Search

Mechanical fault diagnosis under nonstationary conditions is one of the hotspots in the condition monitoring research field, presenting still several difficulties. Vibration signals of faulty bearings exhibit second-order cyclostationarity when rotating speed and load are constant. Due to the time-varying rotating speed, the original cyclostationary signal becomes a special type of nonstationary signal. Considering the superiority of cyclostationary analysis in bearing fault diagnosis, it is interesting to extend the cyclostationary paradigm to make it suitable for particular signals that show irregular statistical cyclicity (ISC). Modeling bearing vibrations with speed fluctuation, however, is a controversial topic, which further sparked a debate on how to process bearing signals with ISC. Therefore, in this article, a detailed comparison of two signal models is made, the pace irregularity and the time warping, to illustrate the advantages and disadvantages of the modeling ideas. Based on these models, two novel tacholess diagnostic methods, the synchronization-sequence method and the dewarping method, are proposed for bearing diagnosis in nonstationary operations. By identifying the impact points with the same cyclic feature, the synchronization-sequence method shifts the impulse patches to the same interval, so the signal becomes cyclostationary. Using another strategy, the dewarping method recovers the regular statistical property by maximizing the squared modulus of the cyclic autocorrelation of specific feature points. The results show that both models can help diagnose local bearing faults, and the classical cyclostationary diagnostic tools become more powerful when extending to nonstationary operation conditions. In addition, the speed fluctuation curve can also be estimated by the two methods without using a tachometer. The characteristic of the proposed methods for speed estimation is that only the second-order cyclostationary information is utilized. Besides, the methods are robust to noise, so they are suitable for processing bearing data in the incipient failure stage. Numerical simulations and experimental data analyses corroborate the theoretical results. [ABSTRACT FROM AUTHOR]

Condition monitoring and fault diagnosis of bearings play important roles in production safety and limiting the cost of maintenance on a reasonable level. Nowadays, artificial intelligence and machine learning make fault diagnosis gradually become intelligent, and data-driven intelligent algorithms are receiving more and more attention. However, many methods use the existing deep learning models directly for the analysis of mechanical vibration signals, which is still lack of interpretability to researchers. In this paper, a method based on multilayer bidirectional gated recurrent units with attention mechanism is proposed to access the interpretability of neural networks in fault diagnosis, which combines the convolution neural network, gated recurrent unit, and the attention mechanism. Based on the attention mechanism, the attention distribution of input segments is visualized and thus the interpretability of neural networks can be further presented. Experimental validations and comparisons are conducted on bearings. The results present that the proposed model is effective for localizing the discriminative information from the input data, which provides a tool for better understanding the feature extraction process in neural networks, especially for mechanical vibration signals. • The attention mechanism is introduced in bearing fault diagnosis. • A deep learning-based bearing fault diagnosis model via attention is proposed. • The results of the model have clearer interpretability compared with other models. • Experiments on bearing datasets verified the effectiveness of the model. [ABSTRACT FROM AUTHOR]

Owing to the rotation and reciprocation of mechanical equipment, their vibration signals inherently exhibit cyclic stationary characteristics. This paper provides a phenomenological signal model of the planetary gear vibrations under the cyclostationarity paradigm. The model is helpful to give comprehensive demodulation information which is the limit of classic spectral analysis of deterministic signals, so that the diagnosis of local faults in planetary gear sets can be more explicit. First, the effects of rotating speed fluctuation on periodic gear meshing vibrations are analyzed. It is shown that the stationary speed fluctuation will turn the periodicity of signals into second-order cyclostationarity. Next, an impulsive oscillation sequence with interval timing jitter is used to describe the dynamic response of the meshing stiffness change caused by local defects. Meanwhile, the influence of a vibration transfer path is also considered in the model. Based on the above analysis, the following procedure is utilized to analyze planetary gearbox vibrations: remove the periodic components by the self-adaptive noise cancellation method and then, apply the second-order frequency descriptor to reveal fault features of the signals. Both the simulation and experiment are used to validate the correctness of model derivation and the superiority of the cyclostationary analysis. • A cyclostationary phenomenological signal model of the planetary set is proposed. • Periodic components are removed by the self-adaptive noise cancellation method. • Modulation features are revealed by the second-order cyclostationary descriptor. • The model is verified by the numerical and experimental analysis. [ABSTRACT FROM AUTHOR]

Speed fluctuations will affect the statistical characteristics of rotating machinery vibrations, resulting in fuzzy fault features that are difficult to extract. This type of feature ambiguity can be interpreted as the vibrations are warped in time, and the related signal model is described as a time-warped almost-cyclostationary (ACS) process. By employing this sophisticated model in a de-warping procedure, the regular statistical properties of the signal are restored. In this paper, we exploit the de-warping method of Napolitano for restoring fault features of rotating machinery under fluctuating speed conditions. Notice that the time-warped ACS process belongs to a subclass of the oscillatory ACS signal model. More specifically, under moderate speed fluctuation conditions, rotating machinery signals are further characterized as modulated cyclical signals. Therefore, the de-warping can be achieved by demodulating the time-varying autocorrelation function. The approach avoids directly estimating the warping function by the complex non-convex optimization problem. Meanwhile, the accuracy of the solution is guaranteed. More clustered features lead to a more efficient diagnosis. In addition, the rotating speed can also be derived from the warping function without using a tachometer. Case studies of the experiments on a conventional bearing and a turbopump bearing with a large DN-value confirm the feasibility of the oscillatory ACS signal model. Furthermore, the superiority of the de-warping method is corroborated by the enhanced fault features in the classical cyclic spectrum analysis. • The regular statistical cyclicity in vibrations of rotating machinery are restored. • The de-warped signals are achieved by demodulating the time-varying autocorrelation function. • The rotating speed of the shaft can also be derived from the warping function without using a tachometer. [ABSTRACT FROM AUTHOR]

Blade crack is one of the most common blade failures of rotating machines which may lead to catastrophic accidents. Although many nonlinear vibrations of breathing crack have been investigated, there are few reports on the axial-bending coupling vibration caused by crack. With the aim of providing physical insight into the mechanisms of axis-bending coupling due to crack, a novel axial-bending coupled breathing crack model (ABCBCM) for the rotating blade is proposed in this study. The proposed ABCBCM is analytically formulated based on the Timoshenko beam theory and Castigliano's principle, and validated by comparing the natural and dynamic characteristics with the finite element model (FEM) and experimental tests. In addition, the effects of boundary conditions (rotational speed), crack parameters (crack depth and crack location) and loading conditions (excitation load) on the vibration characteristics of the cracked blade are investigated. The results indicate that the blade axial response is more sensitive to the nonlinearity caused by the breathing crack than the bending response; the axial-bending coupling caused by the crack changes the equilibrium position of the axial and bending displacement; the phase portrait of axial response and the orbit of axial-bending displacement are effective methods to detect breathing crack; and the axial displacement amplitude ratio and the axial acceleration amplitude ratio also are valuable indicators to estimate the severity of the blade crack. The proposed model can provide valuable insights for fault diagnosis and safety monitoring of cracked blade. • A novel axial-bending coupled breathing crack model (ABCBCM) for the cracked blade with rotational effects is proposed. • The proposed model can simulate the crack-caused axial-bending vibration without additional bending moment. • The phase portrait of axial response and orbit of axial-bending displacement are used to investigate the cracked blade. [ABSTRACT FROM AUTHOR]

Using the sparse and redundant representation model to de-noise vibration signals is an effective method for mechanical fault diagnosis. The selection of the regularization parameter in the sparse de-noising algorithm plays a vital role in reconstructing fault features. The current choices, however, are primarily based on fixed-form parameters derived under the assumption of additive white noise or based on the "trial and error" method. As a result, the extracted de-noised signals often fail to show fault features. To solve this difficulty, a self-adaptive regularization parameter selection strategy is proposed, which directly measures the significance of fault features in sparse de-noised signals. Firstly, the entire solution path is obtained by the computationally efficient least angle regression (LARS) algorithm. Besides, a feature presence probability index (FP) is well designed to measure the significance of fault features in the reconstructed signal envelope demodulation spectrum. Accordingly, with the help of the maximal FP indicator, the optimal regularization parameter and LARS solution is determined. The proposed method is not limited by noise probability distributions; therefore, it shows the superiority of extracting fault features in both the simulation and experimental data analysis. [ABSTRACT FROM AUTHOR]