Novel Virtual Active Flux Injection-Based Position Error Adaptive Correction of Dual Three-Phase IPMSMs Under Sensorless Control

In this paper, a novel signal injection method is proposed to correct the position error under sensorless control, with the help of two sets of windings in a dual three-phase interior permanent magnet synchronous machine (DTP-IPMSM). By injecting two opposite virtual active fluxes with the same amplitudes into two position observers, the corresponding difference of estimated positions between two position observers can be utilized to derive the correlation between the position error and the constructed cost function (composed of <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">dq</i>-axis currents/voltages). It is found the cost function is independent of stator resistance, PM flux linkage, and inverter nonlinearity. Consequently, the position error can be adaptively corrected by minimizing the constructed cost function. Compared with conventional parameter estimation methods for position error correction separately, the total position estimation errors caused by mismatches in temperature-dependent parameters and inverter nonlinearity can be collectively corrected. Moreover, utilizing the DTP winding characteristic of DTP-PMSMs, the torque fluctuation can be avoided with the opposite virtual active flux injections. The experiments are given to validate the proposed method.