An adaptive quasi-sliding-mode observer-based sensorless drive for heavy-duty interior permanent magnet synchronous machines

Abstract

Extended electromotive force (EMF)-based sliding-mode observer (SMO) is a promising solution for sensorless control of interior permanent magnet synchronous machines (IPMSMs) in the medium- and high-speed region. However, due to machine saliency, the magnitude of the extended EMF will change with load variations, and a phase lag will be observed in the estimated rotor position if the observer gains are chosen improperly. In the applications of heavy-duty, off-road hybrid electric vehicles, such load-dependent phase lags will significantly affect power/torque generation of an IPMSM when the load changes abruptly. Furthermore, considering switching losses, inverter size and noise, the favorable sampling ratio for control system implementation is in a low range, e.g., 15 samples per electric revolution, which will degrade the performance of the conventional SMO. To overcome these issues, an adaptive quasi-SMO (QSMO) using an online parameter adaption scheme is proposed to estimate the extended EMF quantities, which are then used to estimate the rotor position of the IPMSM. The resulting estimated position has zero phase lags and is highly robust to fast load variations. The effectiveness of the proposed adaptive QSMO is validated by experiments for a practical 150 kW IPMSM drive system under various conditions, such as four-quadrant operations and complete torque reversals.