No abstract

No abstract

e aim of this paper is to propose a simple method for detecting a faulty bearing. Frequently, the time domain and data-based methods used are noise a ected, sample size dependent, machine and load dependent, involve complex mathematical calculations or require training of the algorithm. To overcome the above issues, the symbolic dynamics approach is used. e time series vibration data is converted into a symbolic series om which a dictionary of the signal is constructed. e common signal index (CSI) parameter is computed based on the dictionary constructed om the reference signal and the test signal. Deviations in the computed CSI value om the CSI value of the healthy state serve as an indicator for the presence of a bearing fault. Healthy vibration data is used as a reference signal to detect a bearing in a healthy or faulty condition. e method is tested using three sets of data: two are experimentallygenerated data (one is the bearing fault vibration signal with additive noise and the other is a mixture of bearing fault vibration signals and vibration interference om the gear with additive noise) and the third is bearing fault simulation data. e e ect of the fault index on the variation of sample points and noise levels is studied. e algorithm is even tested for bearing faults at di erent locations, of various sizes and at varying loads. e advantages of the proposed method over the existing time domain and data-based methods are discussed. e results demonstrate the applicability of the proposed method for bearing fault detection.