In this study, a fault diagnosis approach for the hydraulic system of chain jacks based on multi-source sensor data fusion is proposed. We developed a hydraulic test rig for the chain jacks with a special measurement and control system to measure and collect real-time data on pressure, temperature and flow in different operating conditions. The proposed approach integrates convolutional neural networks (CNN) and long and short-term memory (LSTM) at the network level to extract the spatial–temporal features of the time-series data measured by sensors. Compared with the artificial neural network (ANN), the accuracy of the CNN-LSTM hierarchical diagnosis model was 96.4%, which improved the diagnosis accuracy by 4.4% and enhanced the generalization ability and stability. This study provides a hierarchical monitoring approach for the service status of marine spread mooring systems and chain jack equipment, which is essential for the safe operation of marine equipment.