Research on hydraulic system fault diagnosis method based on machine learning

Abstract This study proposes a machine learning-based fault diagnosis method for hydraulic systems, focusing on using K-means clustering algorithm for data preprocessing, use Support Vector Machine (SVM) and Convolutional Neural Network Gated Recurrent Unit (CNN-GRU) models respectively for fault classification. By performing cluster analysis on sensor data, the data dimensions can be effectively reduced and the efficiency of fault analysis can be improved. The results show that the accuracy of the SVM algorithm in the cooler status and valve status classification tasks reached 99.77% and 100.00% respectively. After the introduction of the clustering algorithm, its training time was significantly reduced, showing extremely high real-time fault diagnosis capabilities. The CNN-GRU model performs particularly well in handling complex fault classification tasks, especially in the hydraulic accumulator fault classification task, with an accuracy rate as high as 96.60%, which is significantly better than the SVM model. Although the training time of CNN-GRU is longer, its advantages in complex pattern recognition give it obvious advantages in high-accuracy application scenarios. Multi-fault diagnosis was conducted, and the results show that the CNN-GRU model achieves the best performance without employing clustering, emphasizing the importance of preserving the integrity of the original sensor data in complex fault scenarios.