Motor Speed Control With Convex Optimization-Based Position Estimation in the Current Loop

This article presents a sensorless machine drive scheme, combining the motion model predictive control (MPC) in the speed loop and the position estimation in the current loop. The current-loop-based position estimation algorithm is to estimate the present rotor position with the phase current information of the previous samples. These methods cannot perform perfectly in the low speed when a heavy load suddenly occurs, as the aggressive motion disturbance can impact the convergence of the low-speed position estimation, or even make divergence and estimation failure. This issue is especially terrible for the surface permanent magnet synchronous machines (SPMSMs), as the magnetic saliency is weak at heavy loads. The target of this article is to solve this issue with respect to the servo system, where the machines need to frequently start and stop with heavy loads, and hence the above risk must be considered. The convex optimization is applied in the position estimation, and the optimization cost value is extended to adjust the injection amplitude and the reference trajectory of the MPC control. With this strategy, the sensorless control can operate with a high stability and accuracy. The scheme is validated in an SPMSM test bench.