Eddy current displacement sensing is an effective method for non-contact position measurement through conductive metallic shielding layers and has important applications in enclosed metallic structures such as magnetic levitation pumps, bearingless motors, and ultra-pure fluid transportation systems. Its detection performance is highly dependent on the excitation frequency. Frequency selection based solely on empirical experience may lead to insufficient displacement response and reduced detection sensitivity. To address the lack of a quantitative criterion for excitation frequency selection in eddy current displacement detection through fixed metallic shielding layers, an optimal excitation frequency determination method based on an analytical coil impedance model is proposed. First, a variable-gap multilayer conductive structure model for detecting the displacement of an internal composite metallic moving target through a fixed metallic shielding layer is established. An analytical expression of the coil complex impedance under different air-gap heights and excitation frequencies is then derived. Subsequently, an experimental platform for through-metal eddy current displacement detection is developed to validate the numerical results of the analytical model, and a first-order equivalent parasitic parameter correction model is introduced to compensate for model deviations in the high-frequency range. The experimental results demonstrate that the proposed model can accurately characterize the variation of coil impedance with air-gap height, and good agreement between the calculated and measured results is achieved below 1 000 kHz. Finally, the relative variation rate of the coil electrical parameters is adopted as the evaluation index for excitation frequency selection, and the responses of coil resistance, inductance, and complex impedance to target displacement are analyzed. The results indicate that the relative variation rate of the complex impedance first increases and then decreases with increasing excitation frequency, reaching its maximum at 400 kHz. Therefore, the optimal excitation frequency for the investigated through-metal displacement detection structure is determined to be 400 kHz. The proposed method transforms empirical frequency selection into a quantitative evaluation process based on analytical modeling and experimental responses, providing a reference for excitation frequency optimization in displacement detection of moving targets behind conductive shielding layers and for position estimation in enclosed structures.