Machine-Learning-Based Fatigue Trend Analysis on IMU Wearable Sensor Data from Construction Site Workers.

Area of Science:

• Occupational Health and Safety
• Biomedical Engineering
• Wearable Technology

Background:

• Physical fatigue is a significant risk factor for accidents and injuries in the construction industry.
• Wearable inertial measurement units (IMUs) show promise for fatigue detection, but research is limited to laboratory settings.
• A gap exists in applying IMU-based fatigue detection to real-life work environments.

Purpose of the Study:

• To bridge the gap between laboratory findings and real-world application of IMU-based fatigue detection.
• To investigate fatigue trend detection using IMU data from actual construction workers on a construction site.
• To enhance fatigue detection by incorporating frequency domain analysis and machine learning.

Main Methods:

• Utilized wearable IMU sensors to collect data from construction workers during simulated tasks on an actual construction site.
• Employed frequency domain investigations to extract detailed, fatigue-relevant features from sensor data.
• Applied machine learning algorithms to predict fatigue trends based on the collected IMU data.

Main Results:

• Achieved state-of-the-art accuracy in fatigue trend detection using IMU data from a real-world construction setting.
• Identified key sensor locations and features crucial for effective fatigue monitoring.
• Demonstrated the feasibility of IMU-based fatigue detection in practical work environments.

Conclusions:

• IMU technology can be effectively applied for fatigue detection in real-life construction work settings.
• Frequency domain analysis and machine learning significantly improve the accuracy of fatigue trend prediction.
• This research provides valuable insights for developing advanced safety measures to mitigate fatigue-related risks in construction.