Iterative Gradient Descent-Based Finite Control Set Predictive Current Control With Least-Squares Optimized Duty Cycles

Abstract

Finite Control Set Predictive Current Control (FCS-PCC) is widely recognized as a competitive control strategy in the field of electrical drives, due to its superiority of fast dynamic response and low switching frequency. However, FCS-PCC is penalized by its inherent drawback that the discrete nature of switching states leads to relatively high torque and current deviations. In this paper, an iterative gradient descent method combined with least squares optimized duty cycles is presented to improve the steady-state performance of FCS-PCC. Unlike the cost function optimization in the conventional FCS-PCC, the quadratic programming problem is solved from a geometric perspective, by obtaining the gradient descent which minimizes the tracking deviation in the fastest manner. To synthesize the gradient descent, the optimal stator current derivatives in the current and previous iteration are employed, and their duty cycles are determined by the least squares method. The abovementioned procedures are iteratively repeated in the dichotomy-based periods. The experimental performance of the proposed gradient descent based FCS-PCC is verified at a 8 kHz sampling frequency, which is compared with that of conventional and dichotomy-based FCS-PCC. It is validated that the proposed algorithm outperforms the conventional and dichotomy-based FCS-PCC at both the steady-state and transient state.