A Neuro-Physical Framework for Three-Dimensional Elastodynamic Wave Prediction in Mechanical Components
Year: 2026
Type: article
Abstract: Predicting vibration-wave propagation in mechanical components requires solvingtransient, three-dimensional elastodynamic equations across complex geometries, material interfaces, and frequency-dependent damping. This paper proposes a researchframework—termed the Causal Energy-Constrained Neural Operator (CENOp)—for learning mappings from material properties, geometry, initial conditions, andexternal excitation to displacement, velocity, and stress fields.The framework combines continuum mechanics with a neural operator, enforcingmomentum balance, constitutive consistency, boundary and interface conditions, andan energy-dissipation inequality. A memory-variable constitutive model representslinear viscoelastic damping, while optional stochastic fields represent materialuncertainty. The proposed additions are hypotheses for future testing, not establishedphysical laws.A reproducible validation program is outlined using analytical wave solutions,finite-element simulations, and measurements from instrumented components. Theframework is intended as a fast surrogate for repeated forward predictions andas a tool for parameter identification, damage localization, and uncertainty-awarestructural monitoring. Its performance must be demonstrated against independentnumerical and experimental benchmarks before engineering use.
Author: ALDHUFRI, KHALED
Language: English
Subfield: Statistical and Nonlinear Physics
Field: Physics and Astronomy
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7223560379