IoT-Aimed Health Monitoring Scheme for Power Semiconductor Devices Based on Mahalanobis Processing

Health condition monitoring is critical to secure the availability and reliability of the power semiconductor devices in power converter. Massive efforts have been devoted to the development of condition monitoring methods in the past decade. However, most of them are limited to simplifying laboratory testing conditions, without considering the dynamic operational and environmental conditions in real applications, or field implementation aspects such as the inconvenient terminal connecting of real converter products. To facilitate the implementation of condition monitoring in field applications, first, this article creatively proposes a bipolar self-power circuit. It can provide the power supply to the chips in the health indicator sensing unit, including the isolation chip, operational amplifier, and analog-to-digital (AD) converter, enabling a plug-and-play edge condition monitoring module. Second, an effective data processing procedure is proposed to suppress the effects caused by the dynamic conditions of the converter, avoiding the complex data calibration conducted in conventional studies. Then, based on the creations, an Internet of Things (IoT)-aimed systematic health monitoring scheme is presented, consisting of the edge health indicator monitoring unit, a wireless data communication, and a user-friendly software embedded with a data processing algorithm. The proposed scheme is demonstrated by a three-phase inverter-based motor drive mission-profile emulator. The final results give that the proposed indicator is increased by about 25 times for the degraded devices and is up to 30% only for new devices.