Showing papers on "Continuous wavelet transform published in 2020"

Bearing defect size assessment using wavelet transform based Deep Convolutional Neural Network (DCNN)

Abstract: Bearing, an importunate component of any rotary machinery, is jeopardized to its failure during its operation in tough working conditions. The condition monitoring of bearing, to avoid its unforeseen failure, is important for its smooth working. Bearing damage assessment is mostly done by selecting features from the vibration signals, which is usually, a time consuming process. Consequently, it becomes importunate for us to achieve full automation for the safety purpose and reduction in the maintenance cost of the machinery. Towards this omnifarious effort, a wavelet transformed based Deep Convolutional Neural Network (DCNN) is proposed for the automatic identification of defective components and damage assessment of bearing, which is achieved by, firstly, processing vibration signals using continuous wavelet transform to form 2D grey scale images of time-frequency representation. Secondly, DCNN is trained using images for learning of defects severity. Through convolution and pooling operation layers, high level features are automatically extracted from images itself. Thereafter, trained 2D grey images are applied to DCNN so that defect severity assessment can be accurately carried out. The overall accuracy achieved using the proposed method is 100%.

Assessment of drought trend and variability in India using wavelet transform

Abstract: This paper presents an analysis of trends in six drought variables at 566 stations across India over the period 1901–2002. Six drought variables were computed using standardized precipitation index...

Convolutional neural networks for microseismic waveform classification and arrival picking

Abstract: Microseismic data have a low signal-to-noise ratio (S/N). Existing waveform classification and arrival-picking methods are not effective enough for noisy microseismic data with low S/N. We ...

Transforming Spectrum and Prosody for Emotional Voice Conversion with Non-Parallel Training Data

Abstract: Emotional voice conversion aims to convert the spectrum and prosody to change the emotional patterns of speech, while preserving the speaker identity and linguistic content. Many studies require parallel speech data between different emotional patterns, which is not practical in real life. Moreover, they often model the conversion of fundamental frequency (F0) with a simple linear transform. As F0 is a key aspect of intonation that is hierarchical in nature, we believe that it is more adequate to model F0 in different temporal scales by using wavelet transform. We propose a CycleGAN network to find an optimal pseudo pair from non-parallel training data by learning forward and inverse mappings simultaneously using adversarial and cycle-consistency losses. We also study the use of continuous wavelet transform (CWT) to decompose F0 into ten temporal scales, that describes speech prosody at different time resolution, for effective F0 conversion. Experimental results show that our proposed framework outperforms the baselines both in objective and subjective evaluations.

Detection of Canopy Chlorophyll Content of Corn Based on Continuous Wavelet Transform Analysis

Abstract: The content of chlorophyll, an important substance for photosynthesis in plants, is an important index used to characterize the photosynthetic rate and nutrient grade of plants. The real-time rapid acquisition of crop chlorophyll content is of great significance for guiding fine management and differentiated fertilization in the field. This study used the method of continuous wavelet transform (CWT) to process the collected visible and near-infrared spectra of a corn canopy. This task was conducted to extract the valuable information in the spectral data and improve the sensitivity of chlorophyll content assessment. First, a Savitzky–Golay filter and standard normal variable processing were applied to the spectral data to eliminate the influence of random noise and limit drift on spectral reflectance. Second, CWT was performed on the spectral reflection curve with 10 frequency scales to obtain the wavelet energy coefficient of the spectral data. The characteristic bands related to chlorophyll content in the spectral data and the wavelet energy coefficients were screened using the maximum correlation coefficient and the local correlation coefficient extrema, respectively. A partial least-square regression model was established. Results showed that the characteristic bands selected via local correlation coefficient extrema in a wavelet energy coefficient created a detection model with optimal accuracy. The determination coefficient (Rc2) of the calibration set was 0.7856, and the root-mean-square error (RMSE) of the calibration set (RMSEC) was 3.0408. The determination coefficient (Rv2) of the validation set is was 0.7364, and the RMSE of the validation set (RMSEV) was 3.3032. Continuous wavelet transform is a process of data dimension enhancement which can effectively extract the sensitive variables from spectral datasets and improve the detection accuracy of models.

Transforming Spectrum and Prosody for Emotional Voice Conversion with Non-Parallel Training Data.

Abstract: Emotional voice conversion aims to convert the spectrum and prosody to change the emotional patterns of speech, while preserving the speaker identity and linguistic content. Many studies require parallel speech data between different emotional patterns, which is not practical in real life. Moreover, they often model the conversion of fundamental frequency (F0) with a simple linear transform. As F0 is a key aspect of intonation that is hierarchical in nature, we believe that it is more adequate to model F0 in different temporal scales by using wavelet transform. We propose a CycleGAN network to find an optimal pseudo pair from non-parallel training data by learning forward and inverse mappings simultaneously using adversarial and cycle-consistency losses. We also study the use of continuous wavelet transform (CWT) to decompose F0 into ten temporal scales, that describes speech prosody at different time resolution, for effective F0 conversion. Experimental results show that our proposed framework outperforms the baselines both in objective and subjective evaluations.

Diagnosis and Classifications of Bearing Faults Using Artificial Neural Network and Support Vector Machine

Abstract: The research paper presents a comparative study of artificial neural network (ANN) and support vector machine (SVM) using continuous wavelet transforms and energy entropy approaches for fault diagnosis and classification of rolling element bearings. An experimental test rig is used to acquire the vibration signals of healthy and faulty bearings. Four real-valued base wavelets are considered. Out of these wavelets, mother base wavelet is selected on behalf of maximum energy and minimum entropy criterions and extracts the statistical features from wavelet coefficient of raw vibration signals. These statistical features are used as input of ANN and SVM for classifying the faults of bearings. Finally, Morlet wavelet is selected on the basis of energy and entropy criterions. The test results show that SVM gives the better fault diagnosis and classification accuracy than ANN.

Wavelet transforms in separation science for denoising and peak overlap detection

Abstract: Wavelet transform is a versatile time-frequency analysis technique, which allows localization of useful signals in time or space and separates them from noise. The detector output from any analytical instrument is mathematically equivalent to a digital image. Signals obtained in chemical separations that vary in time (e.g., high-performance liquid chromatography) or space (e.g., planar chromatography) are amenable to wavelet analysis. This article gives an overview of wavelet analysis, and graphically explains all the relevant concepts. Continuous wavelet transform and discrete wavelet transform concepts are pictorially explained along with their chromatographic applications. An example is shown for qualitative peak overlap detection in a noisy chromatogram using continuous wavelet transform. The concept of signal decomposition, denoising, and then signal reconstruction is graphically discussed for discrete wavelet transform. All the digital filters in chromatographic instruments used today potentially broaden and distort narrow peaks. Finally, a low signal-to-noise ratio chromatogram is denoised using the procedure. Significant gains (>tenfold) in signal-to-noise ratio are shown with wavelet analysis. Peaks that were not initially visible were recovered with good accuracy. Since discrete wavelet transform denoising analysis applies to any detector used in separation science, researchers should strongly consider using wavelets for their research.

Bearing fault identification based on convolutional neural network by different input modes

Abstract: Convolutional neural networks (CNNs) have been applied to the field of fault diagnosis as one of the most widely used deep learning architectures. Different input modes of CNN for bearing fault identification were analyzed by researchers to improve recognition accuracy, such as time-domain diagram, grayscale diagram, short-time Fourier transform diagram, and continuous wavelet transform diagram. However, for the data with small sample size and high background noise, the performance of the CNN is constrained. In this paper, one CNN input mode for bearing fault recognition is proposed based on time-domain color feature diagram (TDCF) through adding red color to diagrams. The method significantly enhanced the fault characteristics of the signal, which is beneficial to the CNN extraction of bearing fault features. Convolution visualization illustrates the effectiveness of the proposed method that provides more bearing fault recognition information. Different sample size and color rate were analyzed by bearing vibration data with high noise. The results showed that the bearing fault identification method based on CNN with 0.4 TDCF obtained a highest fault identification accuracy compared with other input mode methods. The feasibility of the proposed method has been validated, which also provides one reference for other faults identification and pattern recognition.

Time-Synchroextracting General Chirplet Transform for Seismic Time–Frequency Analysis

Abstract: Synchrosqueezing transform (SST) is an effective time–frequency analysis (TFA) approach for the processing of nonstationary signals. The SST shows a satisfactory ability of the TF localization of the nonlinear signal with a slowly time-varying instantaneous frequency (IF). However, for the signal of which ridge curves in the TF domain are fast varying, or even almost parallel to the frequency axis, the SST will provide a blurred TF representation (TFR). To solve this issue, the transient-extracting transform (TET) was recently put forward. The TET can effectively characterize and extract transient features in the much concentrated TFR for the strongly frequency-modulated (FM) signal, especially the impulse-like signal. However, contrary to the SST, it is not suitable for weak FM modes. In this study, we propose a TFA method called the time-synchroextracting general chirplet transform (TEGCT). The TEGCT can achieve a highly concentrated TFR for strong FM signals as well as weak FM ones. Quantized indicators, the concentration measurement and the peak signal-to-noise ratio, are used to analyze the performances of the proposed method compared with those of other methods. The comparisons show that the TEGCT can provide a result with better TF localization. Then, the proposed method was applied to the spectrum analysis of the seismic data for oil reservoir characteristics. The horizontal slices of the offshore 3-D seismic data show that the TEGCT delineates more distinct and continuous subsurface channels in a fluvial-delta deposition system. All the results illustrate that our proposed method is a good potential tool for seismic processing and interpretation in the geoscience.

Seismic Time–Frequency Analysis via Adaptive Mode Separation-Based Wavelet Transform

Abstract: To better reveal time-varying spectral components of nonstationary seismic signals, time–frequency analysis (TFA) has been widely applied in seismic processing and analysis. In this letter, we propose an advanced seismic TFA method based on an optimal spectral mode separation and an adaptive wavelet bank design. The proposed adaptive mode separation-based wavelet transform (AMSWT) generates a superior time–frequency resolution. In addition, because the wavelet bank is adaptively built on the intrinsic spectral modes, the ability to accurately characterize geophysical structures has been significantly improved. To demonstrate the effectiveness of the proposed AMSWT method, we apply it on both synthetic and field data. Compared with the results from continuous wavelet transform (CWT), empirical mode decomposition (EMD), variational mode decomposition (VMD), and empirical wavelet transform (EWT), AMSWT provides a higher resolution and offers potentials in precisely highlighting stratigraphy boundaries.

Discrete-Wavelet-Transform and Stockwell-Transform-Based Statistical Parameters Estimation for Fault Analysis in Grid-Connected Wind Power System

Abstract: Detection and assessment of unbalanced conditions in an early stages are of utmost importance for reliable and smooth operation of a grid-connected wind system. This article presents fault assessment in the grid-connected wind system. For this purpose, the grid-connected wind system has been simulated in MATLAB. All symmetrical and unsymmetrical faults have been considered for three different wind systems. The system current signal has been taken and normalized, then using discrete wavelet transform (DWT)-based statistical parameter analysis, unbalanced conditions have been detected. Total harmonic distortion (THD), interharmonics groups, and Stockwell transform (S-transform) based statistical parameter analysis has also been used for total assessment of unbalanced conditions, like presence of harmonics, classification of faults, etc. This article emphasizes fast detection and classification of all unbalanced conditions of the grid-connected wind system. Then, severity of different unbalanced conditions has been assessed by investigating the presence of harmonics using advanced signal-processing-based approach.

Accurate Demagnetization Faults Detection of Dual-Sided Permanent Magnet Linear Motor Using Enveloping and Time-Domain Energy Analysis

Abstract: In this article, dual-sided permanent magnet linear motors (DPMLM) have been wildly applied in linear motion occasions such as high-precision laser cutting machines. This article researches a new way for accurate demagnetization fault detection of DPMLM, and this proposed method based on signal enveloping and time-domain energy analysis can be suitably used in some typical industrial occasions such as motor batch demagnetization inspection before delivery and periodic maintenance. First, three magnetic signals in motor air-gap region are selected as demagnetization fault index, and finite element analysis (FEA) is used to obtain these signals. Second, complex continuous wavelet transform (CCWT) is introduced to preprocess the fault signal and extract signal envelop for next-step fault feature extraction. Comparison experiments show that CCWT is better than frequently used extreme value and Hilbert–Huang transform methods. Then, Teager–Kaiser energy operator (TKEO) is applied to detect time-domain energy of fault signal envelop as fault feature. In addition to this, Hanning window is used to optimize TKEO to enhance fault feature and help with the accurate detection of demagnetization fault. Finally, motor prototype is manufactured for actual experiment and the results under different noise environments can certify the effectiveness and robustness of this proposed method.

Application of wavelet for seismic wave analysis in Kathmandu Valley after the 2015 Gorkha earthquake, Nepal

Abstract: In this paper, we estimate the seismogenic energy during the Nepal Earthquake (25 April 2015) and studied the ground motion time-frequency characteristics in Kathmandu valley. The idea to analyze time-frequency characteristic of seismogenic energy signal is based on wavelet transform which we employed here. Wavelet transform has been used as a powerful signal analysis tools in various fields like compression, time-frequency analysis, earthquake parameter determination, climate studies, etc. This technique is particularly suitable for non-stationary signal. It is well recognized that the earthquake ground motion is a non-stationary random process. In order to characterize a non-stationary random process, it is required immeasurable samples in the mathematical sense. The wavelet transformation procedures that we follow here helps in random analyses of linear and non-linear structural systems, which are subjected to earthquake ground motion. The manners of seismic ground motion are characterized through wavelet coefficients associated to these signals. Both continuous wavelet transform (CWT) and discrete wavelet transform (DWT) techniques are applied to study ground motion in Kathmandu Valley in horizontal and vertical directions. These techniques help to point out the long-period ground motion with site response. We found that the long-period ground motions have enough power for structural damage. Comparing both the horizontal and the vertical motion, we observed that the most of the high amplitude signals are associated with the vertical motion: the high energy is released in that direction. It is found that the seismic energy is damped soon after the main event; however the period of damping is different. This can be seen on DWT curve where square wavelet coefficient is high at the time of aftershock and the value decrease with time. In other words, it is mostly associated with the arrival of Rayleigh waves. We concluded that long-period ground motions should be studied by earthquake engineers in order to avoid structural damage during the earthquake. Hence, by using wavelet technique we can specify the vulnerability of seismically active region and local topological features out there.

Bridging the gap between the short-time Fourier transform (STFT), wavelets, the constant-Q transform and multi-resolution STFT

Abstract: The short-time Fourier transform (STFT) is extensively used to convert signals from the time-domain into the time–frequency domain. However, the standard STFT has the drawback of having a fixed window size. Recently, we proposed a variant of that transform which fixes the window size in the frequency domain (STFT-FD). In this paper, we revisit that formulation, showing its similarity to existing techniques. Firstly, the formulation is revisited from the point of view of the STFT and some improvements are proposed. Secondly, the continuous wavelet transform (CWT) equation is used to formulate the transform in the continuous time using wavelet theory and to discretize it. Thirdly, the constant-Q transform (CQT) is analyzed showing the similarities in the equations of both transforms, and the differences in terms of how the sweep is carried out are discussed. Fourthly, the analogies with multi-resolution STFT are analyzed. Finally, the representations of a period chirp and an electrocardiogram signal in the time–frequency domain and the time-scale domain are obtained and used to compare the different techniques. The analysis in this paper shows that the proposed transform can be expressed as a variant of STFT, and as an alternative discretization of the CWT. It could also be considered a variant of the CQT and a special case of multi-resolution STFT.

First Arrival Time Identification Using Transfer Learning With Continuous Wavelet Transform Feature Images

Abstract: In our work, the deep learning technique has been used to develop an automatic method for identifying the first arrival times of seismic waves. This method introduces transfer learning to train a deep neural network, given a limited number of continuous wavelet transform (CWT) feature images as input. The application of the CWT for feature extraction, aimed at detecting abrupt changes in the amplitude, phase, and frequency produced by first arrivals as a whole rather than any single characteristic, provides the most informative images. First, we apply the CWT to each seismic trace to obtain the CWT feature images and split them into a set of subimages. Then, a pretrained convolutional neural network (CNN) is fine-tuned with limited labeled subimages. The resulting model can be used to predict probability distributions of noise, first-break, and post first-break. Finally, the first arrival times are extracted from the peaks of the probability distributions. We have tested the performance of the method using vibroseis, dynamite, and air gun shot records, which include various types of seismic waves and noise. More accurate and robust results can be obtained with the proposed method compared with the short-time and long-time average (STA/LTA) algorithm and the adaptive multiband picking algorithm (AMPA).

Spectral feature extraction based on continuous wavelet transform and image segmentation for peak detection

Abstract: Peak detection is a crucial step in spectral signal pre-processing. The accurate locations of characteristic peaks are prerequisite for chemical identification. However, measured spectra inevitably contain both noise and baseline signals. These interference signals will generate a series of false peaks, which is a challenge for spectral analyses. For this purpose, a spectral peak detection algorithm named CWT-IS, based on continuous wavelet transform (CWT) and image segmentation (IS), is proposed. First, a wavelet coefficient matrix is obtained by CWT, then the matrix is converted to a gray image and the image is segmented by a machine vision method, an improved version of the Otsu method based on fuzzy mathematical theory. The desired threshold can be determined based on the membership grades of segmentation performance and the features of false peaks are eliminated. The fuzzy Otsu method solves the problem that the traditional Otsu method cannot effectively handle an image with a unimodal histogram, thus, this method can accurately separate peak regions from the CWT coefficient matrix and ridges representing the peaks are complete. This method has been successfully applied to the peak detection of the simulated spectra and the MALDI-TOF spectra. The experimental results show that CWT-IS can effectively eliminate the adverse effect of noise and baseline and make the features of spectral peaks become more obvious, which is conducive to determine characteristic peak positions.

Analysis of adaptive synchrosqueezing transform with a time-varying parameter

Abstract: The synchrosqueezing transform (SST) was developed recently to separate the components of non-stationary multicomponent signals. The continuous wavelet transform-based SST (WSST) reassigns the scale variable of the continuous wavelet transform of a signal to the frequency variable and sharpens the time-frequency representation. The WSST with a time-varying parameter, called the adaptive WSST, was introduced very recently in the paper “Adaptive synchrosqueezing transform with a time-varying parameter for non-stationary signal separation.” The well-separated conditions of non-stationary multicomponent signals with the adaptive WSST and a method to select the time-varying parameter were proposed in that paper. In addition, simulation experiments in that paper show that the adaptive WSST is very promising in estimating the instantaneous frequency of a multicomponent signal, and in accurate component recovery. However, the theoremretical analysis of the adaptive WSST has not been studied. In this paper, we carry out such analysis and obtain error bounds for component recovery with the adaptive WSST and the 2nd-order adaptive WSST. These results provide a mathematical guarantee to non-stationary multicomponent signal separation with the adaptive WSST.