摘要
为了研究煤矿中夜班工人的疲劳状况和疲劳检测方法,通过对夜班矿工脑电和行为指标进行分析,建立BP神经网络模型预测矿工疲劳程度。选择15名夜班矿工作为被试,通过脑电图(EEG)测量其夜班前后执行Oddball任务时的脑电信号,通过快速傅里叶变换提取脑电特征值(α+θ)/β、(α+θ)/(α+β)、θ/β和α/β,收集认知任务期间的准确性和反应时间,作为疲劳检测指标建立BP神经网络模型,对夜班矿工的疲劳等级进行预测。对夜班前后矿工脑电信号中所有通道的特征值进行配对t检验,共得到6个显著通道的特征值,与反应时间、准确率等建立BP神经网络模型,总体检测准确度为89.685%。该模型对检测夜班矿工的疲劳程度具有较高准确性,可为矿工疲劳干预措施和检测装备的研发提供技术手段和理论基础。
In order to study the fatigue status and fatigue detection methods of night shift workers in coal mines,a BP neural network model should be established through the night miners’EEG and behavioral indicators to predict the degree of fatigue miners.Fifteen night-shift miners were selected as subjects in the experiment,their EEG signals,behavioral indicators,including reaction time,correctness rate,and a number of omissions,were recorded before and after the night shift.Besides,the scores of the subjects’Karolinska Sleepiness Scale were collected.The FFT code was written by MATLAB to extract PSD of different frequency bands,and the EEG feature values,(α+θ)/β,(α+θ)/(α+β),θ/βandα/β,were calculated.To analyze the difference in fatigue of night shift miners before and after shift,the channels with significant characteristics were selected in the 60 channels,the behavior data and EEG characteristic values of night shift miners before and after shift were analyzed by paired t-test.The EEG feature values of the six channels F5,F3,FT7,FT8,C5,and CP6 were selected,and the EEG characteristic values of six channels and behavior indicators of all subjects were used as fatigue detection indicators to establish a BP neural network model so that the fatigue levels were predicted.The results show that miners’KSS scale scores increase significantly after the night shift(p<0.001),the reaction time and the number of omissions increase(p<0.001),and the accuracy of miners after shifts decrease markedly.The EEG characteristic values of the six channels show obvious differences before and after the shift(p<0.05),indicating that the fatigue level of the miners increases significantly after the night shift.Based on the EEG data and behavioral data of 15 subjects,the overall detection accuracy of the BP neural network model is 89.685%.The model has high accuracy for detecting the fatigue of night shift miners and provides technical means and theoretical basis for the research and development of fatigue intervention measures
作者
田水承
胥静
田方圆
路伟
孙璐瑶
TIAN Shui-cheng;XU Jing;TIAN Fang-yuan;LU Wei;SUN Lu-yao(School of Safety Science and Engineering,Xi'an University of Science and Technology,Xi'an 710054,China;Institute of Safety&Emergency Management,Xi'an University of Science and Technology,Xi'an 710054,China)
出处
《安全与环境学报》
CAS
CSCD
北大核心
2022年第4期2034-2040,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)。
