Abstract: In this study, finite element simulations were conducted using COMSOL Multiphysics software to model common defects, such as notches and inclusions, in powder metallurgy (PM) guide rod components. The effects of excitation frequency and defect location on the output voltage of the receiving coils in the electromagnetic tomography (EMT) system were systematically investigated. Furthermore, two image reconstruction algorithms, namely linear back projection (LBP) and Tikhonov regularization, were employed to reconstruct defect distribution images, and their imaging accuracy was comparatively analyzed. The results show that the output voltage of the receiving coils increases with excitation frequency, and a clear linear relationship was observed between the frequency and the imaginary part of the output voltage. Therefore, the imaginary part of the coil voltage is more suitable for characterizing the variation trend of the output voltage. In addition, the variation in the imaginary part of the output voltage becomes more significant when the defect is closer to either the receiving or excitation coil. Specifically, at an excitation frequency of 10 kHz, defects at different positions in the PM guide rod components can be more readily identified. A comparison of imaging quality indicates that the Tikhonov regularization algorithm provides superior reconstruction performance, as the notches and inclusions in the reconstructed images align more closely with the original model distributions. Hence, the Tikhonov regularization algorithm is more suitable for defect image reconstruction in PM guide rod components. This study provides a theoretical and methodological reference for enhancing the defect detection accuracy and imaging quality of EMT systems for PM guide rod products.