中国安全科学学报 ›› 2018, Vol. 28 ›› Issue (4): 109-114.doi: 10.16265/j.cnki.issn1003-3033.2018.04.019

• 安全工程技术科学 • 上一篇    下一篇

基于RVM的装配式建筑吊装作业安全预警模型

刘名强1, 李英攀**1 副教授, 王芳2 工程师, 陈晓3 工程师, 李瑞格1, 李晓喆1   

  1. 1 武汉理工大学 土木工程与建筑学院,湖北 武汉 430070
    2 中建三局第二建设工程有限责任公司, 湖北 武汉 430074
    3 中建三局绿色产业投资有限公司,湖北 武汉 430040
  • 收稿日期:2018-01-28 修回日期:2018-03-17 出版日期:2018-04-28 发布日期:2020-09-28
  • 通讯作者: 李英攀(1978—),女,江西上饶人,博士,副教授,硕士生导师,从事建设项目管理及其信息化、建筑产业化、项目投融资管理与PPP模式等方面的研究。E-mail:liyingpan@whut.edu.cn。
  • 作者简介:刘名强(1994—),男,江西萍乡人,硕士研究生,研究方向为建设项目管理及其信息化、建筑产业化、PPP模式等。E-mail:liuyangmai@whut.edu.cn。
  • 基金资助:
    湖北省自然科学基金资助(2013CFB346);中央高校基本科研业务费专项资金资助(WUT:2014-IV-122)。

An RVM based safety early warning model for hoisting operation in fabricated building project

LIU Mingqiang1, LI Yingpan1, WANG Fang2, CHEN Xiao3, LI Ruige1, LI Xiaozhe1   

  1. 1 School of Civil Engineering and Architecture, Wuhan University of Technology, Wuhan Hubei 430070, China
    2 Second Construction Co., Ltd. of China Construction Third Engineering Bureau, Wuhan Hubei 430074, China
    3 Green Industry Investment Co., Ltd. of China Construction Third Engineering Bureau, Wuhan Hubei 430040, China
  • Received:2018-01-28 Revised:2018-03-17 Online:2018-04-28 Published:2020-09-28

摘要: 为提高装配式建筑施工安全水平,准确判断吊装作业安全状况,建立基于相关向量机(RVM)的预警模型。根据装配式建筑吊装作业特点,对比传统施工模式,分析致使吊装事故发生的主要因素,按人-机-料-法-环(4M1E)5要素确定预警指标体系,并通过粗糙集(RS)属性约简算法确定模型安全预警因子;选用混合核函数构建RVM预警模型,并通过改进粒子群算法(IPSO)寻优确定核参数,给出计算方法及模型流程;以华中地区5个项目的相关数据完成模型学习训练和预警仿真。结果表明:用该模型所得结果与实际情况基本一致,判定正确率为94%、平均相对误差为3.667%,预警分析效果良好,较其他3种机器学习方法泛化拟合能力更强、效率更高。

关键词: 装配式建筑, 塔机吊装作业, 相关向量机(RVM), 安全预警模型, 改进粒子群算法(IPSO)

Abstract: To improve the safety level of construction and get accurate prediction of the safe operation condition of the hoisting operation for fabricated building project, a pre-warning model based on RVM was developed. According to the features of the hoisting operation for fabricated building project, major factors causing frequent accidents were discussed in comparison with the traditional mode, and a early warning index system was established according to 4M1E. The early warning factor was ascertained by attribute reduction algorithm in rough set(RS). A mixed kernel function was introduced and an RVM model was built, whose kernel parameters were determined by IPSO optimization. The model was applied to 5 projects as an example. The data on these projects were input for training and simulation. As the case study shows, the results obtained by using the model are basically consistent with the actual situation.The fitting accuracy, generalization ability and efficiency of the model are better than those of the three other machine learning algorithms.

Key words: fabricated building, crane hoisting operation, relevance vector machine (RVM), safety early warning model, improved particle swarm optimization (IPSO)

中图分类号: