Abstract: Micro-motion features are regarded as critical characteristics of space targets, providing essential references for target classification and identification. However, traditional micro-motion feature extraction methods rely on complicated signal processing techniques, which are computationally intensive and exhibit limited robustness, particularly for composite micro-motions such as precession. To address these challenges, a precision period estimation method for conical space targets (with and without empennages) based on an enhanced convolutional neural network (ECNN) is proposed in this paper. Time-frequency representation of narrowband radar echoes are utilized to accurately estimate the micro-motion periods in the proposed method. Notably, reliable estimation results are achieved even under low signal-to-noise ratio (SNR) conditions. The feasibility and robustness of the proposed method are validated by the simulation experiments.