Abstract:This paper explores the use of physics-informed neural networks (PINNs) to estimate motion and identify system parameters of a moored buoy under three different sea states. PINNs are a deep learning architecture that incorporates physical information to provide an interpretable and physically-meaningful neural network model, making it well-suited for modeling offshore moored structures with high nonlinearity. The moored buoy is modeled as a nonlinear ordinary differential equation (ODE), and the general formulation of PINN for the system of ODEs is established. Two new metrics for motion estimation and system identification are proposed to evaluate the accuracy and efficiency of the implementation of PINN. The results demonstrate that PINN can accurately estimate motion and identify system parameters by choosing appropriate hyperparameters (HPs). This paper also investigates the effects of the number of layers, nodes, and learning rate on motion estimation and system identification to provide a benchmark for selecting optimal HPs. This study finds that hyperparameter optimization can reduce the relative error of identified parameters by up to two thousand times compared to no optimization. Motion estimation prefers large neural networks for all sea states, while at least three layers of neural networks are needed for accurate parameter identification. The paper also provides a look-up table to investigate further implementing PINN on moored floating offshore structures. Proper selection of HPs is crucial as it can incur up to three orders of magnitude PINN loss and exceedingly-high identification error. Overall, this study highlights the applicability of PINN in modeling complex offshore structures and provides insights into selecting optimal HPs for accurate and efficient estimation of motion and system parameters.

Physics-informed Neural Network for System Identification of Rotors

Physics-Informed Neural Networks for the Condition Monitoring of Rotating Shafts

Improved High Resolution Estimation Approach for Rotor Fault Diagnosis

Physics Informed Neural Networks for Fault Severity Identification of Axial Piston Pumps

Induction Motor Failure Identification Based on Multiscale Acoustic Entropy Feature Selection and Hierarchical Adaptive Neuro-Fuzzy Inference System with Localized Recurrent Input

Hybrid physics-embedded recurrent neural networks for fault diagnosis under time-varying conditions based on multivariate proprioceptive signals

Data-driven system identification and optimal control of an active rotor-bearing system

Rotor Faults Diagnosis for Induction Motors Considering Effects of the Load and Inertia

Analysis of Inter-Turn Short-Circuit Faults in Brushless DC Motors Based on Magnetic Leakage Flux and Back Propagation Neural Network

ResNet diagnosis of rotor faults in oil transfer pumps

Unbalanced Position Recognition of Rotor Systems Based on Long and Short-Term Memory Neural Networks

Bayesian model‐based fault diagnosis for the rotor

Probabilistic Model-Based Fault Diagnosis Of The Rotor System

Bayesian model-based fault diagnosis for the rotor

A Fault Diagnosis Method of Rotor System Based on Parallel Convolutional Neural Network Architecture with Attention Mechanism

A Neural Network Identification Technique for a Foil-Air Bearing Under Variable Speed Conditions and Its Application to Unbalance Response Analysis

Motion estimation and system identification of a moored buoy via physics-informed neural network

Response Estimation and System Identification of Dynamical Systems via Physics-Informed Neural Networks

Research on Fault Diagnosis of Steam Turbine Rotor Unbalance and Parallel Misalignment Based on Numerical Simulation and Convolutional Neural Network

Nonlinear mechanical response analysis and convolutional neural network enabled diagnosis of single-span rotor bearing system

On-line Rotor Systems Condition Monitoring Using Nonlinear Output Frequency Response Functions under Harmonic Excitations