Open-Circuit Fault Identification Method for Interleaved Converters Based on Time-Domain Analysis of the State Observer Residual

Abstract

Fault-tolerance capability is a critical matter in power conversion systems that require continuous operation. Multiphase converters provide an inherent redundancy that gives them fault-tolerant capability. In addition, the interleaved operation of these converters provides them with some advantages regarding total current ripple, which otherwise would be lost under a fault condition. This issue can be solved by means of an identification and reconfiguration system. This paper proposes a method for open-circuit fault identification, based on the time-domain comparison of the residual of a state-space observer with signatures per phase. Signature online generation, together with a proper normalization and thresholding strategy, provides immunity with respect to the operation point. The use of the phase information associated to the interleaved operation allows to reduce the number of sensed variables, making it possible to identify the faulty leg by only sensing input and output signals, which are conventionally used for control purposes. Hence, the measurement of signals within every leg is avoided. Experimental results on a four-phase interleaved buck converter validate the method capacity to identify faults in two switching periods. Robustness regarding abrupt changes in the operational conditions was tested.