Air-Gap Permeance and Reluctance Network Models for Analyzing Vibrational Exciting Force of In-Wheel PMSM

There is recognized interest in in-wheel motors for vehicle traction. Studies have gradually focused the design of in-wheel motors on electro-mechanical vibration, subject to the demand of driving comfort. It is crucial to model, analyze, and minimize the air-gap exciting force, the dominant source of vibration. In order to qualitatively and quantitatively study the air-gap vibrational exciting force and evaluate its characteristic for an in-wheel outer-runner permanent magnet synchronous machine (PMSM), this research proposes an air-gap permeance model (APM) and a unique adaptive reluctance network model (ARNM). In the process of motor design, APM is a quick means of determining dominant radial force density (RFD) harmonics and minimizing specific components that may produce considerable vibration. Morphing surface-mounted magnets and nonuniform air-gap length are guided by design optimization which complicates the analysis. Yet, proposed ARNM can account for this geometry variation and handle deformed gap geometry by using modified equation-based local permeance and residual flux. The speed and accuracy of proposed model are verified through the comparison of flux density and force mapping data between the calculation by proposed analytical model and the simulation by finite element tools. A 3-kW prototype is fabricated and tested to explore the effectiveness of ARNM-based performance prediction.

Automated summary

There is a recognized interest in using in-wheel motors for vehicle traction. This research focuses on the design of in-wheel motors in relation to electro-mechanical vibration, with the aim of improving driving comfort. The air-gap exciting force is identified as the main source of vibration, and the proposed models aim to qualitatively and quantitatively evaluate this force for an in-wheel outer-runner permanent magnet synchronous machine (PMSM). The proposed models are verified through comparison with simulation data and tested using a prototype, demonstrating their effectiveness in performance prediction.