Human-Centric Digital Twin Model of Electric System Maintenance Technicians for Near Real-Time Health and Ergonomic Postural Risk Monitoring
Restricted (Penn State Only)
- Author:
- Sadat Mohammadi, Milad
- Graduate Program:
- Architectural Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- June 10, 2022
- Committee Members:
- Somayeh Asadi, Professor in Charge/Director of Graduate Studies
Somayeh Asadi, Co-Chair & Dissertation Advisor
John Messner, Major Field Member
Houtan Jebelli, Co-Chair & Dissertation Advisor
Mehdi Kiani, Outside Unit & Field Member - Keywords:
- Human Centric Digital Twin Model
Health Monitoring
Machine Learning
Deep Learning
Ergonomic Risk Assessment - Abstract:
- Increasing attention has been paid to energy management on the generation, transmission, and demand sides in recent decades. Novel control techniques have been introduced in the literature to ensure a reliable energy supply and prevent energy shortages on the demand side. However, there is a lack of studies on reducing human-related failures in energy systems. Operators and maintenance crew are a vital part of energy systems, and their efficient and effective performance is essential to the reliable operation of the energy system. Early detection of the most frequently occurring human factors contributing to human-related failures in energy systems is a promising step toward the reliable operation of electric systems. Human factors such as occupational stress, high physical load, and ergonomic risks affect operators' and maintenance crew's ability to assess the situation and take appropriate actions, consequently increasing the potential of human-related failure in the energy system. Thus, monitoring and detecting the mentioned human factors among the operators and maintenance technicians is crucial to ensure their reliable performance. In this research, a human-centric digital twin (DT) model is introduced to monitor the health status of operators and maintenance crew in a near real-time setting and detect the human factors prior to impacting their performance. The proposed human-centric DT model in this research consists of three layers: (1) physical layer, (2) digital layer, and (3) communication layer. The physical layer consists of wearable sensors to record physiological signals and body posture; the recorded signals are transferred to the digital layer through the communication layer. The digital layer hosts a virtual model for visualization and multiple deep learning/machine learning models for health risk assessments. The optimal configuration for each layer is introduced to achieve a computationally less expensive and near real-time DT system. The proposed human-centric DT model can be used to visualize and detect occupational stress, high physical load, and ergonomic risks among workers in electric systems.