A Review of Data-Driven Surrogate Models for Design Optimization of Electric Motors

Electric motor is one of the core components of electronic propulsion systems and play an essential role in the industry. The optimal design of an electric motor poses a complex nonlinear problem, often challenging traditional methods to strike a balance between accuracy and efficiency. Achieving accurate analysis and holistic optimization typically entails significant computational requirements, particularly when dealing with massive individuals. As a result, researchers began to explore the utilization of data-driven surrogate models to resolve this dilemma. This review paper focuses on investigating the leading techniques employed for constructing data-driven surrogate models to assist and facilitate the design optimization process of electric motors. These techniques encompass statistical models, machine-learning (ML) models, deep-learning (DL) models, and other artificial intelligence-based technologies. The article provides a comprehensive survey of the underlying principles and offers detailed examples of studies that have utilized these diverse models. Besides, the performances and potentials of these models are highlighted with comments, shedding light on their respective strengths and limitations. Furthermore, the research challenges that lie ahead and promising avenues for future improvements under this topic are discussed.