基于机器学习的建筑施工人员安全能力预测模型

doi:10.16265/j.cnki.issn1003-3033.2023.07.1630

摘要/Abstract

摘要：

为探究建筑施工人员安全能力的作用机制和关键影响因素,以社会资本和行为动机理论为依据,建立基于组织网络和个体行为的多指标数据集;借鉴成熟量表构建安全能力指标体系,通过问卷调查向中国境内若干建筑工程项目的457名施工人员获取数据;采用基于熵权法与变异系数法的多指标综合评价法进行指标权重计算和筛选;采用梯度提升决策树(GBDT)等5种机器学习方法分别建立回归预测模型,利用网格搜索法进行超参数优化,以提升模型的预测能力。结果表明:集成学习方法优于传统机器学习方法,GBDT算法在所有方法中拟合性能最佳;权重优化后的数据集可在一定程度上提升模型的预测精度;施工人员的安全参与行为和组织中社会资本的结构与认知维度对其安全能力影响较大;此外,个体特征中的工作年限和教育水平也与安全能力存在一定相关性。

关键词: 机器学习, 建筑施工人员, 安全能力, 预测模型, 梯度提升决策树(GBDT)

Abstract:

In order to explore the mechanism and key influencing factors of the safety competency of construction workers, this study established a multi-indicator dataset based on organizational network and individual behavior, referring to social capital and behavioral motivation theory. A safety competency indicator system was constructed based on previous scales, and data were obtained from 457 construction workers in China through a questionnaire survey. The multi-indicator comprehensive evaluation method based on the entropy weight and coefficient of variation method was used to calculate and screen the indicators. Then, five machine learning methods, such as GBDT, were used to establish regression models, respectively. The grid search method was used for hyperparameter optimization to improve the prediction ability of the model. The results show that the ensemble learning method is superior to the traditional machine learning method, and the GBDT algorithm has the best fitting performance. The optimized dataset can improve the prediction accuracy of the model. The safety participation and the structural and cognitive dimension of social capital significantly affect the construction workers' safety competency. In addition, their working experience and education level also have a correlation with their safety competency.

Key words: machine learning, construction workers, safety competency, prediction model, gradient boosting decision tree (GBDT)

赵伟, 李书全. 基于机器学习的建筑施工人员安全能力预测模型[J]. 中国安全科学学报, 2023, 33(7): 51-57.

ZHAO Wei, LI Shuquan. Prediction model of safety competency of construction workers based on machine learning[J]. China Safety Science Journal, 2023, 33(7): 51-57.

图/表 6

表1

表2

表3

图1

图2

图3

参考文献 19

[1]	住房和城乡建设部. 2020年房屋市政工程生产安全事故情况[EB/OL]. (2022-10-27). https://www.mohurd.gov.cn/gongkai/zhengce/zhengcefilelib/202210/20221026_768565.html.
[2]	Health and Safety Executive. Managing health and safety in construction[EB/OL]. (2021-08-16). https://www.hse.gov.uk/index.htm.
[3]	LUO Xiaochun, LI Heng, HUANG Ting, et al. A field experiment of workers' responses to proximity warnings of static safety hazards on construction sites[J]. Safety Science, 2016, 84:216-224. doi: 10.1016/j.ssci.2015.12.026
[4]	HASLAM R A, HIDE S A. Contributing factors in construction accidents[J]. Applied Ergonomics, 2005, 36(4):401-415. doi: 10.1016/j.apergo.2004.12.002 pmid: 15892935
[5]	成连华, 赵旭东, 郝杰. 集成ISM与G1的建筑工人安全能力影响因素研究[J]. 中国安全科学学报, 2023, 33(4):61-68. doi: 10.16265/j.cnki.issn1003-3033.2023.04.0962
	CHENG Lianhua, ZHAO Xudong, HAO Jie. Research on influencing factors of construction workers' safety competency integrating ISM and G1 method[J]. China Safety Science Journal, 2023, 33(4):61-68. doi: 10.16265/j.cnki.issn1003-3033.2023.04.0962
[6]	杨鑫刚, 王起全. 建筑安全管理机制博弈分析与改进[J]. 中国安全科学学报, 2021, 31(11):26-31. doi: 10.16265/j.cnki.issn 1003-3033.2021.11.004
	YANG Xingang, WANG Qiquan. Game analysis and improvement of construction safety management mechanism[J]. China Safety Science Journal, 2021, 31(11):26-31. doi: 10.16265/j.cnki.issn 1003-3033.2021.11.004
[7]	NYKNEN M, SUND R, VUORI J. Enhancing safety competencies of young adults: a randomized field trial (RCT)[J]. Journal of Safety Research, 2018, 67:45-56. doi: S0022-4375(17)30723-5 pmid: 30553429
[8]	SUN Linhui, ZHANG Yimeng, CUI Kai, et al. Construction of safety competency evaluation system for high-risk industry[C]. International Conference on Enterprise Systems (ES), 2017: 46-53.
[9]	SULLIVAN D T, HIRST D, CRONENWETT L. Assessing quality and safety competencies of graduating prelicensure nursing students[J]. Nursing Outlook, 2015, 57(6):323-331. doi: 10.1016/j.outlook.2009.08.004
[10]	阳长征. 突发事件中网络社会资本对受众认知偏差的影响研究:基于信息级联的中介效应视角[J]. 情报杂志, 2021, 40(2):123-130.
	YANG Changzheng. Research of effect of social capital on cognitive bias in network emergency[J]. Journal of Intelligence, 2021, 40(2):123-130.
[11]	NAHAPIET J, GHOSHAL S. Social capital, intellectual capital, and the organizational advantage[J]. Academy of Management Review, 1998, 23(2):242-266. doi: 10.2307/259373
[12]	CURCURUTO M, CONCHIE S M, MARIANI M G. The role of prosocial and proactive safety behaviors in predicting safety performance[J]. Safety Science, 2015, 80:317-323. doi: 10.1016/j.ssci.2015.07.032
[13]	NEAL A, GRIFFIN M A. The impact of organizational climate on safety climate and individual behavior[J]. Safety Science, 2000, 34(1): 99-109. doi: 10.1016/S0925-7535(00)00008-4
[14]	王亮, 孙一伟, 王凯. 安全工程人才多维能力培养体系构建与实践[J]. 中国安全科学学报, 2021, 31(5):52-57. doi: 10.16265/j.cnki.issn1003-3033.2021.05.008
	WANG Liang, SUN Yiwei, WANG Kai. Construction and practice of multi-dimensional ability training system for safety engineering talents[J]. China Safety Science Journal, 2021, 31(5):52-57. doi: 10.16265/j.cnki.issn1003-3033.2021.05.008
[15]	MAITI J. Development of worksystem safety capability index (WSCI)[J]. Safety Science, 2010, 48(10):1369-1379. doi: 10.1016/j.ssci.2010.05.010
[16]	胡喆, 胡昊, 徐峰, 等. 基于认知方式的建筑工人安全行为能力仿真研究[J]. 安全与环境学报, 2022, 22(1):194-200.
	HU Zhe, HU Hao, XU Feng, et al. Research on the safety behavior capability of construction workers based on cognitive style[J]. Journal of Safety and Environment, 2022, 22(1):194-200.
[17]	ZHOU Zhipeng, GUO Wenya. Applications of item response theory to measuring the safety response competency of workers in subway construction projects[J]. Safety Science, 2020, 127: 114-126.
[18]	王盼盼, 李启明, 邓小鹏. 施工人员安全能力模型研究[J]. 中国安全科学学报, 2009, 19(8): 40-45.
	WANG Panpan, LI Qiming, DENG Xiaopeng. Study on safety competency model of construction workers[J]. China Safety Science Journal, 2009, 19(8): 40-45.
[19]	LI Shuquan, FAN Meng. Effect of social capital between construction supervisors and workers on workers' safety behavior[J]. Journal of Construction Engineering & Management, 2018, 144(4):79-88.

指标维度	题项	权重
结构维度	SCS1:与项目部其他成员间的私人关系	0.043 4
	SCS2:对项目部内其他成员了解的程度	0.040 4
	SCS3:在食堂、休息室、走廊等非正式场合与项目部其他成员交谈的次数	0.037 8
	SCS4:与项目部其他成员在工作之外熟悉的程度	0.037 2
	SCS5:与项目部内其他成员间的合作	0.035 1
关系维度	SCR1:在与同事合作的过程中能够倾尽自己所能来完成某一项工作	0.040 6
	SCR2:与项目部其他成员能真诚合作	0.039 5
	SCR3:在与同事进行合作的过程中彼此不会投机取巧	0.037 9
认知维度	SCC1:与项目部其他成员拥有一致的集体目标	0.041 1
	SCC2:针对工作中的问题使用的交流方式是大家都能接受和理解的	0.039 3
	SCC3:与项目部其他成员交流时使用专业术语的次数	0.039 0
	SCC4:与项目部其他成员有共同语言并能有效沟通	0.037 9
	SCC5:对工作中的专业符号、用语、词义都很清楚	0.035 5
安全遵守	SBC1:能够积极地配合安全管理人员的指挥和安排	0.033 1
安全遵守	SBC2:工作中,您遵守安全相关规定及操作规程	0.033 0
安全参与	SBP1:参与制定组织的安全目标、安全计划等工作	0.044 7
	SBP2:参与项目安全风险评价等工作	0.038 6
	SBP3:主动与工友讨论施工的安全问题	0.036 9
	SBP4:主动地制止、纠正同事的错误操作或想法	0.036 8
	SBP5:主动与上级领导沟通施工安全问题	0.036 6
	SBP6:主动积极地参加安全会议	0.036 5
统计信息	工作年限:1(≤10年)、2(>10年且≤15年)、3(>15年且≤20年)、4(>20年)	—
	年龄:1(18~30)、2(31~40)、3(41~50)、4(51~60)岁	—
	教育水平:小学、初中、高中、大专及以上	—
	工程类型:民用建筑工程、工业建筑工程、市政公用行业建筑项目、其他	—

模型	参数	取值
Lasso回归	L1项常数	0.01
SVR	惩罚系数	10.2
SVR	核函数	径向基核函数
MLP	权重优化求解器	lbfgs优化器
	隐藏层神经元数量	120
	学习率	自适应学习
	惩罚系数	1
RF	基评估器数量	190
	不纯度衡量	MSE
	最大深度	11
GBDT	弱学习器数量	200
	最大深度	3
	学习率	0.02
	子采样	0.9

模型	原模型			优化模型
模型	MAE	R²	时间/s	MAE	R²	时间/s
Lasso	0.470	0.522	4.8	0.390	0.606	4.6
SVR	0.329	0.732	9.3	0.305	0.769	8.9
MLP	0.326	0.726	27.1	0.276	0.802	22.4
RF	0.263	0.825	50.7	0.250	0.843	46.1
GBDT	0.203	0.903	60.5	0.192	0.907	52.6