基于RF-PSO的塔吊事故可能发生阶段预测与分析  

Prediction and analysis of possible stages of tower crane accidents based on RF-PSO

作　　者：刘冬华 赵星[1] 赵江平[1] 杨震[1] LIU Donghua;ZHAO Xing;ZHAO Jiangping;YANG Zhen(School of Resource Engineering,Xi'an University of Architecture and Technology,Xi'an Shaanxi 710055,China)

机构地区：[1]西安建筑科技大学资源工程学院,陕西西安710055

出　　处：《工业安全与环保》2024年第2期59-63,67,共6页Industrial Safety and Environmental Protection

基　　金：陕西省软科学研究计划项目(2019KRM091)。

摘　　要：塔吊在高层建筑的施工现场较为常见,为了探索塔吊事故的影响因素,对塔吊事故可能发生阶段进行预测,建立了塔吊事故的人为因素分析分类系统(HFACS)模型框架,明确了塔吊事故的影响因素;利用卡方检验对事故致因进行特征选择及量化分析,提出了一种基于粒子群优化算法优化随机森林预测模型(RF-PSO)。结果表明:RF-PSO模型的预测是可靠的,训练后的RF-PSO模型可以预测塔吊事故可能发生的阶段,影响因素塔吊安装不符合设计要求、缺乏沟通和个人防护设备缺失发生的概率越大,塔吊事故分别在安拆阶段、吊装阶段和攀爬阶段发生事故的可能性越大。研究结果可为塔吊施工现场管理提供理论依据。Tower cranes are common in the construction site of high-rise buildings.In order to explore the influencing factors of tower crane accidents and predict the possible stages of tower crane accidents,the human factor analysis and classification system(HFACS)model framework of tower crane accidents is established,and the influencing factors of tower crane accidents are clarified.The chi-square test is used for feature selection and quantitative analysis of accident causes,and a stochastic forest prediction model based on particle swarm optimization algorithm(RF-PSO)is proposed.The results show that the prediction of RF-PSO model is reliable,and the trained RF-PSO model can predict the possible stages of tower crane accidents.The greater the probability of tower crane installation not meeting the design requirements,lack of communication and lack of personal protective equipment,the greater the probability of tower crane accidents in the installation and disassembly stage,lifting stage and climbing stage.The research results can provide theoretical basis for tower crane construction site management.

关 键 词：塔吊事故 随机森林 粒子群优化算法 事故预测 

分 类 号：TU712.4[建筑科学—建筑技术科学]