改进离散Hopfield神经网络在煤矿人因评估中的应用  

Improved discrete Hopfield neuralnetwork for human factors assessmentin coal mines

作　　者：李红霞[1,2] 张倩 田水承 张丹 LI Hongxia;ZHANG Qian;TIAN Shuicheng;ZHANG Dan(School of Management,Xi'an University of Science and Technology,Xi'an 710054,China;School of Safety Science and Engineering,Xi'an University of Science and Technology,Xi'an 710054,China;Shanxi Orchid Kechuang Yuxi Coal Mine Co.,Ltd.,Jincheng 048214,Shanxi,China)

机构地区：[1]西安科技大学管理学院,西安710054 [2]西安科技大学安全科学与工程学院,西安710054 [3]山西兰花科创玉溪煤矿有限责任公司,山西晋城048214

出　　处：《安全与环境学报》2023年第6期1978-1984,共7页Journal of Safety and Environment

基　　金：国家自然科学基金面上项目(51874237);国家自然科学基金重点支持项目(U1904210);国家社科基金项目(20XGL025)。

摘　　要：为了预防和控制煤矿人因事故发生,将离散Hopfield神经网络应用于煤矿人因安全评估中。首先根据人因分析与分类系统(Human Factors Analysis and Classification System,HFACS)模型建立煤矿人因安全风险评估体系,体系包含多个指标;其次是构建输入矩阵,借助模糊综合评价法对评估指标量化编码;最后,运用学习率优化的离散Hopfield神经网络展开煤矿人因安全评估,以模型输出结果确定评估对象风险等级。将构建的煤矿人因安全评估模型与传统评估方法进行比较,结果表明该模型合理有效,可用于煤矿人因安全评估,为煤矿人因安全管理提供依据。In this paper,a coal mine human-caused safety assessment model based on an improved discrete Hopfield neural network is put forward.Firstly,according to the theory of Human Factors Analysis and Classification System(HFACS)model,the human-caused safety accidents in coal mines and their influencing factors are analyzed.This process selects“5M1E”model according to the characteristics of coal mine production which includes six factors:human,machine,material,method,environment,and measurement.On this basis,the influencing factors are divided into five categories:poor organization and management,unsafe supervision,unsafe premise,unsafe behavior,and improper emergency response.Secondly,the index system of coal mine human-caused safety is established by using the risk factors.During the assessment process,to eliminate the irrationality of subjective factors and conclusion error caused by the lack of training samples of neural network models,a discrete Hopfield neural network optimized by learning rate is applied.This improved model uses learn-memory patterns instead of a large number of training samples.Besides,it has high output accuracy.The process uses the fuzzy comprehensive evaluation method to quantify the assessment indexes and construct the input matrix.The risk level of the assessment determines by the output of the model.Finally,to verify the rationality and effectiveness of the coal mine human factors safety assessment model based on the discrete Hopfield neural network,this paper compares and analyzes it with the traditional neural network assessment models(BP neural network and support vector machine).In this case,the results show that this model is reasonable and effective,and can be used for coal mine human factors safety assessment.The assessment result is almost following the actual situation of the coal mine.It provides a scientific basis and reference for human factors safety management in coal mines.

关 键 词：安全社会工程 人因安全 离散HOPFIELD神经网络 模糊综合评价 

分 类 号：X913[环境科学与工程—安全科学]