Abstract:Machine learning has drawn growing attention from the areas of fatigue, fracture, and structural integrity. However, most current studies are fully data-driven and may contradict the underpinning physical knowledge. To address this issue, we propose a physically consistent framework for fatigue life prediction that uses a probabilistic physics-informed neural network (PINN) to incorporate the physics underpinning the fatigue mechanism. Particularly, we consider the scatter of the fatigue life using a probabilistic neural network with the output to parametrize the fatigue life distribution. Then use neural networks' inherent backpropagation capa-bilities to automatically compute the derivatives that represent the physical knowledge. Finally, construct a composite loss function to encode the derivatives with certain physical constraints and uses a negative log -likelihood function to consider both failure data and run-out data. This enforces the network training process to learn a continuous function that describes the stress-life relationship satisfying both experimental data and physical knowledge. We demonstrate the proposed framework with sensitivity analysis and a comparison to the fully data-driven neural networks and the conventional statistical methods using the fatigue test data of three different materials. The results show that the proposed framework has a robust performance to effectively reflect the underlying physical knowledge and prevent overfitting issues. The findings provide a better understanding of neural networks' application to fatigue life prediction and suggest that one should be cautious when using a fully data-driven approach in scientific applications.

Improving Fatigue Life Prediction of Natural Rubber Using a Physics‐Informed Neural Network Model

A unified probabilistic fatigue life prediction model for natural rubber components considering strain ratio effect

A physics-informed neural network approach to fatigue life prediction using small quantity of samples

Prediction of Rubber Fatigue Life Using an Assimilation‐based Learning Approach and Incremental Crack Propagation Model

A Physically Consistent Framework for Fatigue Life Prediction Using Probabilistic Physics-Informed Neural Network

Probabilistic fatigue life prediction model of natural rubber components based on the expanded sample data

Residual Fatigue Life Prediction of Natural Rubber Components under Variable Amplitude Loads

A Hybrid Mechanism and Data‐Driven Approach for Predicting Fatigue Life of MEMS Devices by Physics‐Informed Neural Networks

A Process-Data-driven BP Neural Network Model for Predicting Interval-Valued Fatigue Life of Metals

A Method to Develop A Unified Fatigue Life Prediction Model for Filled Natural Rubbers under Uniaxial Loads

Intelligent prediction of fatigue life of natural rubber considering strain ratio effect

MFLP-PINN: A physics-informed neural network for multiaxial fatigue life prediction

Fatigue Life Prediction of Vulcanized Natural Rubber under Proportional and Non‐proportional Loading

Development of a cascaded multitask physics-informed neural network (CM-PINN) to construct the muti-physical field model of rubber bushing press fitting

Fatigue life prediction of natural rubber components using an artificial neural network

Experiment and Fatigue Life Prediction of Filled Natural Rubber under Variable Amplitude Loading Conditions

A physics-guided modelling method of artificial neural network for multiaxial fatigue life prediction under irregular loading

Experiment and Fatigue Life Prediction of Filled Natural Rubber Emphasizing on the Effect of Strain Ratio

PINN-AFP: A Novel C-S Curve Estimation Method for Asphalt Mixtures Fatigue Prediction Based on Physics-Informed Neural Network

Fatigue life prediction for a natural rubber anti-vibration component with mean strain effect

A physically informed neural network based on Basquin's formula for predicting fatigue life