This paper constructs a fault diagnosis model for antenna drive reducer bearings based on the fusion of Successive Variational Mode Decomposition (SVMD) and Convolutional Long Short-Term Memory Neural Network (CNN-LSTM). SVMD is utilized to decompose bearing vibration signals, obtaining multiple Intrinsic Mode Functions (IMFs) which are used to construct the dataset. The CNN extracts local spatial features from the signals, while the LSTM captures temporal characteristics, enabling deep analysis of fault features. Experimental results demonstrate that the proposed model achieves a classification accuracy of 97.63%. Compared to CNN, CNN-LSTM, Transformer, and AlexNet models, the classification accuracy is improved by 5.94%, 2.42%, 1.12%, and 2.45%, respectively. This model provides a novel solution approach for fault diagnosis in antenna drive reducer bearings.