An Integrated Framework for Condition Monitoring and Fault Diagnosis of Electric Machine Drive Systems

Funded by NSF EPCN program (Single PI, 360K, 06/2021-5/2024)

This NSF project aims to design and demonstrate an innovative physics-guided signature-based approach for condition monitoring and fault diagnosis (CMFD) of electric machine networks. As the number of electric machines grows rapidly in response to less carbon dioxide emissions, a large number of electric machines are connected to form electric machine networks. However, traditional CMFD were developed based on individual machine sensors, which requires a large number of sensors and do not comprehensively consider fault and degradation propagation. This limitation will be in part resolved by coordinated monitoring and analysis of strategically-placed electrical waveform sensors in power networks. The intellectual merits of the project include integrating high-fidelity physical model of electric machine networks to signature-based CMFD for improved accuracy and robustness. The broader impacts of the project include advancing the research experiences for K-12 and undergraduate students including underrepresented students. The research will be integrated into the undergraduate and graduate electric power engineering curriculum to educate future engineers who will have the skills and knowledge to meet the emerging needs of the industry.

The proposed physics-guided signature-based approach for condition monitoring and fault diagnosis (CMFD) of electric machine networks will advance the literature in a new direction compared to the traditional approach based on individual machine sensors. Four specific objectives will be pursued: (1) Assess the electrical waveform signature due to faults and degradation that will build a technical foundation for the proposed CMFD approach. (2) Create a high-fidelity physical model of electric machine networks that reflects not only the fault, degradation and nonlinearity of the machine, but also waveform propagation due to faults and degradation. (3) Design a physics-guided signature-based method that leverages the waveform propagation model to more effectively and efficiently identify and locate faults or degradation source in electric machine networks. (4) Build a semi-virtual testbed of electric machine networks to evaluate the performance of the proposed CMFD approach. The proposed CMFD approach can be generally used in a variety of applications, including manufacturing and industrial systems, smart buildings, wind farms, and electrified transportation.