Intelligent Fault Diagnosis of Hydraulic Pumps Based on Multi-Source Signal Fusion and Dual-Attention Convolutional Neural Networks.

As a core component of hydraulic systems, hydraulic pumps generate vibration signals that contain abundant key features reflecting the operational state of internal machinery. However, most existing fault diagnosis methods rely solely on single-channel vibration data, neglecting the correlations and complementarities among multi-channel signals, which results in unstable and less accurate diagnostic outcomes. To address this limitation, this study proposes an intelligent fault diagnosis approach for hydraulic pumps based on multi-source signal fusion and a dual attention mechanism. First, vibration, pressure, and acoustic signals are transformed into time-frequency feature images, and an RGB image fusion strategy is applied to map the time-frequency representations of different signals into the individual channels of a color image. Subsequently, a convolutional neural network incorporating enhanced channel and spatial attention mechanisms is constructed to extract features from the fused images and perform classification. Experimental results demonstrate that the proposed method significantly improves fault diagnosis performance and outperforms other deep learning-based approaches, offering a novel strategy for intelligent hydraulic pump diagnostics with promising engineering applications.