Contactor Fault Detection and Classification System Using Optical Fiber Bragg Grating Sensors

Electrical switching devices control and protect systems at various voltages. Monitoring them ensures safety and reliability. This study introduces a method to instrument and analyze these devices, using ABB AX40 AC contactors and Fiber Bragg Grating (FBG) sensors. The dynamic strain sensing of the FBG was used for acquiring signals for the analysis of the switching event. The devices were subjected to three simulated fault conditions: the inner contact blockage, pressure spring wear-off, and load contact wear-off. For recognizing the degradation patterns of the mechanisms, the data acquired during the switching events were submitted to several steps, such as data augmentation, feature selection, and classification. With a Support Vector Machine as the classifier, a score of 80% for fault detection in training and validation was achieved. Within this detection, a score of 80.2% for fault classification was achieved. Regarding the repeatability test data set, it was able to achieve results of fault detection of 72.1% and within this detection, a score of 85% for fault classification was achieved. We also used both, the CN2 Rule classifier and the Decision Tree classifier, to extract human-comprehensible information from the frequency spectrum features. The results presented in this paper suggest the suitability of FBG and machine learning methods for the predictive maintenance of switching devices and the importance of repeatability for future field applications.