Incorporating Segmentive & Augmentive Tools in Cosine KNN for Bearing Intelligent Fault Diagnosis

To address challenges in Intelligent Fault Diagnosis (IFD) of bearings in rotating machinery, this study aims to propose a model that reduces computational resources and expert dependence while enhancing performance under diverse operating conditions.

The study introduces the Segmentation-based Augmented Cosine K—nearest neighbors (SACK) model, integrating raw vibration signal segmentation and data augmentation techniques. Unlike traditional approaches, SACK eliminates frequency or time-frequency signal processing, simplifying augmentation using existing tools.

Evaluation of the SACK model reveals satisfactory performance and improvements compared to other Machine Learning (ML) models. Enhanced robustness across various operating conditions suggests potential efficiency gains in computational resources and reduced reliance on human experts. The proposed model presents a promising solution for cost-effective IFD implementation in resource-constrained environments.