中国安全科学学报 ›› 2019, Vol. 29 ›› Issue (1): 167-172.doi: 10.16265/j.cnki.issn1003-3033.2019.01.028

• 公共安全 • 上一篇    下一篇

安全管理水平量化评估指标权重优化方法

杜军威1 教授, 于洋洋1, 胡强1 副教授, 刘勃2   

  1. 1 青岛科技大学 信息科学技术学院,山东 青岛 266061;
    2 中国石化安全工程研究院,山东 青岛 266071
  • 收稿日期:2018-09-10 修回日期:2018-11-03 出版日期:2019-01-28 发布日期:2020-11-23
  • 作者简介:杜军威 (1974—),男,山东青岛人,博士,教授,主要从事软件工程、人工智能等方面的研究。E-mail:djwqd@163.com。
  • 基金资助:
    国家自然科学基金资助(61273180);山东省重点研发计划项目资助(2016GGX101031,2018GGX101052)。

A method for optimizing index weight in quantitative evaluation of safety management level

DU Junwei1, YU Yangyang1, HU Qiang1, LIU Bo2   

  1. 1 College of Information Science and Technology, Qingdao University of Science and Technology, Qingdao Shandong 266061, China;
    2 Sinopec Research Institute of Safety Engineering, Qingdao Shandong 266071, China
  • Received:2018-09-10 Revised:2018-11-03 Online:2019-01-28 Published:2020-11-23

摘要: 为使石化安全管理评估指标体系更好地适用于时间、国家政策以及实际管理需求的转变,利用领域专家对已有事故案例集中的诱因指标重要度评价信息,基于二元语义分析方法,获取各样本的修正贡献度;综合样本的修正贡献度、时间、样本危害评分等要素,提出基于熵权法的指标权重计算方法;在该方法的基础上,开发相应的石化安全管理评估指标体系的指标权重修正系统,采用 1 874 个样本分批次修正某研究院已有指标权重。结果表明:随样本数量增加,修正后的指标权重曲线与优秀实践指标均值曲线的K-L散度越来越小,修正后的指标体系权重接近优秀实践指标均值权重,验证了该方法的合理、有效性。

关键词: 安全管理, 二元语义, 样本贡献度, 熵权法, 指标权重调整

Abstract: In order to make the indicators of the petrochemical safety management evaluation index system more applicable to the variation of time, national policy and practical management requirement, based on the important evaluation information on incentive indicators in the existing sample accident cases by domain experts and the two-tuple linguistic information analysis method, the revised contribution degrees of the samples were obtained. Then the entropy weight method was introduced to compute the weights of indicators based on the amended contribution degree, time and hazard rating of all the sample accidents. An indicator adjusting system was developed based on the proposed method. The adjusting system was employed to optimize the existing weight of the indicators in the risk assessment system for a certain safety engineering research institute.The adjustments were performed in batches based on 1 874 samples. The amended results show that the K-L divergence of revised index weight curve and excellent practice index mean curve is getting smaller, and that the revised index system weight is close to the average weight of excellent practice index, so the effectiveness of the proposed method is demonstrated.

Key words: safety management, two-tuple linguistic, sample contribution degree, entropy weight method, index weight adjusting

中图分类号: