Scholarlink

This study investigates the use of Physics-Informed Neural Networks (PINNs) for modeling underdamping and overdamping in spring mass damper systems, employing a novel approach that forgoes traditional data point evaluations. Instead, the PINN framework developed here relies solely on residual calculations and initial conditions, embedding the physical laws governing the system dynamics directly into the learning algorithm. This method highlights the comparative performance of different activation functions and optimizers. Findings demonstrate that, with suitable choices of these parameters, PINNs can effectively capture the intricate dynamics of underdamping and overdamping conditions.

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

Comprehensive Analysis of Damping Conditions Using PINNs with Different Activation Functions and Optimizers

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Prasant Sahay,
  • Krishna Kant Mishra,
  • Aryan Verma,
  • Rajnish Mallick,
  • Dineshkumar Harursampath,
  • Manoj Sahni

摘要

This study investigates the use of Physics-Informed Neural Networks (PINNs) for modeling underdamping and overdamping in spring mass damper systems, employing a novel approach that forgoes traditional data point evaluations. Instead, the PINN framework developed here relies solely on residual calculations and initial conditions, embedding the physical laws governing the system dynamics directly into the learning algorithm. This method highlights the comparative performance of different activation functions and optimizers. Findings demonstrate that, with suitable choices of these parameters, PINNs can effectively capture the intricate dynamics of underdamping and overdamping conditions.