The Rapid Detection Method of Lubricant Oxidation State Based on Artificial Olfactory System

Lubricant quality is a critical factor affecting the healthy operation of rotating equipment such as engines. During prolonged operation or abnormal conditions, lubricants undergo oxidation that compromises equipment safety. Therefore, monitoring lubricant oxidation status is crucial. However, current oxidation detection methods primarily rely on offline laboratory sampling, which is inefficient and unsuitable for real-time monitoring. This study proposes an olfactory-based lubricant oxidation detection method using artificial olfactory sensor arrays to collect volatile component response data during accelerated oxidation cycles. A support vector machine (SVM) model was established to identify oxidation status, achieving an average accuracy rate of 99.6%. Additionally, near-infrared (NIR) absorption spectra were simultaneously acquired. SVM models based on NIR data and integrated AOS-NIR data were developed for comparative analysis with our proposed method, demonstrating the superiority of the AOS-SVM model. The study also investigated the impact of kernel functions on machine learning performance and validated the AOS-SVM model's generalization capability through cross-validation experiments with two different lubricant brands. Consequently, this research provides a practical solution for rapid detection and monitoring of lubricant oxidation status.