike information criterion (AIC), the strongly correlated factors of lifeyears loss were identified and furthermore QRNN model was constructed. Then the nonlinear relationship between the predicted value of lifeyears loss and the strongly correlated factors was obtained, the predicted loss under different quantile points was outputted, and the lifeyears loss probability density was predicted by adopting Gaussian kernel function. Finally, with the damage loss data of 189 Chinese earthquakes from 1996 to 2014 as training samples, the lifeyears loss of 10 earthquakes in 2015 was predicted and compared with quantile regression B-spline (QRBS) model and three linear models. The results show that the damage loss probability density prediction based on the proposed model reduces data dependency while improves evaluation efficiency, and the average absolute error of the prediction is less than 7.5%, which is effective for damage assessment.

Key words: lifeyears loss, stepwise regression, quantile regression neural network (QRNN), correlated factors identification, probability density function

中图分类号:

X913.4

武佳佳, 王威, 朱强强, 马东辉. 基于QRNN模型的生命年损失概率密度预测[J]. 中国安全科学学报, 2019, 29(5): 37-43.

WU Jiajia, WANG Wei, ZHU Qiangqiang, MA Donghui. Lifeyears loss probability density prediction based on QRNN model[J]. China Safety Science Journal, 2019, 29(5): 37-43.

参考文献

[1] 孙龙飞. 城市地震灾害损失评估方法及系统开发研究[D]. 西安:西安建筑科技大学, 2016.
SUN Longfei. Research on the earthquake disaster loss assessment method for urban areas and system development[D]. Xi'an: Xi'an University of Architecture & Technology, 2016.
[2] ILAN N. A non-monetary global measure of the direct impact of natural disasters[R]. Victoria University of Wellington, 2014.
[3] AKAIKE H. Markovian representation of stochastic processes and its application to the analysis of autoregressive moving average processes [J]. Annals of the Institute of Statistical Mathematics, 1974, 26(1): 363-387.
[4] KOENKER R W, BASSETT J G. Regression quantiles[J]. Econometrica, 1978, 46(1): 33-50.
[5] TAYLOR J W. A quantile regression neural network approach to estimating the conditional density of multiperiod returns[J]. Journal of Forecasting, 2000,37(9): 299-312.
[6] 汪志红, 王斌会, 刘明. 基于B样条曲线的GM(2,1)动态组合预测模型应用[J]. 中国安全科学学报, 2011, 21(7): 157-162.
WANG Zhihong, WANG Binhui, LIU Ming. A dynamic combination forecast model based on B-spline curve and GM(2,1) and Its application[J]. China Safety Science Journal, 2011, 21(7): 157-162.
[7] 卫星君, 赵晓萌, 马长玲,等. 降雨型滑坡灾害的约简和逻辑回归预测模型[J]. 中国安全科学学报, 2018, 28(8): 1-6.
WEI Xingjun, ZHAO Xiaomeng, MA Changling, et al. Reduction and logistic regression model for prediction of rainfall landslides disaster[J]. China Safety Science Journal, 2018, 28(8): 1-6.
[8] 许启发, 蒋翠侠. 分位数局部调整模型及应用[J].数量经济技术经济研究, 2011, 28(8): 115-133.
XU Qifa, JIANG Cuixia. Quantile partial adjustment model and its application[J]. Research on Quantitative Economy and Technical Economy, 2011, 28(8): 115-133.
[9] 何耀耀, 郑丫丫, 杨善林. 基于Box-Cox变换分位数回归与负荷关联因素辨识的中长期概率密度预测[J]. 系统工程理论与实践, 2018, 38(1): 197-207.
HE Yaoyao, ZHENG Yaya, YANG Shanlin. Medium and long term probability density forecasting based on Box-Cox transformation quantile regression and load relaltion factor identification[J]. Systems Engineering-Theory & Practice, 2018, 38(1): 197-207.
[10] 颜伟, 李丹, 朱继忠,等. 月前日负荷曲线的概率预测和随机场景模拟[J]. 电力系统自动化, 2017, 41(17): 155-162.
YAN Wei, LI Dan, ZHU Jizhong. et al. Probabilistic forecasting and stochastic scenario simulation of month-ahead daily load curve [J]. Automation of Electric Power Systems, 2017, 41(17): 155-162.
[11] WHO. WHO methods and data sources for global burden of disease estimates 2000-2011[R],2013.
[12] 中国统计年鉴[EB/OL]. [2018-09-03]. http://www.stats.gov.cn/tjsj/ndsj/.
[13] 阎海琴. 年龄中位数指标计算公式的分析[J]. 人口研究, 1985,9(2): 51-52.
[14] CANNON A J. Quantile regression neural networks: Implementation in R and application to precipitation downscaling[J]. Computers & Geosciences, 2011, 37(9): 1 277-1 284.
[15] 许启发, 蒋翠侠. R软件及其在金融定量分析中的应用[M]. 北京:清华大学出版社, 2015: 400-402.
[16] 郑通彦, 郑毅. 2013年中国大陆地震灾害损失述评[J]. 自然灾害学报, 2015, 24(1): 239-246.
ZHENG Tongyan, ZHENG Yi. Review of earthquake damage losses in Chinese mainland in 2013[J]. Journal of Natural Disasters, 2015, 24(1): 239-246.
[17] 郑通彦, 冯蔚, 郑毅. 2014年中国大陆地震灾害损失述评[J].世界地震工程, 2015, 31(2): 202-208.
ZHENG Tongyan, FENG Wei, ZHENG Yi. Review of earthquake damage losses in Chinese main land in 2014 [J]. World Earthquake Engineering, 2015, 31(2): 202-208.
[18] 陈通, 郑通彦. 2015年中国大陆地震灾害损失述评[J].灾害学, 2016, 31(3): 133-137.
CHEN Tong, ZHENG Tongyan. Review of earthquake damage losses in mainland China in 2015 [J].Journal of Catastrophology, 2016, 31(3): 133-137.