Search

This article uses acoustic emission (AE) analysis to diagnose an industrial-scale and slow-speed wind turbine blade bearing. The main challenge for AE analysis is that the fault signals are mingled with heavy noise. As a result, the objective of this article is to filter the raw AE signals and extract weak fault signals. To achieve this goal, a general linear and nonlinear auto-regressive (GLNAR) model is first developed to exploit the nonlinear characteristics of the AE signals. Then, the sparse augmented Lagrangian (SAL) algorithm is applied to learn the built GLNAR model and filter the raw AE signals. The characteristics of SAL is that it is a novel sparse representation technique which can convert the original filtering problem into a number of suboptimization problems, and these subproblems can be solved separately. Finally, as the blade bearing rotates at fluctuating speeds, the filtered signals are resampled so that the bearing fault type can be diagnosed in the order domain. The proposed diagnostic framework was validated in several experiments under time-varying low-speed and heavy-blade-load conditions. The results indicate that our proposed methods are effective and accurate. [ABSTRACT FROM AUTHOR]

Motor fault diagnosis is imperative to enhance the reliability and security of industrial systems. However, since motors are often operated under nonstationary conditions, the high complexity of vibration signals raises notable difficulties for fault diagnosis. Therefore, considering the special physical characteristics of motor signals under nonstationary conditions, in this article, we propose a multiscale kernel based residual convolutional neural network (CNN) for motor fault diagnosis. Our contributions mainly fall into two aspects. First, we notice that each motor fault category has various patterns in vibration signals due to the changing operational conditions of the motor. To capture these patterns, a multiscale kernel algorithm is applied in the CNN architecture. Second, since the motor vibration signals are made up of many different components from different transfer paths, they are very complex and variable. To enable the architecture to extract fault features from deep and hierarchical representation spaces, sufficient depth of the network is needed, which will lead to the degradation problem. In the proposed method, residual learning is embedded into the multiscale kernel CNN to avoid performance degradation and build a deeper network. To validate the effectiveness of the proposed networks, a normal motor and five motors with different failures are tested. The results and comparisons with state-of-the-art methods highlight the superiority of the proposed method. [ABSTRACT FROM AUTHOR]

Fault diagnosis and remaining useful life (RUL) prediction are always two major issues in modern industrial systems, which are usually regarded as two separated tasks to make the problem easier but ignore the fact that there are certain information of these two tasks that can be shared to improve the performance. Therefore, to capture common features between different relative problems, a joint-loss convolutional neural network (JL-CNN) architecture is proposed in this paper, which can implement bearing fault recognition and RUL prediction in parallel by sharing the parameters and partial networks, meanwhile keeping the output layers of different tasks. The JL-CNN is constructed based on a CNN, which is a widely used deep learning method because of its powerful feature extraction ability. During optimization phase, a JL function is designed to enable the proposed approach to learn the diagnosis–prognosis features and improve generalization while reducing the overfitting risk and computation cost. Moreover, because the information behind the signals of different problems has been shared and exploited deeper, the generalization and the accuracy of results can also be improved. Finally, the effectiveness of the JL-CNN method is validated by run-to-failure dataset. Compared with support vector regression and traditional CNN, the mean-square-error of the proposed method decreases 82.7 $\%$ and 24.9 $\%$ , respectively. Therefore, results and comparisons show that the proposed method can be applied for the intercrossed applications between fault diagnosis and RUL prediction. [ABSTRACT FROM AUTHOR]

Fault diagnosis under time-varying operational conditions is always the challenge in industrial systems. The chirplet transform (CT) provides an effective first-order approximation approach, but the slow operation speed and the drawback for high-order signal analysis limit its uses. Focusing on this problem, a fast nonlinear chirplet dictionary-based sparse decomposition (FNC-SD) method for nonlinear signal analysis is proposed. By replacing the time–frequency inclination parameter in the CT with a frequency bending parameter, the proposed method can track the nonstationary signals by arbitrary order polynomial law with time by an additional degree. Considering that too many parameters will slow down the operation speed, a frequency bending operator estimation algorithm is also proposed in this paper based on the calculated rotating frequency. Therefore, a laser vibrometer is needed in the experiment to collect the impulse signals, which can represent rotating speed signals. Then, the FNC-SD method is used to construct a redundant dictionary for signal sparse time–frequency representation. The high similarity between signals and atoms can make the method prefer to choose the right atoms, instead of being influenced by noise. Therefore, the proposed method also shows a strong noise robustness. Two simulation experiments are constructed to verify the performance of the proposed method. Finally, the FNC-SD method is used to diagnose a faulty motor with broken bar. The analyzed results and comparisons with respect to the state-of-the-art methods are illustrated in detail, as well as the convergence accuracy and speed, which highlight the superiority of the proposed method. [ABSTRACT FROM AUTHOR]

As wind power attracts increasing attention and wind turbines (WTs) capacity expands, fault diagnosis of WT is playing a more and more important role in improving reliability, minimizing down time, reducing maintenance costs, and providing reliable power generation. In this paper, a novel sparse time-frequency representation (STFR) method is proposed to increase the diagnostic precision of incipient faults. The proposed method can be applied once the condition is detected as abnormal according to the VDI3834 vibration threshold standard in WT fault diagnosis systems. The proposed method is a novel signal representation method based on the sparse representation theory and Wigner–Ville distribution (WVD), which can overcome the limitations of traditional basis functions expansion and time-frequency analysis methods. In this method, a union of redundant dictionary (URD) is constructed on the basis of the underlying prior information of the oscillate characteristics with multicomponent coupling effect and different morphological waveforms. Therefore, the vibration signal can be sparsely represented over the URD. Then, the sparse coefficients and corresponding atoms can be obtained by solving the basis pursuit denoising problem via alternating direction method of multipliers. Based on the combination of the WVD of each atom and corresponding sparse coefficient, the time-frequency distribution of the vibration signal can be obtained. To verify the effectiveness of the STFR method, a simulation and two field tests in the wind farm are performed. The comparison results with state-of-the-art methods illustrate the superiority and robustness of the proposed method in the engineering applications. [ABSTRACT FROM AUTHOR]

Fault diagnosis of rotating machinery plays a significant role for the reliability and safety of modern industrial systems. As an emerging field in industrial applications and an effective solution for fault recognition, artificial intelligence (AI) techniques have been receiving increasing attention from academia and industry. However, great challenges are met by the AI methods under the different real operating conditions. This paper attempts to present a comprehensive review of AI algorithms in rotating machinery fault diagnosis, from both the views of theory background and industrial applications. A brief introduction of different AI algorithms is presented first, including the following methods: k -nearest neighbour, naive Bayes, support vector machine, artificial neural network and deep learning. Then, a broad literature survey of these AI algorithms in industrial applications is given. Finally, the advantages, limitations, practical implications of different AI algorithms, as well as some new research trends, are discussed. [ABSTRACT FROM AUTHOR]

Bearing plays an essential role in the performance of mechanical system and fault diagnosis of mechanical system is inseparably related to the diagnosis of the bearings. However, it is a challenge to detect weak fault from the complex and non-stationary vibration signals with a large amount of noise, especially at the early stage. To improve the anti-noise ability and detect incipient fault, a novel fault detection method based on a short-time matching method and Support Vector Machine (SVM) is proposed. In this paper, the mechanism of roller bearing is discussed and the impact time frequency dictionary is constructed targeting the multi-component characteristics and fault feature of roller bearing fault vibration signals. Then, a short-time matching method is described and the simulation results show the excellent feature extraction effects in extremely low signal-to-noise ratio (SNR). After extracting the most relevance atoms as features, SVM was trained for fault recognition. Finally, the practical bearing experiments indicate that the proposed method is more effective and efficient than the traditional methods in weak impact signal oscillatory characters extraction and incipient fault diagnosis. [ABSTRACT FROM AUTHOR]

Machine fault diagnosis collects massive amounts of vibration data about complex mechanical systems. Performing feature detection from these data sets has already led to a major challenge. Compressive sensing theory is a new sampling framework that provides an alternative to the well-known Shannon sampling theory. This theory enables the recovery of sparse or compressible signals from a small set of nonadaptive linear measurements. However, it is suboptimal to recover the whole signals from the compressive measurements and then solve feature identification problems through traditional DSP techniques. Thus, a novel mechanical feature identification method is proposed in this paper. Its main advantage is that fault features are extracted directly in the compressive measurement domain without sacrificing accuracy, while a significant reduction in the dimensionality of the measurement data is achieved. Moreover, Gaussian white noises are significantly alleviated, which dramatically enhances the reliability of machine fault diagnosis. Parameter analysis is also profoundly investigated through a set of numerical experiments. Numerical simulations and experiments are further performed to prove the reliability and effectiveness of the proposed method. [ABSTRACT FROM AUTHOR]

It is a fundamental task in the machine fault diagnosis community to detect impulsive signatures generated by the localized faults of bearings. The main goal of this paper is to exploit the low-rank physical structure of periodic impulsive features and further establish a weighted low-rank sparse model for bearing fault detection. The proposed model mainly consists of three basic components: an adaptive partition window, a nuclear norm regularization and a weighted sequence. Firstly, due to the periodic repetition mechanism of impulsive feature, an adaptive partition window could be designed to transform the impulsive feature into a data matrix. The highlight of partition window is to accumulate all local feature information and align them. Then, all columns of the data matrix share similar waveforms and a core physical phenomenon arises, i.e., these singular values of the data matrix demonstrates a sparse distribution pattern. Therefore, a nuclear norm regularization is enforced to capture that sparse prior. However, the nuclear norm regularization treats all singular values equally and thus ignores one basic fact that larger singular values have more information volume of impulsive features and should be preserved as much as possible. Therefore, a weighted sequence with adaptively tuning weights inversely proportional to singular amplitude is adopted to guarantee the distribution consistence of large singular values. On the other hand, the proposed model is difficult to solve due to its non-convexity and thus a new algorithm is developed to search one satisfying stationary solution through alternatively implementing one proximal operator operation and least-square fitting. Moreover, the sensitivity analysis and selection principles of algorithmic parameters are comprehensively investigated through a set of numerical experiments, which shows that the proposed method is robust and only has a few adjustable parameters. Lastly, the proposed model is applied to the wind turbine (WT) bearing fault detection and its effectiveness is sufficiently verified. Compared with the current popular bearing fault diagnosis techniques, wavelet analysis and spectral kurtosis, our model achieves a higher diagnostic accuracy. [ABSTRACT FROM AUTHOR]