中国安全生产科学技术
中國安全生產科學技術
중국안전생산과학기술
JOURNAL OF SAFETY SCIENCE AND TECHNOLOGY
2014年
12期
185-191
,共7页
李书全%吴秀宇%袁小妹%周远
李書全%吳秀宇%袁小妹%週遠
리서전%오수우%원소매%주원
安全行为%影响因素%决策模型%遗传算法( GA)%支持向量机( SVM)
安全行為%影響因素%決策模型%遺傳算法( GA)%支持嚮量機( SVM)
안전행위%영향인소%결책모형%유전산법( GA)%지지향량궤( SVM)
safety behaviors%influencing factor%decision model%genetic algorithm ( GA)%support vector machine ( SVM)
导致施工人员不安全行为的因素众多,如何保证员工进行安全施工是施工企业亟待解决的问题。为分析施工人员安全行为的影响因素及作用机理,从社会资本理论、认知心理学理论和安全行为理论分析了安全行为的影响因素,并设计了相关调查问卷。在进行仿真分析前,采用遗传算法优化计算的方法筛选出了13个关键影响因素,降低了自变量之间的相关性,之后利用支持向量机( SVM)的方法对决策模型进行了仿真分析,并与BP神经网络模型做了对比。仿真结果表明:基于筛选出的关键影响因素的SVM仿真模型的精度和有效性大于BP神经网络模型,模型精度为0.00887,相关系数为88.4%,说明影响因素与安全行为之间具有较好的拟合关系。研究结论为企业衡量员工安全行为水平,提高员工安全行为能力和企业安全管理能力提供理论支持。
導緻施工人員不安全行為的因素衆多,如何保證員工進行安全施工是施工企業亟待解決的問題。為分析施工人員安全行為的影響因素及作用機理,從社會資本理論、認知心理學理論和安全行為理論分析瞭安全行為的影響因素,併設計瞭相關調查問捲。在進行倣真分析前,採用遺傳算法優化計算的方法篩選齣瞭13箇關鍵影響因素,降低瞭自變量之間的相關性,之後利用支持嚮量機( SVM)的方法對決策模型進行瞭倣真分析,併與BP神經網絡模型做瞭對比。倣真結果錶明:基于篩選齣的關鍵影響因素的SVM倣真模型的精度和有效性大于BP神經網絡模型,模型精度為0.00887,相關繫數為88.4%,說明影響因素與安全行為之間具有較好的擬閤關繫。研究結論為企業衡量員工安全行為水平,提高員工安全行為能力和企業安全管理能力提供理論支持。
도치시공인원불안전행위적인소음다,여하보증원공진행안전시공시시공기업극대해결적문제。위분석시공인원안전행위적영향인소급작용궤리,종사회자본이론、인지심이학이론화안전행위이론분석료안전행위적영향인소,병설계료상관조사문권。재진행방진분석전,채용유전산법우화계산적방법사선출료13개관건영향인소,강저료자변량지간적상관성,지후이용지지향량궤( SVM)적방법대결책모형진행료방진분석,병여BP신경망락모형주료대비。방진결과표명:기우사선출적관건영향인소적SVM방진모형적정도화유효성대우BP신경망락모형,모형정도위0.00887,상관계수위88.4%,설명영향인소여안전행위지간구유교호적의합관계。연구결론위기업형량원공안전행위수평,제고원공안전행위능력화기업안전관리능력제공이론지지。
There are many factors leading to the unsafe behaviors of construction personnel, and how to ensure the staff to conduct safety construction is an urgent problem in construction enterprises.To analysis the influencing fac-tors and action mechanism of safety behaviors for construction personnel, combining with the theories of social cap-ital, cognitive psychology and safety behavior, the influencing factors of safety behaviors were analyzed, and the corresponding questionnaire was designed.Before the simulation, the genetic algorithm was applied to screen out 13 key influencing factors, and the correlation between independent variables was reduced.The simulation analysis on decision model was conducted by SVM, and a comparison with BP neural network model was presented.The simu-lation results showed that based on the selected key influencing factors, the accuracy of the SVM model was greater than that of BP neural network, the model precision was 0.00887, and the correlation coefficient was 88.4%, and there had a better fitting relationship between influencing factors and safety behaviors.The results can provide theo-retical support for enterprise to measure the safety behaviors level of employee, and improve the safety behaviors a-bility of employee and safety management ability of enterprise.