Scholarlink

<p>This article presents a comprehensive methodology for designing and optimizing a multi-ring radial permanent magnet bearing (MRPMB) utilizing a mathematical model and artificial neural networks (ANN). A mathematical model is presented to calculate MRPMB force and stiffness characteristics. The mathematical model results were validated using the results of ANSYS mechanical APDL. A discrete optimization process was employed for the optimization of single-ring permanent magnet bearings (PMB). Subsequently, an ANN was developed to predict optimal design variables, along with corresponding maximum force and stiffness values, based on the mathematical model results. When compared to the mathematical model, the proposed ANN model gave similar prediction accuracy with a decrease in computational time. The mean deviation between the prediction and the mathematical model for force and stiffness remained 3–5% across the design variables. A comparative analysis validated the effectiveness of the proposed approach, demonstrating substantial improvements in prediction accuracy and computational efficiency compared to existing literature methods. This research presents an advanced computational tool that combines a mathematical model with an ANN-based prediction for an optimized MRPMB.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Pragmatic optimization of axially polarized multi-ring radial permanent magnet bearings using artificial neural networks

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年12月8日

查看全文
  • D. K. Supreeth,
  • Dhruv Deshwal,
  • Siddappa I. Bekinal

摘要

This article presents a comprehensive methodology for designing and optimizing a multi-ring radial permanent magnet bearing (MRPMB) utilizing a mathematical model and artificial neural networks (ANN). A mathematical model is presented to calculate MRPMB force and stiffness characteristics. The mathematical model results were validated using the results of ANSYS mechanical APDL. A discrete optimization process was employed for the optimization of single-ring permanent magnet bearings (PMB). Subsequently, an ANN was developed to predict optimal design variables, along with corresponding maximum force and stiffness values, based on the mathematical model results. When compared to the mathematical model, the proposed ANN model gave similar prediction accuracy with a decrease in computational time. The mean deviation between the prediction and the mathematical model for force and stiffness remained 3–5% across the design variables. A comparative analysis validated the effectiveness of the proposed approach, demonstrating substantial improvements in prediction accuracy and computational efficiency compared to existing literature methods. This research presents an advanced computational tool that combines a mathematical model with an ANN-based prediction for an optimized MRPMB.