Topology-oriented multi-objective optimization for interior permanent magnet synchronous motors using AutoML-based surrogate models

With the rapid expansion of the electric vehicle market, interior permanent magnet synchronous motors (IPMSMs) have become increasingly critical for high-performance drive systems. However, the diversity of rotor topologies and the computational intensity of traditional finite element analysis (FEA) present significant challenges for efficient motor design optimization. This paper proposes a multi-topology optimization framework based on automated machine learning (AutoML) that enables simultaneous optimization across diverse rotor configurations. The system integrates a unified representation method for multiple topologies with an ensemble-based surrogate modeling approach that automatically selects optimal algorithms and hyperparameters. cross-topology validation across four representative rotor topologies demonstrates that the proposed framework achieves prediction accuracies exceeding 95% while reducing computational time to less than 1/8000 of traditional FEA methods. The optimization system generates Pareto-optimal solutions for torque maximization and ripple minimization within 19 minutes, providing multiple viable design alternatives that meet specific performance requirements across different topological configurations.