Identification of Series Fault Arc Occurred in Motor with Inverter Circuits Under Vibration Conditions

One of the safety risks in electricity supply systems is the series fault arc (SAF). The method for identifying SAFs in electrical connectors for industrial motor with inverter circuits under vibrational situations was suggested in this study. Firstly, we adopted a motor with an inverter as a load, and a SAF experiment was conducted under vibrational conditions. To eliminate high-frequency harmonic interference from the inverter, the fault-phase current signal at the back-end of the inverter collected by the experiment was filtered with a finite impulse response low-pass filtering. After that, the filtered current signal was processed by phase space reconstruction (PSR), and the obtained signal was normalized, visualized and gray scaled. Next, the gray level co-occurrence matrix (GLCM) of the processed signal in the 0°, 45°, 90° and 135° directions were calculated, and 24 matrix parameters were selected as the preliminary characteristics of the SAF. Finally, the random forest (RF) algorithm was used to screen out 13 effective features and completed the training of the recognition model. The test results show that the identification method can not only distinguish between the normal data and the SAF data with 100% accuracy, but also classify the normal data and the SAF data with 99.25% accuracy under three different vibration frequencies.