Efficient dynamic modal load reconstruction using physics-informed Gaussian processes based on frequency-sparse Fourier basis functions

2025-02-01

Mechanical systems and signal processing

AI Summary

In structural health monitoring, direct measurement of external dynamic loads is often infeasible due to sensor placement constraints, inaccessible loading points, or unknown excitation characteristics. To address this challenge, this paper proposes a physics-informed Gaussian process (PI-GP) framework that jointly incorporates physical constraints and sparse frequency-domain priors. The method introduces a modal load inversion model driven by frequency-sparse Fourier basis functions, ensuring interpretability, robustness under limited data, and computational efficiency. Compared with conventional Tikhonov regularization and standard Gaussian process approaches, the proposed method reduces load reconstruction error by 42% and accelerates computation by 8.3× across multiple experimental scenarios. These improvements significantly enhance the capability for real-time, high-accuracy dynamic load identification.

Technology Category

• Intelligent Robots: State Estimation
• Machine Learning: Learning with Manifolds
• Planning, Routing, and Scheduling: Activity and Plan Recognition

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• User Modeling, Personalization and Recommendation: Psychology-informed user models and recommender systems
• Security and Privacy: Large-scale security measurements

Problem

• Reconstructs dynamic loads without direct force measurements.
• Uses physics-informed Gaussian processes for structural response modeling.
• Reduces computational complexity via frequency-sparse Fourier basis functions.

Innovation

• Physics-informed Gaussian processes for load reconstruction
• Frequency-sparse Fourier basis functions reduce complexity
• Probabilistic dynamic load model without force data