Abstract
In this article, two kinds of finite control set-model predictive voltage control (FCS-MPVC) methods have been proposed for linear induction machine to eliminate the cumbersome selection of the weighting factor (WF), achieve lower ripples of both thrust and flux, and reduce the heavy calculation steps. The first method, FCS-MPVC I, uses one voltage vector (VV) during the whole control period, and the second method, FCS-MPVC II, uses two VVs. Moreover, the proposed methods use a single cost function without any WF. This function consists of the errors between primary voltage references with their predicted values in the alpha beta-axis. Depending on the distance between the reference VV (RVV), as calculated using the deadbeat control method, and all candidate VVs (CVVs), the first proposed method, FCS-MPVC I, can select the nearest VV to the RVV as the optimal VV (OVV), which can reduce the redundant evaluation work greatly. Meanwhile, the second proposed method, FCS-MPVC II, applies two VVs, active VV and null VV, to adjust the duty ratio effectively. To avoid the iterated evaluation for all CVVs, one simplified search order excluding unsuitable CVVs has been proposed to determine the optimal OVV, which selects only one active VV from the two adjacent active VVs in the sector owning the RVV. Afterward, based on a prototype of 3-kW arc induction machine with a large radius, extensive simulation and experiments have fully demonstrated that the proposed methods can get much lower thrust ripples and less calculation steps effectively, compared with those of conventional finite control set-model predictive thrust control strategy.