Scholarlink

This article presents control of quarter-car passive suspension system using an Artificial neural network (ANN) controller. A mathematical model of the proposed system has been developed and simulated in Proportional-integral-derivative (PID) controlled environment. The ANN controller has been designed using the outputs of PID controller and trained using three different training algorithms namely Levenberg–Marquardt (LM), Bayesian-Regularization (BR) and Scaled-Conjugate-Gradient (SCG). The performance of these machine learning algorithms has been analyzed in terms of settling time and maximum percentage overshoot response. The results clearly indicate superior performance of BR-algorithm compared to other proposed algorithms. The settling time taken to stabilize the complete system using BR, LM and SCG algorithm were 4.0 s, 8.0 s and 14.0 s respectively. The maximum percentage overshoot responses obtained for sprung mass displacement using BR, LM and SCG algorithm were 0.7%, 1.4% and 6.7% respectively.

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

Quarter-Car Passive Suspension System Control Using Different Machine Learning Algorithms

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Ashwani Kharola,
  • Syed Farrukh Rasheed,
  • Vishwjeet Choudhary,
  • Arvind Kumar,
  • Tarun Kumar Dhiman,
  • Abhijat Karn

摘要

This article presents control of quarter-car passive suspension system using an Artificial neural network (ANN) controller. A mathematical model of the proposed system has been developed and simulated in Proportional-integral-derivative (PID) controlled environment. The ANN controller has been designed using the outputs of PID controller and trained using three different training algorithms namely Levenberg–Marquardt (LM), Bayesian-Regularization (BR) and Scaled-Conjugate-Gradient (SCG). The performance of these machine learning algorithms has been analyzed in terms of settling time and maximum percentage overshoot response. The results clearly indicate superior performance of BR-algorithm compared to other proposed algorithms. The settling time taken to stabilize the complete system using BR, LM and SCG algorithm were 4.0 s, 8.0 s and 14.0 s respectively. The maximum percentage overshoot responses obtained for sprung mass displacement using BR, LM and SCG algorithm were 0.7%, 1.4% and 6.7% respectively.