Monitoring of water contamination in lubricants for rotating machines using acoustic emission signals based on the IVY-VMD-CNN

Effective lubrication is essential for the smooth operation and safety of rotating machinery. However, water contamination will significantly reduce the lubrication effectiveness, accelerate equipment wear, and raise maintenance costs. For instance, the lubricating oils in journal bearings are particularly vulnerable to water ingress during operations, emphasizing the need for precise and reliable detection of contamination. This study captures acoustic emission (AE) signals generated from oil film shearing using a rotational rheometer and analyzes lubricant samples with controlled moisture to model hydrodynamic behaviors in water-affected journal bearings. A hybrid method integrating the IVY-VMD and CNN is proposed for accurate AE pattern recognition. The framework decomposes AE signals from the oil film shear process into adaptive IMFs via the IVY-VMD. Sensitivity-based IMF selection utilizes KLD to isolate moisture-related features for signal reconstruction. These improved signal profiles are subsequently classified using a CNN architecture. Simulation and experimental validations demonstrate that the methodology is effective in detecting contamination-induced AE characteristics, indicating its potential for enhanced lubrication monitoring.