The appearance of repetitive transients in a vibration signal is one typical feature of faulty rolling element bearings. However, accurate extraction of these fault-related characteristic components has always been a challenging task, especially when there is interference from large amplitude impulsive noises. A frequency domain multipoint kurtosis (FDMK)-based fault diagnosis method is proposed in this paper. The multipoint kurtosis is redefined in the frequency domain and the computational accuracy is improved. An envelope autocorrelation function is also presented to estimate the fault characteristic frequency, which is used to set the frequency hunting zone of the FDMK. Then, the FDMK, instead of kurtosis, is utilized to generate a fast kurtogram and only the optimal band with maximum FDMK value is selected for envelope analysis. Negative interference from both large amplitude impulsive noise and shaft rotational speed related harmonic components are therefore greatly reduced. The analysis results of simulation and experimental data verify the capability and feasibility of this FDMK-based method
27 January 2023 in Engineering Applications of Artificial Intelligence
The vibration response of faulty bearing is always characterized by periodic transient impulses in the signal. Generally, these fault-related features are inevitably submerged in noise and harmonic components. Mathematical morphology is an excellent method of noise reduction, which can retain the detail information of impulses in the time domain. However, the filtering effect of traditional morphological operator (MO) might be easily affected by random impulses, and the proper selection of the structure element (SE) depends heavily on the experience of researchers. In order to effectively remove these interferences and extract the fault features accurately, an improved method, named adaptive morphological filter (AMF), is proposed in this article. This method utilizes autocorrelation to lift MO in time domain to enhance periodic components, and the scale of SE can, therefore, be calculated with the local maximum of the autocorrelation spectrum. Since the selection of the optimal SE scale is adaptive, researchers’ experience is no longer needed, and there is also no need to calculate the fault characteristic frequency (FCF) for the determination of maximum scale of SE. The vibration and acoustical signals of faulty locomotive wheel set bearing are analyzed with this method, and the results verify its effectiveness and ability.
The properties of the time domain parameters of vibration signals have been extensively studied for the fault diagnosis of rolling element bearings (REBs). Parameters like kurtosis and Envelope Harmonic-to-Noise Ratio are the most widely applied in this field and some important progress has been made. However, since only one-sided information is contained in these parameters, problems still exist in practice when the signals collected are of complicated structure and/or contaminated by strong background noises. A new parameter, named Shock Pulse Index (SPI), is proposed in this paper. It integrates the mutual advantages of both the parameters mentioned above and can help effectively identify fault-related impulse components under conditions of interference of strong background noises, unrelated harmonic components and random impulses. The SPI optimizes the parameters of Maximum Correlated Kurtosis Deconvolution (MCKD), which is used to filter the signals under consideration. Finally, the transient information of interest contained in the filtered signal can be highlighted through demodulation with the Teager Energy Operator (TEO). Fault-related impulse components can therefore be extracted accurately. Simulations show the SPI can correctly indicate the fault impulses under the influence of strong background noises, other harmonic components and aperiodic impulse and experiment analyses verify the effectiveness and correctness of the proposed method.