Monitoring of Ball Bearing Based on Improved Real-Time OPTICS Clustering

This paper presents a new methodology of the Real-Time monitoring (IRT-OPTICS) for the detection of defect in rolling bearing by combining three domain features (time, frequency and scale), and reducing dimension by two methods: Principal Component Analysis (PCA) and Kernel Principal Component Analysis (KPCA), then classifying data by OPTICS method (Ordering Points To Identify the Clustering Structure). This methodology generated in three loops: initialization, detection and follow-up. The initialization loop is the fundamental and the based loop to start the surveillance. The detection loop is the phase where machines defects can be detected, in case there is defect, the third loop will start. The third loop is the follow-up; it aims to observe the degradation state. The decision to replace a component can be taken in the follow-up. These three loops use the combination of three features extraction methods: time domain, frequency domain and time scale domain also they use reduction of dimension by two methods, to combine features in three components and then to classify them by the OPTICS method. The proposed method has been validated numerically and experimentally.