基于DPSIR和BP神经网络的安全绩效评估模型  

Safety performance evaluation model based on DPSIR and BP neural network

作　　者：张力[1,2] 陈文[2,3] 蒋建军[2] 

机构地区：[1]湖南工学院,湖南衡阳421002 [2]南华大学人因研究所,湖南衡阳421001 [3]南华大学环境保护与安全工程学院,湖南衡阳421001

出　　处：《中国安全科学学报》2014年第12期76-82,共7页China Safety Science Journal

摘　　要：为更好地评估安全绩效,使企业充分了解其安全工作系统的整体运行状态,借鉴环境评估中DPSIR模型系统分析的思想,从5个相互联系的层面分类和筛选企业安全方面指标,建立企业安全绩效评估指标体系,其中共含5个1级指标,25个2级指标。采用问卷调查和数据统计分析法对指标进行分级量化,并结合BP神经网络梯度下降法进行模式识别。将所提出的DPSIR+BP神经网络的企业安全绩效评估方法,应用于M企业的安全绩效评估。实例结果表明,M企业安全绩效等级水平为3级,评估结论符合该企业实际情况。To provide enterprises with rules while selecting the safety performance indicators, to help them better understand the establishing process of indicator system and overall operation state of their safety systems, the systematical idea from DPSIR model in environmental evaluation system was used for reference. Indicators were classified and extracted according to five related aspects, and an indicator system was established, which includes 5 first class indicators and 25 second level indicators. Classification and quantification of indicators were carried out by using questionnaire and statistical analysis method. Pattern recognition was carried out by using the gradient descent method of BP neutral network. Then a safety performanee evaluation model based on both DPSIR and BP neural network was built. The model was applied to M company. The result shows that safety performance of the company is at level 3. This evaluation con- forms with reality.

关 键 词：安全绩效 指标体系 DPSIR模型 BP神经网络 MATLAB 

分 类 号：X921[环境科学与工程—安全科学]