Arc Duration Measurement of Vacuum Interrupter Based on Multi-sensor Fusion

With an increasing share of distributed generation, vacuum interrupters see more switching operations, which makes them more susceptible to contact wear and reduced life expectancy. Past studies have shown the correlation between arc duration and loss of contact mass during load switching. Contact wear estimation for condition-based maintenance can be based on measuring the accumulative arc duration. However, one problem that remains unresolved is how to reliably measure arc duration. Existing methods are associated with implementation challenges, limited accuracy, and inadequate reliability. To achieve an accurate and robust arc duration measurement, this paper proposes a non-invasive online arc duration measurement method based on multi-sensor fusion. Multiple sensors are set in different locations to capture the very-low-frequency and low-frequency magnetic field and electric field, vibration, and acoustic signals during the load-switching event of a vacuum interrupter. Various sensor fusion algorithms, including 1D Convolutional Neural Networks, are used for data fusion and for determining the arc duration. The results demonstrate that the proposed method of multi-sensor fusion can effectively determine arc duration with a low error rate of less than 0.2ms, and the method is robust to noise interference and sensor failure.