Papers by Keyword: Wavelet Packet Decomposition

Abstract: Residual stress has a significant influence on mechanical strength of a manufactured part and is considered to be related with process parameters and grinding signals. This paper investigates the relationship between residual stress and forces in the grinding of maraging steel 3J33. Features in time and frequency domains are extracted from tangential and normal grinding forces via various signal processing techniques. A two-round selection based on the statistical criterion is proposed to choose the best features that are related to the residual stress in the surface layer. The selected features are combined linearly in order to develop an empirical regression model that is capable of predicting residual stress well. The predicted residual stress values are compared with those measured from the experiment performed under different process parameters, and the result shows a favorable agreement quantitatively.

Abstract: In this paper, a method of wavelet packet decomposition and multi-frequency band analysis is considered to study the amplitude attenuation characteristics of stress wave propagation in a viscoelastic polycarbonate (PC) rod. The reconstruction coefficients of different frequency bands could be acquired through wavelet packet decomposition and reconstruction. Associate conjugate gradient method with modified Sadaovsk formula to fit the curve of peak amplitude for each frequency band, from which the amplitude attenuation characteristics and attenuation rates of each frequency band with the increasing of the propagation distance were obtained. Predicted the waveform in the light of viscoelastic attenuation behavior of different frequency bands, it gives better mean square error (MSE) compared with the actual waveform. This result indicated that this method can effectively predict the propagation disciplinarian of the viscoelastic stress wave.

Abstract: Through the mechanism of the gearbox’s vibration signal and establish the corresponding mathematical model, then establish a fault diagnosis method based on the wavelet theory and Hilbert demodulation spectrum. First, the wavelet threshold de-noising can be used to reducing noise of the gearbox’s vibration signal. Then, use the wavelet packet decomposition to decomposing the de-noising signal into different frequency band. After that, use the Hilbert transform to demodulate the frequency band that focused power. Finally, extract the fault characteristic value for the fault diagnosis. Through a fault simulation vibration signal test the method, the results show that the method can effectively extract the fault information of the wind turbine gearbox.

Abstract: Intelligent predictive maintenance (IPdM) is a maintenance strategy that makes maintenance decisions automatically and dynamically based on Artificial Intelligence and Data mining techniques through condition monitoring of machines, equipment and production processes. IPdM system consists of the following main modules: sensor and data acquisition, signal and data processing, feature extractions, maintenance decision-making, key performance indicators, maintenance scheduling optimization and feedback control and compensation. Among them, the most important part of IPdM is maintenance decision-making, which includes diagnostics and prognostics. This paper proposes a framework of intelligent faults diagnosis and prognosis system (IFDaPS) and discuss some key technologies for implement IPdM policy in manufacturing and industries. A case study focus on the vibration signals collected from the sensors mounted on a pressure blower for critical components monitoring. We decompose the pre-processed signals into several signals using Wavelet Packet Decomposition (WPD), and then the signals are transformed to frequency domain using Fast Fourier Transform (FFT). The features extracted from frequency domain are used to train Artificial Neural Network (ANN). Trained ANN model is able to identify the fault of the components and predict its Remaining Useful Life (RUL). The case study demonstrates how to implement the proposed framework and intelligent technologies for IPdM and the result indicates its higher efficiency and effectiveness comparing to traditional methods.

Abstract: Port surveillance videos are degraded seriously in foggy conditions. This paper presented a clearness algorithm based on wavelet packet decomposition. Firstly, we extracted the background image from degraded videos and established the updated model; Secondly, we detected the moving objects as foreground images; Thirdly, we defogged these images based on wavelet packet decomposition; Finally, we fused the background and foreground images together. The experimental results show that our method is more effective.

Abstract: A noveol neural network of Elman is typically dynamic recurrent neural network. A novel method of flow regime identification based on Elman neural network and wavelet packet decomposition is proposed in this paper. Above all, the collected pressure-difference fluctuation signals are decomposed by the four-layer wavelet packet, and the decomposed signals in various frequency bands are obtained within the frequency domain. Then the wavelet packet energy eigenvectors of flow regimes are established. At last the wavelet packet energy eigenvectors are input into Elman neural network and flow regime intelligent identification can be performed.

Abstract: Engine mechanical fault usually causes abnormal change of the body surface vibration signal. Cylinder surface vibration signals under normal condition, piston knocking fault condition and main bearing wear fault condition are analyzed with wavelet packet decomposition method, relative energy value of each frequency band can be calculated and then be regarded as the input vector to form the training sample, BP neural network model is used to identify the fault state, test data shows that this method can effectively recognize the fault types.

Abstract: The wavelet packet decomposition can be used to extract the frequency band containing bearing fault feature, because the fault signal can be decomposed into different frequency bands by using the wavelet packet decomposition, that is to say the optimal wavelet packet decomposition node needs to be found. A method applying the average Euclidean distance to find the optimal wavelet packet decomposition node was presented. First of all, the bearing fault signals were decomposed into three layers wavelet coefficients by which the bearing fault signals were reconstructed. The peak values extracted from the reconstructing signal spectrum constructed a feature space. Then, the minimum average Euclidean distance calculated from the feature space indicated the optimal wavelet packet node. The optimal feature space could be constructed by the feature points extracted from the signals reconstructed by the optimal wavelet packet nodes. Finally, the optimal feature space was used for the K-means clustering. The feature extraction and pattern recognition test of the four kinds of bearing conditions under four kinds of rotation speeds was detailed. The test results show this method, which can extract the bearing fault feature efficiently and make the fault feature space have the lowest within-class scatter, wons a high pattern recognition accuracy.

Abstract: In this paper, a new Intelligent Identification method based on wavelet packet decomposition and APSO-SVM was put forward. As is known the characteristic of pressure drop is nonlinear and non-stationary. The wavelet packet transform can decompose signals to different frequency bands according to any time frequency resolution ratio, the features are extracted from the differential pressure fluctuation signals of the air-water two-phase flow in the horizontal pipe and the wavelet packet energy features of various flow regimes are obtained. The adaptive particle swarm ptimization support vector machine was trained using these eigenvectors as flow regime samples, and the flow regime intelligent identification was realized. The test results show the wavelet packet energy features can excellently reflect the difference between four typical flow regimes, and successful training the support vector machine can quickly and accurately identify four typical flow regimes. So a new way to identify flow regime by soft sensing is proposed.

Abstract: In this paper, shaft monitoring data in condition monitoring system of hydropower units was used to build the fault classification model based on the least square support vector machine (LS-SVM). By the wavelet packet signal decomposition for unit vibration signal, setting the signal energy components as the study sample, learning of fault diagnosis classifier was conducted, to achieve the diagnosis of common faults in shaft running of hydropower unit.