Energy-Efficient MPCC Using Centroid-Synthesized Virtual Voltage Vectors for IM Drives in Electric Vehicles

Abstract

This paper presents an improved, energy-efficient Model Predictive Current Control (MPCC) strategy based on centroid-based virtual voltage vector synthesis for three-phase inverter-fed induction motor drives in electric vehicle (EV) applications. Unlike conventional finite-set MPCC methods that rely on cost function evaluation over discrete switching states, the proposed approach eliminates the need for look-up tables by employing a pre-defined set of virtual vectors. These centroid-based virtual voltage vectors are synthesized by combining two adjacent active vectors and two nonzero voltage vectors in opposite directions adjacent to the sector replacing the traditional switching set. They approximate the reference voltage vector in both magnitude and phase angle, thereby reducing current tracking error through a simplified cost function. The number of candidate vectors is reduced, preserving computational efficiency. Furthermore, the scheme ensures zero average common-mode voltage (CMV) per sampling interval by completely avoiding zero-voltage vectors (ZVVs). The proposed method reduces torque ripple by up to 17% compared to the conventional approach and lowers stator current total harmonic distortion (THD) by 37%, while ensuring evenly distributed switching transitions among inverter legs. This results in reduced switching losses and enhanced drive efficiency-particularly advantageous in EV applications. Experimental validation under the high-speed extra urban driving cycle (EUDC) and low-speed ECE-R15 cycle, including torque ripple and energy consumption analysis, confirms the effectiveness of the approach, achieving an overall efficiency of 83.3%.