Scholarlink

This paper presents the application of the Physics-Informed Neural Network (PINN) as a method for predicting the operating conditions of a planetary transmission. The method is based on training a neural network with the transmission’s nonlinear dynamic model. The neural network is added as a perturbation of the differential equations, and the neural network’s parameters are determined using the loss function. The loss function is computed as the mean square error between the original solution and the perturbed calculation. The parameters of the dynamic model are estimated from experimental data, which are then compared to the numerical predictions. The Physics-Informed Neural Network is an algorithm that trains and produces a numerical model capable of predicting the behavior of a nonlinear dynamic system with greater flexibility than one that relies solely on experimental data, and a simple dynamic model of a complex planetary transmission showed the advantages of PINN as a tool for monitoring the operation conditions of mechanical systems.

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

Prediction of the Remaining Lifespan of Planetary Gearboxes Using Physics-Informed Neural Networks

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Juan Carlos Jauregui-Correa,
  • Jovan Basaldua-Sanchez,
  • Jose de Jesus Santana,
  • Marco Cecarrelli

摘要

This paper presents the application of the Physics-Informed Neural Network (PINN) as a method for predicting the operating conditions of a planetary transmission. The method is based on training a neural network with the transmission’s nonlinear dynamic model. The neural network is added as a perturbation of the differential equations, and the neural network’s parameters are determined using the loss function. The loss function is computed as the mean square error between the original solution and the perturbed calculation. The parameters of the dynamic model are estimated from experimental data, which are then compared to the numerical predictions. The Physics-Informed Neural Network is an algorithm that trains and produces a numerical model capable of predicting the behavior of a nonlinear dynamic system with greater flexibility than one that relies solely on experimental data, and a simple dynamic model of a complex planetary transmission showed the advantages of PINN as a tool for monitoring the operation conditions of mechanical systems.