Innovative Machine Learning Method for Nuclear Power Plant Fault Diagnosis

Researchers at Sanming University have developed a novel machine learning method to improve the reliability of fault diagnosis in nuclear power plant gear lubrication systems. This integrated data-driven approach combines sensor measurements with advanced machine learning algorithms to detect faults more accurately.

Key Takeaways:

• The proposed method, called Integrated Data-Driven Machine Learning (IDDML), uses sensor measurements to identify critical fault paths and calculate failure probabilities.
• The IDDML method consists of two major components: fault tree analysis and adaptive sparse principal component analysis based on variable projection combined with proximal gradient optimization (VPPGO-ASPCA).
• The VPPGO-ASPCA method incorporates a modified principal component analysis technique and an optimization algorithm with an adaptive threshold, improving detection accuracy and sensitivity.
• Compared to traditional diagnostic methods, IDDML offers higher detection accuracy and improved sensitivity.
• The method was validated through experimental and computational results, achieving a fault detection success rate of up to 99%.
• A unique real-time measurement system integrating multiple high-sensitivity sensors and four network architectures was developed for the fault diagnosis of the gear lubrication system in a nuclear power plant.

Statistics:

• The IDDML fault diagnosis method achieves a fault detection success rate of up to 99%.
• The VPPGO-ASPCA method incorporates an adaptive threshold, improving detection accuracy and sensitivity.
• The method was validated through experimental and computational results, demonstrating the effectiveness of IDDML in real-time measurement systems.
• The research was funded by the Program for Key Science and Technology Project of Industry-University-Research Collaborative Innovation in Sanming City, China, and other organizations.

Sources:

• "Fault Diagnosis of Gear Lubrication Systems Using Sensor Measurements and Data-driven Machine Learning: a Case Study of a Nuclear Power Plant." Sensors and Materials, 2025;37(6):2325-2349.
• Sanming University Research Program, China.
• Major Project of the Fujian STS Program Supporting the Cooperation Between the Academy and the Province, Sanming University of Fujian Province, China.
• Department of Education of Fujian Province, China.