Parameter estimation of multilayer metal specimens using electromagnetic acoustic resonance

Copper–aluminum multilayer plates are used in electric vehicles and power-transmission systems, requiring nondestructive estimation of layer thicknesses and material parameters. This study proposes an electromagnetic acoustic resonance method for estimating layer thicknesses and sound velocities from resonance frequencies. An analytical resonance-frequency model is used to construct a frequency-residual objective function, and its parameter-space distribution is analyzed to assess identifiability. Finite-element simulations of thickness configurations are combined with global optimization to evaluate inversion performance. When either layer thicknesses or sound velocities are known, the remaining parameters can be estimated with relative errors of approximately 2% or less. When both are estimated simultaneously, most simulated errors remain within 5%; however, Monte Carlo analysis shows that frequency uncertainty is strongly amplified for unbalanced layer configurations. Optimizer comparisons indicate that the genetic algorithm achieves accuracy comparable to other global and multi-start methods with moderate computational cost. Preliminary experiments on two anisotropic copper–aluminum specimens show agreement between estimated total thicknesses and micrometer measurements, with directional total-thickness estimates differing by no more than 0.05 mm. These results demonstrate the feasibility of the method while highlighting the effects of parameter identifiability, frequency uncertainty, and material anisotropy.