Scholarlink

<p>The coal mine monitoring dispatchers play a crucial role in the safe operation of coal mines. Fatigue grading detection for coal mine monitoring dispatchers is a key means to effectively prevent accidents. To accurately evaluate the fatigue status of monitoring dispatchers, promptly identify and alleviate their fatigue grades, thereby reducing operational errors and effectively avoiding accidents, this study conducted a coal mine monitoring and dispatching experiment using eye-tracking technology, and collected eye movement indicators from 15 dispatchers, including blink rate, gaze duration, saccade counts, saccade amplitude and pupil size, along with corresponding fatigue assessment indicator data. Subsequently, Principal Component Analysis (PCA) was applied to compute the weights of the fatigue assessment indicators, and the fatigue grades were classified using the K-means ++ algorithm. Based on this, a prediction model for the fatigue grades of coal mine monitoring dispatchers was constructed using an Artificial Neural Network (ANN). The ANN model was compared with seven other machine learning models, including K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Back propagation (BP), Extreme Learning Machine (ELM), Decision Trees (DT), Random Forest (RF) and Linear Discriminant Analysis (LDA). The findings revealed that the optimal classification approach is to categorize fatigue grades into three distinct categories. The ANN model excels in detecting different fatigue grades, achieving an accuracy rate of 93.52%. It can provide instant warnings of dispatchers' fatigue status, effectively curb potential safety hazards and reduce the incidence of accidents caused by personnel fatigue. Meanwhile, it optimizes the scheduling system for dispatchers, providing a scientific basis for decision-making in coal mine safety production.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine Learning Based on Eye Movement Indicators to Detect Fatigue in Coal Mine Monitoring Dispatcher

  • 期刊论文

  • 发表日期：

    2025年4月26日

查看全文
  • Yiman He,
  • Jizu Li

摘要

The coal mine monitoring dispatchers play a crucial role in the safe operation of coal mines. Fatigue grading detection for coal mine monitoring dispatchers is a key means to effectively prevent accidents. To accurately evaluate the fatigue status of monitoring dispatchers, promptly identify and alleviate their fatigue grades, thereby reducing operational errors and effectively avoiding accidents, this study conducted a coal mine monitoring and dispatching experiment using eye-tracking technology, and collected eye movement indicators from 15 dispatchers, including blink rate, gaze duration, saccade counts, saccade amplitude and pupil size, along with corresponding fatigue assessment indicator data. Subsequently, Principal Component Analysis (PCA) was applied to compute the weights of the fatigue assessment indicators, and the fatigue grades were classified using the K-means ++ algorithm. Based on this, a prediction model for the fatigue grades of coal mine monitoring dispatchers was constructed using an Artificial Neural Network (ANN). The ANN model was compared with seven other machine learning models, including K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Back propagation (BP), Extreme Learning Machine (ELM), Decision Trees (DT), Random Forest (RF) and Linear Discriminant Analysis (LDA). The findings revealed that the optimal classification approach is to categorize fatigue grades into three distinct categories. The ANN model excels in detecting different fatigue grades, achieving an accuracy rate of 93.52%. It can provide instant warnings of dispatchers' fatigue status, effectively curb potential safety hazards and reduce the incidence of accidents caused by personnel fatigue. Meanwhile, it optimizes the scheduling system for dispatchers, providing a scientific basis for decision-making in coal mine safety production.