Noninvasive condition monitoring of three-phase four-wire inverter system parameters

Abstract

Current condition monitoring of three-phase inverters is mainly for DC-link capacitance, with little attention paid to AC-side parameters, and some of these methods require the addition of DC current sensors or changes in operating conditions. In fact, the AC side inductance degrades with use. In particular, the neutral inductors of three-phase four-wire inverters need to be condition-monitored because of the high failure rate due to the long term exposure to huge ripples. To address the above problems, this paper proposes a zero-sequence current injection method, which can take into account both the AC and DC sides, satisfy the parameter identification of “large capacitance and small inductance”, and does not need to add additional sensors or sampling circuits. Then, the zero-sequence components are extracted by a high-order band-pass filter and a recursive least squares filter. Next, the extracted components are processed by the proposed adaptive forgetting factor recursive least square observer to obtain the recognized parameter values. The speed and accuracy of parameter identification are greatly accelerated by the adaptive processing of the forgetting factor. The three-phase simultaneous identification and back-propagation iteration for the neutral inductor improves the identification accuracy and speed. Finally, a three-phase four-wire experimental platform is built. The experimental results show that the proposed method is still effective under different types of loads and load fluctuation.

© 2017 Elsevier Inc. All rights reserved.