Feature Extraction Of Fault Signal Research Articles

Complementary Ensemble Empirical Mode Decomposition and Maximum Correlated Kurtosis Deconvolution for Wind Turbine Bearing Fault Feature Extraction

Wind turbine bearing fault signals exhibit characteristics such as nonlinearity, non-stationarity, and susceptibility to external noise interference, making it challenging to extract fault features and identify them accurately. In light of these issues, this paper proposes a fault signal feature extraction method that combines complementary ensemble empirical mode decomposition (CEEMD) and maximum correlated kurtosis deconvolution (MCKD). CEEMD is utilized to decompose the signals, reducing mode mixing and eliminating residual auxiliary noise in the decomposition process, thereby obtaining linear and stationary signals that enhance the significance of the features. MCKD is employed to increase the kurtosis value of the fault signals, facilitating the extraction of continuous transient impacts from weak fault signals. Finally, a Convolutional Neural Network (CNN) is employed to validate that the method enhances fault characteristics and improves the accuracy of fault signal recognition.

Mechanical Fault Feature Extraction under Underdamped Conditions Based on Unsaturated Piecewise Tri-Stable Stochastic Resonance

In the case of the rapid development of large machinery, the research of mechanical fault signal feature extraction is of great significance, it can not only ensure the development of the economy but also ensure safety. Stochastic resonance (SR) is of widespread use in feature extraction of mechanical fault signals due to its excellent signal extraction capability. Compared with an overdamped state, SR in an underdamped state is equivalent to one more filtering of the signal, so the signal-to-noise ratio (SNR) of the output signal will be further improved. In this article, based on the piecewise tri-stable SR (PTSR) obtained from previous studies, the feature extraction of mechanical fault signals is carried out under underdamped conditions, and it is found that the SNR of the output signal is further improved. The simulation signals and experimental signals are used to verify that PTSR has better output performance under underdamped conditions.

Research and application of forward three-dimensional design method of substation

During the operation of digital substations, they are vulnerable to the influence of different fault feature extraction, resulting in weak anti-interference in the detection process. To solve this problem, the singular value decomposition method is combined detection method to establish a prediction model. Firstly, the three-dimensional control mode model is established based on mathematical method to process the current signal with anti-interference; Secondly, the singular value method is used to optimize the current transmission process and denoise the fault signal feature extraction; Finally, the method of singular value decomposition is used to realize fault detection in substation operation.

Fault location of transmission line based on CNN-LSTM double-ended combined model

In the study of transmission line fault location, most of the previous artificial intelligence-based location methods rely heavily on feature extraction of fault signals, which depend on the researcher’s level of analytical understanding of fault characteristics and require some experience. In addition, previous location methods are more sensitive to line parameters, and the machine learning model obtained based on a specific line is not applicable to other lines, which restricts the application of the method. To solve the above problems, this paper proposes a double-ended combined fault location model based on Maximum Mean Discrepancy (MMD), which combined Convolutional Neural Network(CNN) and Long Short-Term Memory(LSTM). First, different transmission lines are categorized by MMD. Second, a double-ended CNN-LSTM combination model is built for similar lines, which autonomously extracts fault features in an end-to-end form, and then the weights of combination model are determined by the Q-learning algorithm. Finally, we obtain the fault distance prediction. Simulation studies show that the CNN-LSTM double-ended combined model based on MMD has good generalization performance for lines with different parameters, cracking the problem of specialized modeling of different lines while meeting the requirement of fault location accuracy.

A novel mechanical fault signal feature extraction method based on unsaturated piecewise tri-stable stochastic resonance

Stochastic resonance (SR) is widely studied in signal feature extraction. The standard tri-stable SR (STSR) method possesses superior performance than the classical bistable SR (CBSR) method in signal feature extraction, but it still has output saturation phenomenon, which will reduce the signal-to-noise ratio (SNR) of the output signal and the signal amplitude of the characteristic frequency as the CBSR method. For the purpose of avoiding the influence of output saturation on the STSR method, a piecewise tri-stable stochastic resonance (PTSR) method is proposed and applied in fault feature signal extraction. Firstly, the simulation signals are processed using the PTSR method and the STSR method separately, then the comparison shows that the output signal has larger signal amplitude and a higher SNR. Ultimately, the devised PTSR method is utilized for extracting fault characteristics of two groups of actual signals, which also has better output characteristics compared with the STSR method.

Construction of a batch-normalized autoencoder network and its application in mechanical intelligent fault diagnosis

Among various fault diagnosis methods, deep learning has shown state-of-the-art performance in processing mechanical big data. This paper investigates a reliable deep learning method known as autoencoder, which is most suitable for automatic feature extraction of fault signals. However, traditional autoencoders have two deficiencies: (1) the multi-layer structure of autoencoder has an internal covariate shift problem, which will cause great difficulty for the network training. (2) The application of autoencoder in the case of rotating speed fluctuation is not mature. To overcome the aforementioned deficiencies, batch normalization strategy is employed in every layer of the autoencoder network to obtain a steady distribution of activation values during training. It can regularize the network without parameter adjustment, and deal with the speed fluctuation problem perfectly. So, a new network named batch-normalized autoencoder is first proposed for intelligent fault diagnosis. The raw vibration signals are directly fed into the network and the extracted features are employed to train a softmax classifier for health state identification. A bearing and a gearbox data set are finally used to confirm the effectiveness of the proposed method. The results manifest that the proposed method can extract salient features from the raw signals and handle the fault diagnosis problem under the speed fluctuation problem.

Rolling element bearing fault diagnosis using convolutional neural network and vibration image

Detecting in prior bearing faults is an essential task of machine health monitoring because bearings are the vital components of rotary machines. The performance of traditional intelligent fault diagnosis methods depend on feature extraction of fault signals, which requires signal processing techniques, expert knowledge, and human labor. Recently, deep learning algorithms have been applied widely in machine health monitoring. With the capacity of automatically learning complex features of input data, deep learning architectures have great potential to overcome drawbacks of traditional intelligent fault diagnosis. This paper proposes a method for diagnosing bearing faults based on a deep structure of convolutional neural network. Using vibration signals directly as input data, the proposed method is an automatic fault diagnosis system which does not require any feature extraction techniques and achieves very high accuracy and robustness under noisy environments.

Generalized Multi-fractal Method for Soft Fault Feature Extraction of Non-linear Analog Circuits Based on EMD

The soft fault signal of non-linear analog circuits is non-linear and unstable. Its inherent characteristics cannot be analyzed by the traditional linear method. Therefore, a new method based on the empirical mode decomposition (EMD) and a generalized multi-fractal dimension algorithm is proposed for soft fault signal feature extraction of non-linear analog circuits. The EMD decomposed the circuits soft fault signal into several intrinsic mode functions (IMF). The EMD then modified a traditional G-P correlation dimension algorithm to simply and reliably use it for calculating the generalized multi-fractal dimensions of the fault signals. According to fractal theory, the generalized fractal dimensions of the original fault signals and IMF can be obtained for an appropriate weight factor q. The principle IMF component can then be filtered using the cross-correlation coefficient criterion. The obtained generalized multi-fractal dimension of the principle IMF component was continually input in the supporting vector machine as a characteristic vector to diagnose the soft fault of the non-linear analog circuits. The feasibility and validity of this proposed method were demonstrated by experiments.

Application in Feature Extraction of AE Signal for Rolling Bearing in EEMD and Cloud Similarity Measurement

Due to the powerful ability of EEMD algorithm in noising, it is usually applied to feature extraction of fault signal of rolling bearing. But the selective correctness of sensitive IMF after decomposition can directly influence the correctness of feature extraction of fault signal. In order to solve the problem, the paper firstly proposes a new method on selecting sensitive IMF based on Cloud Similarity Measurement. By comparing this method in simulation experiment with the traditional mutual information method, it is obvious that the proposed method has overcome the misjudgment in the traditional method and it has higher accuracy, by factually collecting the normal, damage, and fracture fault AE signal of the inner ring of rolling bearing as samples, which will be decomposed by EEMD algorithm in the experiments. It uses Cloud Similarity Measurement to select sensitive IMF which can reflect the fault features. Finally, it sets the Multivariate Multiscale Entropy (MME) of sensitive IMF as the eigenvalue of original signal; then it is classified by the SVM to determine the fault types exactly. The results of the experiments show that the selected sensitive IMF based on Cloud Similarity Measurement is effective; it can help to improve the accuracy of the fault diagnosis and feature extraction.

Adaptive redundant lifting wavelet transform based on fitting for fault feature extraction of roller bearings.

A least square method based on data fitting is proposed to construct a new lifting wavelet, together with the nonlinear idea and redundant algorithm, the adaptive redundant lifting transform based on fitting is firstly stated in this paper. By variable combination selections of basis function, sample number and dimension of basis function, a total of nine wavelets with different characteristics are constructed, which are respectively adopted to perform redundant lifting wavelet transforms on low-frequency approximate signals at each layer. Then the normalized lP norms of the new node-signal obtained through decomposition are calculated to adaptively determine the optimal wavelet for the decomposed approximate signal. Next, the original signal is taken for subsection power spectrum analysis to choose the node-signal for single branch reconstruction and demodulation. Experiment signals and engineering signals are respectively used to verify the above method and the results show that bearing faults can be diagnosed more effectively by the method presented here than by both spectrum analysis and demodulation analysis. Meanwhile, compared with the symmetrical wavelets constructed with Lagrange interpolation algorithm, the asymmetrical wavelets constructed based on data fitting are more suitable in feature extraction of fault signal of roller bearings.