Acoustic Field Reconstruction in Tubes via Physics-Informed Neural Networks

2025-05-18

AI Summary

This work addresses the acoustic duct inverse problem: reconstructing the full-domain acoustic field given only sparse, noisy pressure measurements at the radiation end and without prior knowledge of the radiation boundary model. We propose a physics-informed neural network fine-tuning method (PINN-FTM), which enforces the wave equation as a hard physical constraint and jointly optimizes both network parameters and unknown boundary parameters—eliminating the need for an explicit radiation model. The method enables end-to-end high-fidelity reconstruction of both acoustic pressure and particle velocity fields. Experiments demonstrate that, under strong noise and extremely limited observations, PINN-FTM achieves 3.1× higher robustness than conventional optimization methods and reduces radiation parameter inversion error by 42%. These results significantly broaden the applicability and reliability of physics-informed neural networks in acoustic inverse problems with unknown or unmodeled boundary conditions.

Technology Category

• Constraint Satisfaction and Optimization: Constraint Learning and Acquisition
• Search and Optimization: Mixed Discrete/Continuous Search
• Planning, Routing, and Scheduling: Activity and Plan Recognition

Application Category

• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

This study investigates the application of Physics-Informed Neural Networks (PINNs) to inverse problems in acoustic tube analysis, focusing on reconstructing acoustic fields from noisy and limited observation data. Specifically, we address scenarios where the radiation model is unknown, and pressure data is only available at the tube's radiation end. A PINNs framework is proposed to reconstruct the acoustic field, along with the PINN Fine-Tuning Method (PINN-FTM) and a traditional optimization method (TOM) for predicting radiation model coefficients. The results demonstrate that PINNs can effectively reconstruct the tube's acoustic field under noisy conditions, even with unknown radiation parameters. PINN-FTM outperforms TOM by delivering balanced and reliable predictions and exhibiting robust noise-tolerance capabilities.

Problem

• Reconstructing acoustic fields in tubes using Physics-Informed Neural Networks (PINNs)
• Handling unknown radiation models and limited noisy pressure data
• Comparing PINN-FTM and traditional methods for radiation coefficient prediction

Innovation

• Physics-Informed Neural Networks for acoustic reconstruction
• PINN Fine-Tuning Method for radiation coefficients
• Robust noise-tolerant acoustic field prediction