Multiobjective Torque Ripple Minimization of a BLDC Motor Using NSGA-II

Abstract

In the development of ventricular assist devices (VADs), torque stability in electric motors is critical to ensuring continuous blood flow and minimizing adverse hemodynamic effects. This research describes a multi-objective geometric optimization framework for minimizing ripple torque in a brushless DC (BLDC) motor designed for a VAD with an axial drive but no central shaft. The electromagnetic model was implemented using the finite element method in COMSOL Multiphysics and coupled to MATLAB via LiveLinkTM, allowing for the automatic evaluation of geometric configurations. Five rotor and stator design parameters were designated as choice variables, and an optimization problem was established with three objectives: limiting torque variation, maintaining a minimal average torque of 50 mN·m, and regulating the air gap area. Optimization was performed using the NSGA-II evolutionary algorithm, running approximately 5000 model evaluations. The results showed that it is possible to reduce ripple torque from 14.8% in the original configuration to values below 5% within the Pareto feasible set, while maintaining the minimum axial torque requirement. The analysis revealed an inherent trade-off between ripple reduction and average torque magnitude. The proposed methodology proves to be an effective tool for the electromagnetic optimization of BLDC motors in biomedical applications.