201 results about "Wavelet thresholding" patented technology

The invention discloses a cutter wear state monitoring method based on a deep gated cycle unit neural network. The method comprises the steps that vibration signals generated in the tool machining process are collected in real time through a sensor, after wavelet threshold denoising, the signals are input into a one-dimensional convolutional neural network for single time step time sequence signallocal feature extraction; then, inputting the time series signal into an improved deep gated recurrent unit neural network CABGRUs to carry out time series signal time series feature extraction; an Attention mechanism is introduced to calculate network weights and reasonably distribute the network weights, and finally, signal feature information with different weights is put into a Softmax classifier to classify tool wear states, so that complexity and limitation caused by manual feature extraction are avoided; meanwhile, the problem that a single convolutional neural network ignores correlation before and after a time sequence signal is effectively solved, and the accuracy of the model is improved by introducing an Attention mechanism. Therefore, the method has the characteristic of improving the real-time performance and accuracy of cutter wear state monitoring.

The invention discloses an MEMS (Micro Electronic Mechanical System) gyro data processing method based on wavelet threshold de-noising and an FAR (Finite Automaton Recognizable) model. For solving the problem that large amount of noise exists in the MEMS gyro output signal, the wavelet threshold de-noising method is used to process the gyro output signal to filter the noise and improve the signal-noise ratio; aiming at a static signal formed by de-noising gyro wavelet, a polynomial fitting method is used for the compensation to the gyro deterministic drift, wherein a residual error after the compensation is a random drift of the gyro, that is to say, a sample sequence required by modeling the random drift of the gyro is obtained; and aiming at modeling the random drift of the MEMS gyro at higher precision, the FAR model is used for modeling the random drift. The invention solves the problem of high output noise of the MEMS gyro, improves the signal-noise ratio and can accurately model the random drift of the gyro, thereby improving the output precision of the MEMS gyro.

An improved noise reduction process by wavelet thresholding utilizes a discrete wavelet transform to decompose the image into different resolution levels. A thresholding function is then applied in different resolution levels with different threshold values to eliminate insignificant wavelet coefficients which mainly correspond to the noise in the original image. Finally, an inverse discrete wavelet transform is applied to generate the noise-reduced video image. The threshold values are based on the relationships between the noise standard deviations of different decomposition levels in the wavelet domain and the noise standard deviation of the original image.

The invention discloses an image noise reduction method, overcoming the problem of fixed bias between a wavelet coefficient obtained by an image noise reduction method in the existing image processing technology and a wavelet coefficient of an original image. The image noise reduction method comprises the following steps of carrying out multilevel wavelet decomposition on to-be-processed noise images, thereby acquiring the corresponding multilevel wavelet coefficients; according to the wavelet coefficients at all levels and the corresponding level numbers of the wavelet coefficients, determining the corresponding noise thresholds of the wavelet coefficients at all levels; carrying out noise reduction on the multilevel wavelet coefficients by adopting a corresponding wavelet threshold noise reduction function of multiple noise thresholds based on the multilevel wavelet coefficients; and reconstructing the original images corresponding to the noise images by adopting the noise-reduced multilevel wavelet coefficients. The constant error among the wavelet coefficients obtained by the image noise reduction method and the wavelet coefficients of the original images is relatively small so that pseudo-Gibbs artifacts can be avoided, detailed information of the images are well retained, and the calculated quantity is relatively low. Thus, the image noise reduction method can be widely applied to the wireless broadcasting field.

The invention provides an empirical mode decomposition denoising method based on a arevised wavelet threshold value. The method is characterized by comprising the following steps of first carrying out the empirical mode decomposition on an original signal to acquire a plurality of intrinsic mode functions I with the frequency being gradually reduced and a remainder term; calculating the smoothness of each intrinsic mode function I; calculating a threshold value of each intrinsic function I by utilizing a wavelet threshold value method; revising the threshold value obtained through the wavelet method according to the smoothness and a serial number of each intrinsic mode function I; carrying out the soft threshold value treatment on each intrinsic mode function I by utilizing the revised threshold value to obtain an intrinsic mode function II; finally reconstructing the intrinsic mode function II to obtain a denoised signal. The method is good in self-adaptability, the threshold value calculated by adopting the wavelet threshold value method is revised through the smoothness index, a signal with high signal-to-noise ratio is obtained on the premise of guaranteeing the smoothness, and the method can be used for denoising the ultrasonic signal.

The invention relates to an adaptive redundant lifting wavelet noise reduction analysis-based bearing failure recognition method, belonging to the technical field of failure recognition of a rolling bearing. The technical scheme is that the method comprises the following steps of: (1) performing adaptive redundant lifting wavelet transformation of a bearing vibration signal; (2) performing variable-size threshold noise reduction processing on a high-frequency detail signal obtained by each decomposition process; (3) performing complete reverse reconstruction on a low-frequency approximation signal obtained by final decomposition and a high-frequency detail signal subjected to wavelet threshold noise reduction; and (4) performing Hilbert demodulation processing on a reconstructed signal to obtain an envelope spectrogram of an initial vibration signal, extracting and recognizing a frequency component in the spectrogram, and judging that a bearing fails if frequency conversion or failure characteristic frequency and even corresponding frequency multiplication occurs. The adaptive redundant lifting wavelet noise reduction analysis-based bearing failure recognition method has the beneficial effects that a threshold can be flexibly selected according to the characteristic of change of noise in a wavelet region, so that noise can be filtered better, and meanwhile, the completeness of a real signal can also be guaranteed as much as possible.

A technique for building probabilistic wavelet synopses for multi-measure data sets is provided. In the presence of multiple measures, it is demonstrated that the problem of exact probabilistic coefficient thresholding becomes significantly more complex. An algorithmic formulation for probabilistic multi-measure wavelet thresholding based on the idea of partial-order dynamic programming (PODP) is provided. A fast, greedy approximation algorithm for probabilistic multi-measure thresholding based on the idea of marginal error gains is provided. An empirical study with both synthetic and real-life data sets validated the approach, demonstrating that the algorithms outperform naive approaches based on optimizing individual measures independently and the greedy thresholding scheme provides near-optimal and, at the same time, fast and scalable solutions to the probabilistic wavelet synopsis construction problem.

The invention discloses a hybrid wind power prediction method based on a long-short-term memory neural network. The invention belongs to the technical field of wind power generation power prediction.According to the prediction method, a long-short-term memory neural network is introduced into wind power prediction, firstly, a decision tree method is adopted to carry out feature selection on fan system data, importance sorting is carried out on alternative input variables, feature attributes with the low importance degree are removed, feature attributes with the high importance degree are obtained, and the accuracy of later prediction is improved; wavelet threshold noise reduction is conducted on the selected characteristic attribute values, and original signals are obtained from the signals mixed with the high noise; an LSTM method is adopted to train the hybrid wind power prediction model, and the weight matrix of the prediction model is updated by increasing the number of iterations, so that the prediction precision is improved; and finally, the prediction error of the LSTM model is corrected by using a least square method, thereby further improving the prediction precision of the LSTM hybrid wind power prediction method.

The invention discloses a voice segment detection method based on MFCC similarity of EMD-Wavelet. The method includes: collecting a speaker's voice signal as the source signal; decomposing a noisy voice signal by empirical mode decomposition (EMD) to obtain an all-order intrinsic mode function (IMF); determining the order of a noise-dominated mode IMF by the variance of all-order IMF components' autocorrelation coefficients, conducting wavelet thresholding denoising on the noise-dominated mode IMF, performing reconstruction with the denoised low-order IMF component and remaining high-order IMFcomponent, thus obtaining a de-noised voice signal; calculating the Mel frequency cepstrum coefficient (MFCC) of the voice signal, and adopting Euclidean distance as the measure of the voice signal MFCC similarity; and obviously distinguishing a voice segment and a noise segment from a similarity curve, thus realizing voice segment extraction. Compared with traditional detection methods, the method provided by the invention has better robustness and adaptability, higher voice segment detection accuracy, and can be well applied to voice segment extraction of voice signals.

The invention relates to a CEEMD-improved wavelet threshold denoising-based transformer winding ultrasonic detection three-dimensional imaging method, and belongs to a transformer winding deformationdetection method. The method comprises the following steps: firstly, performing CEEMD decomposition on a target signal to obtain a multi-order IMF component, then calculating a correlation coefficientof the IMF component, performing improved wavelet threshold processing on a high-frequency component with a relatively low correlation coefficient, and finally reconstructing a denoised component, alow-frequency component and a residual component to obtain a denoised signal. According to the method, most noise is suppressed while low-amplitude effective information and high-frequency effective information are reserved, and the denoising effect is ideal. In an ultrasonic detection three-dimensional imaging system, the denoised transformer winding state diagram is better in visual effect and clearer in fault position, and it is shown that the transformer winding state diagram denoising method has a good denoising effect.

The invention discloses an electric energy quality disturbance detection and positioning algorithm. The algorithm comprises the steps of S201, filtering noise of an electric energy quality disturbancesignal by adopting an improved wavelet threshold function, and determining a preset scale K through Fourier transform; S202, solving each intrinsic mode function of the electric energy quality disturbance signal through VMD, and extracting each mode amplitude and frequency characteristic information through Hilbert transform; and S203, realizing effective positioning of start and stop moments ofthe disturbance signal through a singular value decomposition principle. The technical problems that an intrinsic mode function obtained by adopting EMD and EEMD in an existing algorithm is prone to mode aliasing and endpoint effect, and power quality disturbance detection and positioning precision is not accurate enough are solved.

Wavelet thresholding using discrete wavelet transforms is a sophisticated and effective approach for noise reduction. However, usage of integer arithmetic implies that not the full range of input values can be used. A method for selectively reducing noise in a digital signal having a first range of values comprises steps of decomposing the digital signal to a plurality of frequency sub-bands, wherein before, during or after the decomposing the digital signal or at least one sub-band is expanded by one or more bits to a second range of integer values, removing in at least one of the frequency sub-bands values that are below a threshold, re-combining the frequency sub-bands, after removing said values that are below a threshold, into an expanded output signal, and de-expanding the expanded output signal, wherein a signal having the first range of values is obtained.

The invention relates to a method and device for denoising near infrared spectrum by wavelet mid-value. The method comprises the following steps: 1, performing wavelet transform on a spectrum signal to obtain the wavelet coefficient of each layer; 2, performing wavelet threshold denoising on a near infrared spectrum according to set upper and lower thresholds; 3, reconstructing the denoised wavelet coefficients according to invert wavelet transform; 4, performing extremum median filter on the reconstructed spectrum to obtain the denoised spectrum. According to the method and device, the wavelet coefficient of a signal location can be better maintained by combining the intralayer correlation of wavelet transform; the spectrum signal of the extremum location can be effectively maintained by virtue of the improved extremum medium filter, the pulse noise at the extremum location can be filtered. By means of the processing, the signal to noise ratio can be improved, and the robustness of a model can be improved.

The invention discloses an underwater acoustic signal denoising method based on combination of an improved VMD model and an improved wavelet threshold method. The method comprises the following steps: firstly, determining a parameter k in a VMD model based on a kurtosis maximization principle, and determining a secondary penalty term alpha by adopting an improved whale optimization algorithm to obtain an improved VMD model; inputting a source signal into the improved VMD model to obtain a plurality of intrinsic mode functions; and then processing the IMFs by a wavelet threshold method to further remove residual noise, wherein an improved grey wolf optimization algorithm is adopted to optimize a threshold lambda, and finally, the IMFs are reconstructed to obtain a denoised signal. According to the method, signals in a noise environment can be effectively separated, complex ocean noise in an underwater acoustic communication system is suppressed, and the underwater acoustic signal receiving quality and the anti-noise performance are improved.

The invention discloses a parameter wavelet threshold signal denoising method based on an improved artificial bee colony algorithm. The parameter wavelet threshold signal denoising method comprises the steps: firstly obtaining a to-be-denoised signal, carrying out wavelet transformation, and obtaining a wavelet coefficient; designing a new threshold function on the basis of a traditional thresholdfunction, proving the property of the new threshold function through mathematical derivation, and determining threshold parameters to be optimized; improving an original artificial bee colony algorithm; taking a mean square error between the to-be-denoised signal and the denoised signal as a fitness function of the improved artificial bee colony algorithm in S3, and obtaining an optimal thresholdparameter under the condition of obtaining a minimum mean square error; and applying the optimal threshold parameter obtained in the step S4 to the new threshold function in the step S2, performing shrinkage processing on the wavelet coefficient to obtain a new wavelet coefficient, and performing inverse wavelet transform to obtain a denoised signal. According to the parameter wavelet threshold signal denoising method, a smaller mean square error, a higher output signal-to-noise ratio and a larger noise rejection ratio can be obtained.

The invention discloses an underwater acoustic signal denoising method based on self-adaptive window filtering and wavelet threshold optimization. The method comprises the steps: firstly, Gaussian/non-Gaussian pulse noise in an underwater acoustic channel is described by combining S[alpha]S distribution and a normal distribution model; a median filtering method based on a self-adaptive window is designed, the size of the filtering window is corrected according to the number of noise points in the window, and non-Gaussian pulse noise is restrained; then, based on an improved artificial bee colony method GDES-ABC, threshold parameters of a wavelet threshold denoising method are optimized, and the Gaussian noise suppression capacity is improved. According to the method, Gaussian/non-Gaussianpulse noise in a complex underwater acoustic environment can be effectively suppressed, the receiving capability of underwater acoustic communication signals such as 2FSK, QPSK and 16QAM is improved,and a relatively high output signal-to-noise ratio and a relatively high noise suppression ratio are obtained.

The invention relates to a temperature reconstruction method applied to flame light field re-focusing imaging. The invention aims to solve problems that the precision of temperature reconstruction islow because the precision of the existing refocusing image is restricted by the space resolution and that the precision of temperature distribution at a corresponding position inside the flame is lowbecause obviously insufficient blocking of the existing refocusing image. The process is that a light field camera shoots the flame and records the light field imaging; multi-pixel extraction is performed on the light field imaging to obtain a sub-aperture image, a light field refocusing image of the flame is obtained according to the sub-aperture image; nose reduction is performed on the light field refocusing image by applying wavelet threshold transform to obtain a noise reduced image; the noire reduced image is restored by applying a Lucy-Richardson deconvolution method to obtain a restored noise reduced image; and the temperature of the reconstructed flame is obtained. The temperature reconstruction method is applied to the technical field of flame imaging simulation in the process ofhigh-temperature flame temperature reconstruction.

The invention relates to an adaptive enhancement method of brain magnetic resonance volumetric data based on transformed domain HMT model , belonging to the field of medical image processing. The method organically combines wavelet transform and hidden Markov chain. According to the characteristics of non-Gaussian distribution with long tail and high peak value of the probability density functionof single wavelet coefficients, a Gaussian mixture model is established for the randomness of single wavelet coefficients. At the same time, the persistence of wavelet coefficients is described by Hidden Markov Tree (HMT). The wavelet domain Hidden Markov Tree model is established and the EM algorithm is used to solve the model. Using the solution of HMT model, the expectation of wavelet coefficients is estimated in the absence of noise. The inverse wavelet transform of the noise suppressed wavelet coefficients is used to obtain the enhanced MRI volume data. Through subjective and objective evaluation, the wavelet adaptive enhancement method has better visual information fidelity than wavelet threshold enhancement method.

The invention relates to a power harmonic signal denoising method based on an improved wavelet threshold. The method comprises the following steps: collecting an original power harmonic signal; carrying out noise dyeing processing on the original power harmonic signal to obtain a one-dimensional noise dyeing signal; performing five-layer wavelet decomposition on the one-dimensional noise-contaminated signal to obtain a high-frequency wavelet coefficient Wj, k; calculating a threshold value by adopting an improved general method, and performing threshold value quantization processing on a groupof obtained high-frequency wavelet coefficients Wj and k by utilizing the threshold value and an improved wavelet threshold value function to obtain estimated low-frequency wavelet coefficients, namely, high-frequency wavelet coefficients Wj and k from the first layer to the fifth layer after threshold value quantization processing and low-frequency wavelet coefficients of the fifth layer, and performing wavelet inverse transformation; and performing signal reconstruction to obtain a reconstructed signal. According to the improved wavelet threshold function, the problems that a hard thresholdfunction is discontinuous and a soft threshold function is distorted are solved, the improved adaptive threshold quantization rule is used, the signal-to-noise ratio of a power signal is improved, and a better denoising effect is obtained.

The invention provides a belt conveyor fault diagnosis method based on sound signals, which can reduce the labor intensity of inspection personnel and has the characteristics of high detection speed, high real-time performance, high safety and the like. The diagnosis method comprises the following steps: S1, collecting a sound signal of the belt conveyor; S2, carrying out improved wavelet threshold de-noising processing on the collected sound signals; s3, performing MFCC and deep learning feature extraction on the noise-reduced sound signal of the belt conveyor; s4, establishing a support vector machine classification model, and forming a trained SVM model; and S5, putting the extracted feature information data into the trained SVM model to obtain a posterior probability, then carrying out decision-level fusion by utilizing a D-S evidence theory, and finally, matching a fusion output result with the running state of the belt conveyor in the known running state of the SVM, the running state with the highest matching degree with the fusion output result corresponding to the current running state of the belt conveyor, so that the fault diagnosis of the belt conveyor is completed.

The invention discloses a power transmission line icing thickness detection method based on image processing, and the method comprises the steps: employing a preprocessing method based on wavelet threshold denoising to eliminate the fuzzy interference and enhance the boundary contour details based on a power transmission line image obtained through unmanned plane inspection and aerial photography;detecting the boundary contour of the power transmission line region by using an improved Canny operator; and finally, on the basis of boundary contour detection, calculating the icing thickness by comparing the pixel widths of the line area images before and after icing and combining the actual power transmission line diameter. The method for detecting the icing thickness of the power transmission line is completed by a computer image processing technology, can be widely applied to power grid overhead line inspection, assists overhead line field inspection and maintenance operation, reducesthe labor intensity of inspection personnel, reduces dangerous operation in power grid operation, and avoids economic loss.

The invention relates to a method for extracting a weak signal of magnetic resonance spectrum overlap and a device thereof. The method comprises the following steps of: carrying out cycle spinning for an overlapped peak signal, and carrying out orthogonal wavelet transform for a result after every spinning; carrying out threshold processing of a wavelet coefficient for a result of every wavelet transform, and setting the wavelet coefficient of a small peak to be 0; carrying out inverse wavelet transform for the wavelet coefficient after every processing to realize reconstruction for a high and wide peak, wherein the reconstructed high and wide peak does not contain the small peak overlapped on the high and wide peak; carrying out inverse cycle spinning for a result of every inverse transform and averaging, and obtaining a pure high and wide peak component without a small peak component; and in the overlapped peak signal, removing the high and wide peak component, and therefore showing the small peak component. According to the method, an oscillation peak brought by an ordinary wavelet threshold method can be effectively restrained. The method has the characteristics of high accuracy, good repeatability, high arithmetic speed and the like.

A DC electric energy signal denoising method based on an improved wavelet threshold value and correlation detection comprises the steps that firstly, discrete wavelet decomposition is conducted on a signal sampled at a fixed frequency through the multi-resolution analysis theory, and therefore the high-frequency part is separated from the low-frequency part in the signal; then, an improved threshold function is designed to quantify the decomposed wavelet coefficient to reduce the influence of noise signals, and periodic mean filtering is performed on the wavelet coefficient in combination with a correlation detection method, so that the extraction effect of periodic signals is improved; and finally, wavelet inverse transformation is carried out on the reconstructed wavelet coefficient, and useful signals are recovered. Experimental results show that the denoising effect of the new threshold function is better than that of a traditional method, and the mixed algorithm has a better extraction effect on periodic signals.