供电作业人身事故危害多元线性回归预测  

Prediction on human casualty hazard in power-supplying operation based on multiple linear regression

作　　者：魏秀宁[1] 柳亦钢[1] 黄玉龙[2] 陈灌枢 萧镜辉[1] 叶钜芬 王韵陶 张勇军[3] 

机构地区：[1]广东电网公司东莞供电局,广东东莞523008 [2]暨南大学电气信息学院,广东珠海519070 [3]华南理工大学电力学院,广东广州510640

出　　处：《中国安全生产科学技术》2015年第7期178-184,共7页Journal of Safety Science and Technology

基　　金：国家自然科学基金项目(51377060)

摘　　要：为了定量评估电网作业和运行的风险,提出了一种基于5M因素分析法和多元线性回归的供电作业人身事故危害预测方法。借鉴安全行为学、安全心理学和事故致因理论,首先将供电作业中影响人身安全的因素分成5个大类,即人、设备、材料、方法和环境。然后分析所有因素间的相关性,将相关性较强的影响因素中对人身事故危害程度弱的一方剔除,并采用南方某电网公司人身伤亡事故数据为例进行分析,删除了两个人身安全影响因素：工器具配置和作业性质。继而,采用多元线性回归法建立供电作业人身事故危害值与各人身安全影响因素间的关系,对供电作业人身事故危害值进行预测。最后,以南方该电网公司2003～2013年发生的64起供电作业事故验证所提方法的有效性和准确性。In order to evaluate the risk in power grid operation quantitatively,a prediction method on human casualty hazard in power- supplying operation based on 5M factors analysis method and multiple linear regression was put forward. By referencing safety behavior,safety psychology and accident- causing theory,the factors that influence the human safety in power- supplying operation were divided into five types,namely man,machine,material,method and environment. The correlations among all these factors were analyzed and those influencing factors which both have high correlation with others and exert weaker effect on human hazard were deleted,and two influencing factors of human safety, ＂ tools configuration＂ and ＂ operation character＂,were deleted by analyzing the historical human accident data of a Power Grid in South China. The relationship between the human casualty hazard in power- supplying operation and those influencing factors of human safty was established by using the multiple linearregression method. Finally,the effectiveness and accuracy of the proposed method were demonstrated on 64 real accidents happened in this Power Grid of South China from 2003 to 2013.

关 键 词：人身危害预测 多元线性回归 相关性 5M法 

分 类 号：X928.03[环境科学与工程—安全科学]