March 3, 2026

Towards uncertainty-aware physics-informed neural networks for prognostics in bushings

Puntos clave

The hybrid methodology predicts the evolution of accumulated wear precisely in mechanical joints under real-world conditions, indicating advancements in prognostics.
Wear evolution in revolute clearance joints remains a challenge; however, the outcome reveals significant predictive capability using deep learning techniques.
Analysis combines physical principles and modified long short-term memory layers with an Archard wear equation for effective modeling of wear dynamics.
Uncertainty quantification using Monte Carlo dropout enhances the robustness of predictions, emphasizing the need for accurate prognostics in engineering systems.

Resumen

Wear in mechanical joints represents a key factor influencing the reliability and operational lifespan of engineering systems. Revolute clearance joints, in particular, are susceptible to progressive degradation because of inherent gaps that induce dynamic contact and friction. While physical and artificial intelligence models exist, accurately predicting wear evolution under real-world conditions remains an open challenge. To address this, a hybrid modeling framework is introduced that integrates physical principles with deep learning techniques. The approach combines modified long shortterm memory layers with a final Archard wear equation layer, enabling the modeling of nonlinear temporal dynamics associated with wear processes. Monte Carlo dropout is incorporated for uncertainty quantification. The results obtained on a sensorized crankslider system test bench show that the hybrid methodology is capable of predicting the evolution of accumulated wear, offering a significant advance towards prognostics.

Preguntar a la IA

Me gusta

Guardar