Host: The Japan Society of Mechanical Engineers
Name : The 30th International Conference on Nuclear Engineering (ICONE30)
Date : May 21, 2023 - May 26, 2023
In nuclear power plants, high reliability nuclear-grade motors is required drive critical equipment. In order to avoid serious accidents on rotating machinery, accurate rolling bearing fault detection index plays an important role in detecting incipient faults. This paper proposed a new non-manual index which can accurately detect incipient bearing fault based on vibration analysis. The detection method consists of three stages. The first stage is obtaining the residuals between the measured data and the reconstructed data. For obtaining the residuals, Auto-Associate Kernel Regression (AAKR) is adopted to reconstruct the current state based on normal state vibration signals. In the second stage, Sequential Probability Ratio Test (SPRT) is adopted to determine the instantaneous state of the bearing. A new non-manual method, which is named as Regression Distance Index (RDI), is proposed as the third stage to accurately locate incipient bearing failures. The results of experimental verification on the bearing database of the Center for Intelligent Maintenance Systems (IMS), University of Cincinnati show the reliability, accuracy and versatility of the index. It shows following advantages: (a) the index can detect incipient bearing fault accurately; (b) the index is self-adaptive without manual setting; (c) the index is robust for inner race, outer race and roller elements fault detection, as the results will not impact by the quality of training data for not using empirical models.
