A comparison of a signal-injection method and a discrete-search algorithm for MTPA tracking control of an IPM machine

The efficiency of interior permanent magnet synchronous machines can be increased by a loss minimization approach when implementing field-oriented control. Model-based optimization gives operating points that yield maximum torque for a given stator current amplitude. The true loss minimum however can differ due to magnetic saturation and temperature variations. Therefore an online discrete search algorithm using a gradient-free Nelder-Mead simplex method is proposed and compared to an existing signal-injection method that was previously published. Simulation results show that the discrete search algorithm has a faster convergence time than the signal-injection method. Once the MTPA operating point is found no further current nor torque ripple remain that could influence the performance of the drive or excite mechanical vibrancy. Therefore it is better suited to track the MTPA trajectory in highly-exploited IPM machines that have a significant saturation characteristic as it occurs in automotive applications.