基于物元可拓和神经网络的危化品港口危险等级评价模型-《集美大学学报（自然科学版）》

文章信息/Info

Title:: Evaluation Model on Hazard Categories in Dangerous Chemical Port Based on Object Topology and BP Neural Network

Author(s):: LIU Cuilian; WANG Jie; School of Traffic and Transportation Engineering,Dalian Maritime University,Dalian 116000，China

Keywords:: port hazard category evaluation; Wuli-Shili-Renli(WSR); BP neural network; G1 method; matter-element extension theory

摘要:: 为了准确且快速地评判危化品港口危险性程度，减少港口发生风险事故的几率，提出一种可对港口危险等级快速分类的评价模型。根据物理-事理-人理(WSR)方法论构建危险等级评价体系,用物元可拓理论建立模型确定各危险等级的节域和经典域，结合指标权重计算综合关联度，得出国内10个主要危化品港口2018年和2020年的危险性等级，并将其作为数据样本，随机分为训练集和测试集进行BP神经网络训练。结果表明：各危化品港口危险等级的评价结果与实际情况相符，神经网络快速评价模型的输出结果与实际危险等级基本一致。可使用该模型对危化品港口的危险等级进行快速评价，避免人为因素带来的随机误差。

Abstract:: In order to accurately and quickly detemine the hazard categories in dangerous chemical ports and reduce the chance of risky accidents in ports,an evaluation model for fast classification is proposed.According to the WSR methodology,the evaluation system of port hazard level is constructed,and the sectional and classical domains of each hazard level are determined by establishing the material element extension theory model,and the hazard levels of ten major domestic dangerous chemical ports in 2018 and 2020 are derived and used as data samples,which are randomly divided into training and testing sets for BP neural network training.The results show that the selected indicators can comprehensively reflect the hazard levels of the ports,and the evaluation model results after the neural network training are largely consistent with the actual levels.The model can be used to quickly evaluate the hazard level of a port,which can better avoid random errors brought by human factors.