Loss Minimization Algorithm for Surface-Mounted PMSM Using Ripple-Based Extremum Seeking

In this article, a fast and parameter-intensive loss minimization algorithm (LMA) is proposed for the surface-mounted permanent magnet synchronous machine (PMSM). The algorithm utilizes ripple correlation control to steer the operating point toward the optimal solution by evaluating the correlation between the injected ripple on the control variable and its effect on the input power. Unlike model-based LMAs, this method does not rely on the motor loss model, its parameters, or precomputed information. Instead, it employs a high-speed search-based procedure to minimize the input power directly. The theoretical analysis includes a design procedure and a method for determining the upper bound of the injected ripple frequency based on the principles underlying loss minimization of PMSMs. Since the d-axis current does not contribute to torque production in surface-mounted PMSMs, the artificial perturbation introduced by the algorithm does not result in undesirable torque ripple. The analysis is supported by simulations and experimental tests. The results demonstrate that the proposed algorithm enables the system to converge to the optimum point within around 1.5 s, making it suitable for high-dynamic applications.