Machine Learning-Based Robust Feedback Observer for Fault Diagnosis in Bearing

Abstract

Rolling element bearing (REB) represent a class of nonlinear and multiple-degrees-of-freedom rotating machines that have pronounced coupling effects and can be used in various industries. The challenge of understanding complexity in a bearing’s dynamic behavior, coupling effects, and sources of uncertainty presents substantial challenges regarding fault diagnosis (FD) in a REB. Thus, a proposed FD algorithm, based on an TSK fuzzy multi structure feedback observer, is represented. Due to the effect of the system’s complexities and uncertainties for FD, a feedback observer (FO) is proposed. To address the FO drawbacks for FD in the REB such as robustness, the multi structure technique is represented. In addition, the TSK fuzzy algorithm is applied to the multi structure FO (MSFO) to increase the performance of signal estimation and reliability. In addition, the energy residual signals are generated and the machine learning technique known as a support vector machine (SVM) adaptively derives the threshold values that are used for classification the faults. The effectiveness of the proposed technique is validated using a Case Western Reverse University (CWRU) vibration dataset.