Open-Switch Fault Detection in Five-Phase Induction Motor Drives Using Model Predictive Control.

Abstract

Achieving a self-reconfigurable fault-tolerant control in multiphase machines requires a fast fault detection and localization. Most fault detection techniques inherit the three-phase approach by defining fault indices in a per-phase basis. A recent approach suggests an alternative fault detection mechanism based on vector space decomposition (VSD) variables, but the study is limited to open-phase faults (OPFs) for a six-phase drive that is regulated under field-oriented control. It is known, however, that 1) the open-switch faults (OSFs) in the converter are more likely than the OPF in the machine and 2) the drive performance in the event of an open-circuit fault is more critical when model predictive control (MPC) is used. This work extends the study of the VSD fault detectionmethod to multiphase machines with different number of phases (five), control strategy (MPC), and type of faults (OPF and OSF). Although experimental results show that MPC misbehaves after the fault occurrence, the fast detection provided by the VSD approach allows a satisfactory transition to postfault mode of operation.