基于多机器学习算法模型对煤矿VDT作业视觉疲劳预测

Visual Fatigue Prediction for VDT Operations in Coal Mines Based on Multiple Machine Learning Algorithm Models

准确预测智能化煤矿中视觉显示终端(VDT)作业员的视觉疲劳程度,是降低VDT作业失误、确保智能化煤矿安全的关键。通过煤矿VDT作业试验,收集被试者眼动指标数据和主客观视觉疲劳评价指标值,采用主成分分析法(PCA)计算各评价指标的权重,并利用FCM聚类算法计算最佳视觉疲劳分类数。在此基础上,将VDT作业视觉疲劳程度阈值作为输出信息,被试者的眼动数据作为输入信息,构建了一个综合了5种机器学习算法的视觉疲劳程度综合预测模型。结果表明:最佳视觉疲劳分类数为3,所构建的预测模型的准确率为94.4%,对3类视觉疲劳程度预测的平均误差率为3.7%,相较于5种单个机器学习算法,综合模型的预测准确性和稳定性均明显提升。

Accurately predicting the visual fatigue degree of visual display terminal(VDT)operators in intelligent coal mines is the key to reduce VDT operation errors and ensure the safety of intelligent coal mines.Through the coal mine VDT operation experiment,the eye-tracking index data and subjective and objective visual fatigue evaluation index values of the subjects were collected.The weight of each evaluation index was calculated by principal component analysis(PCA),and the optimal visual fatigue classification number was calculated by FCM clustering algorithm.On this basis,the visual fatigue threshold of VDT operation was taken as the output information,and the eye-tracking data of the subject was taken as the input information.A comprehensive prediction model of visual fatigue degree was constructed,which integrated five machine learning algorithms.The results show that the optimal classification number of visual fatigue is 3,the accuracy of the constructed prediction model is 94.4%,and the average error rate of the prediction of the three types of visual fatigue is 3.7%.Compared with five single machine learning algorithms,the prediction accuracy and stability of the comprehensive model have been significantly improved.