303 results about "Wavelet packet transformation" patented technology

The invention relates to an extraction method for early failure sensitive characteristics based on ensemble empirical mode decomposition (EEMD) and wavelet packet transform. The extraction method for the early failure sensitive characteristics based on the EEMD and the wavelet packet transform includes the following steps: (1), collected original vibration signals of mechanical and electrical equipment are decomposed according to the EEMD, white noise is added, and intrinsic mode function (IMF) components are obtained through decomposition; (2), the sensitive IMF components closely related to failure are chosen, and other irrelative IMF components are ignored; (3), the sensitive IMF components chosen through step (2) are decomposed in an orthogonal wavelet packet mode, and a wavelet coefficient of each node is obtained; and (4), envelopes are extracted from the obtained wavelet packet coefficients by adoption of the Hilbert transform and the Fourier transform, power spectrums are calculated, the power spectrum corresponding to each wavelet packet coefficient is obtained and serves as the early failure sensitive characteristic , and the sensitive characteristics are automatically obtained. Self-adapting signals can be decomposed, the sensitive characteristics can be convenient to obtain automatically, diagnosis precision and speed are improved, and a mechanical and electrical system can be diagnosed quickly, accurately and stably. The extraction method for the early failure sensitive characteristics based on the EEMD and the wavelet packet transform can be applied to the field of mechanical and electrical equipment failure diagnosis.

The invention discloses a neural network method for diagnosing analog circuit failures which is based on a particle swarm algorithm, and comprises the following steps: imposing an actuating signal to an analog circuit to be tested, measuring an actuating response signal in the testing nodes of the circuit, extracting the candidate signal of failure characteristics by implementing noise elimination and then wavelet packet transformation on the measured actuating response signal, extracting the failure characteristics information by further implementing orthogonal principal component analysis and normalization processing on the candidate signal of failure characteristics, and sending the failure characteristics information as samples to the neural network for implementing classification. The method adopts the particle swarm algorithm instead of a gradient descent method in traditional BP algorithms, thus leading the improved algorithm to be characterized in that the algorithm avoids the local minimum problem and has better generalization performance. The BP neural network method for diagnosing the analog circuit failures which is optimized on the basis of particle swarm can obviously reduce iteration times in the algorithm, improve the precision of network convergence, and improve diagnosis speed and precision.

A method of detecting and isolating at least one rhythmic component from a discrete-time input signal, comprises subjecting the input signal to discrete wavelet packet transform multi-resolution analysis; applying wavelet packet basis selection criteria to the result of the analysis to evaluate rhythmic signal features of the input signal; and isolating at least one rhythmic signal component from the input signal based on the evaluation.

A method for diagnosing analog circuit fault based on vector-valued regularized kernel function approximation, includes steps of: step (1) acquiring a fault response voltage signal of an analog circuit; step (2) carrying out wavelet packet transform on the collected signal, and calculating a wavelet packet coefficient energy value as a characteristic parameter; step (3) utilizing a quantum particle swarm optimization algorithm to optimize a regularization parameter and kernel parameter of vector-valued regularized kernel function approximation, and training a fault diagnosis model; and step (4) utilizing the trained diagnosis model to recognize circuit faults.

The invention discloses network flow-predicting method and device based on wavelet package decomposition and a fuzzy neural network, wherein the method comprises the steps of: collecting historic measured data of network flow; decomposing the original network flow onto wavelets with different time scales by wavelet package conversion; reconstructing flow signals on all time scales so as to lead the data volume of the flow signals to be identical with that of original signals; by the fuzzy neural network, studying and predicting the flow signals decomposed onto all time scales; and summing thepredicted values of the flow signals on all time scales to obtain the predicted value of network flow.

The invention provides a signal identification and classification method. The method comprises the followings steps of: carrying out noise reduction on initial data containing higher noise by utilizing a wavelet transform method, decomposing signals into high-frequency information and low-frequency information in data analysis, carrying out noise cancelling on the signals by adopting a soft thresholding method and then carrying out signal reconstruction; carrying out further decomposition on the high-frequency part which is not detailedly classified by multiscale analysis while inheriting allthe favorable time-frequency localization advantages of the wavelet transform; analyzing the signals within different frequency bands after multi-layered decomposition by utilizing the wavelet packettransform to extract out characteristic information reflecting a system state; transforming the characteristic vectors of input signals into a high-dimensional characteristic space through non-lineartransform and then solving for an optimal linear classification plane in the high-dimensional characteristic space. The invention overcomes the defects of difficult determination of a network structure, low convergence rate, requirement on large quantities of data samples during training, and the like in neural network learning and enables the neural network learning to be with the characteristics of high precision and strong real time in the aspect of practical application of engineering.

The invention discloses a simulating PCB intelligent test system which is based on nerve network, which includes main control PC machine, processor, storage, communication circuit, function signal generator, multi-channel sampling switch matrix, sequential circuit, decoding circuit, A / D converter, sampling retainer, differential amplifier, function signal generator under the control of processor outputs stimulating signal to the stimulating node of the tested circuit, the responding signal of the tested circuit is transmited to the processor by multi-channel sampling circuit, then is transmitted to the main control PC machine by communication circuit, the main control PC machine treats the sampling signal with wavelet packet transform de-noising treatment, and with principal component analysis and normalization processing for obtaining fault feature vector; the fault feature vector is inputted into the trained BP nerve network, the output of the BP nerve network is fault type. The invention can position the PCB test fault to the component level effectively, and improve the scalability of system greatly with the simple and effective test method of CMOS switch array.

The invention discloses an analog circuit fault diagnosis method based on an improved RBF neural network. The analog circuit fault diagnosis method includes the following steps that excitation is exerted on a circuit to be detected, and response signals are processed through improved wavelet packet transformation to extract fault characteristic signals; the extracted candidate characteristic signals are normalized to obtain fault characteristic vectors; the fault characteristic vectors serving as samples are input into the neural network and classified to obtain a result of fault diagnosis. Extraction of the fault characteristic vectors based on wavelet packet transformation is adopted, so that the distinguishability is improved; through normalization and other preprocessing, influences caused by different dimensions and too large numerical value difference on original variables are effectively eliminated; an LMS method in an RBF algorithm is replaced by a genetic optimization algorithm to train parameters of the neural network, so that the performance of the RBF algorithm is improved, an optimizing starting point of a genetic algorithm is set through a K average clustering learning algorithm, the iterations of the algorithm is effectively reduced, errors are reduced, diagnosis speed is increased, and the fault recognition rate is improved.

The invention provides a fatigue driving detection system and method based on EEG identification. The fatigue driving detection system comprises an inertia measurement unit, an EEG collecting unit, a processor unit and an upper computer. The inertia measurement unit is fixed to the midpoint of a steering wheel and used for measuring the rotating angle of the steering wheel. The EEG collecting unit is worn by a tested person and used for collecting EEG signals of the tested person. The upper computer is used for carrying out fatigue driving detection according to the rotation angle, measured by the inertia measurement unit, of the steering wheel and the EEG signals collected by the EEG collecting unit, and sending the detection result to the processor unit. The processor unit is used for receiving the fatigue driving detection result and sending the received detection result to an upper computer control center in a wireless mode through a mobile base station. According to the fatigue driving detection system and method, a cospace mode and wavelet packet conversion are adopted, a fatigue driving detection state evaluation model based on EEG identification is built by collecting the EEG signals of the tested person and steering wheel driving operation information in the driving process, and fatigue driving detection is accurate.

The invention relates to a distribution network fault classification method based on a convolution depth confidence network. The method comprises the steps of firstly collecting the three-phase voltage, zero-sequence voltage and three-phase current of a low-voltage bus of a main transformer and a low-voltage side of the main transformer, and respectively interceptting the signal waveform data of one cycle wave before and after each fault condition as training samples; secondly, carrying out the time-frequency decomposition on the training sample data of step S1 by using the discrete wavelet packet transform, and obtaining the time-frequency matrix, then constructing the pixel matrix of the time-frequency spectrum map, and constructing the time-frequency spectrum map as the input of the subsequent CDBN model; then constructing the CDBN model to train two convolution-constrained Boltzmann machines in unsupervised learning mode, and adding the softmax classifier after the second CRBM to train the network model to effectively extract and automatically classify the fault features, and finally, using the trained model to realize the fault classification of distribution network. The method of the invention can realize accurate fault location.

The invention relates to a multi-robot system fault diagnosis method, comprising the following steps of: 1) obtaining real-time robot motion data; 2) adopting wavelet packet transform to perform feature extraction on the motion data; 3) inputting to-be-diagnosed data subjected to feature extraction into a trained fault diagnosis model, and calculating a corresponding real-time likelihood probability of current robot motion data; and 4) according to a relation between the real-time likelihood probability and a state threshold value, obtaining a current hidden state of the multi-robot system and obtaining a fault diagnosis result. Compared with the prior art, the method has the advantages of high robustness, accurate diagnosis result, wide application range and the like.

The invention discloses a gearbox vibration signal fault feature extraction method based on wavelet entropy and information fusion. The method comprises the following steps: vibration signals are decomposed through wavelet packet transform, and a wavelet coefficient matrix is obtained; wavelet time-frequency entropy (WTF) and wavelet singular value entropy (WS) can be calculated through the wavelet coefficient matrix; then information entropy of a singular value sequence is calculated according to an information entropy formula and is the WS; the WS is subjected to nonlinear transformation through kernel entropy principal component analysis (KECA), and information fusion is realized; after KECA, a first principal component, a second principal component and a third principal component are taken as fault features after fusion. Experiments prove that the method can effectively extract the fault features of a gearbox in a mixed mode, and fault diagnosis can be performed effectively on the gearbox on the basis of the gearbox fault feature extraction method.

The invention discloses a single-phase grounding line selection method for a small-current grounding system, and the method comprises the following steps: 1, collecting the zero sequence voltage of a bus or an outgoing line and the zero sequence current of the outgoing line, and calculating the amplitude of the zero sequence voltage in real time; 2, determining that the bus or the outgoing line has a single-phase grounding fault when the amplitude of the zero sequence voltage is greater than the amplitude of a line selection start voltage, and switching to step 3; 3, extracting the data of zero sequence current and voltage of a cycle before a moment when a fault happens and a cycle after the moment of the fault happens, and carrying out the wavelet packet transformation of the data of transient zero sequence voltage and current; 4, comparing the phases of the transient zero sequence current and voltage after the wavelet packet transformation: determining that a line is not a fault line if the phases are the same, and determining that the line is a fault line if the phases are opposite, thereby completing the grounding line section. The method solves a problem of a higher line selection misjudgment rate caused by that the phases and amplitudes of the zero sequence currents of the fault line and the non-fault line when a conventional arc suppression coil grounding system has a single-phase grounding fault.

The invention discloses an open circuit fault diagnosis method for an NPC three-level photovoltaic inverter. The method comprises the following steps of (1) building an NPC three-level inverter simulation model for simulating a fault condition; (2) collecting phase voltage time domain signals of three phases; (3) performing wavelet packet transform on the phase voltage signals to obtain frequencyband energy after the transform, and taking the frequency band energy after the transform as signal eigenvectors; (4) performing dimension reduction processing on the extracted eigenvectors by using aprincipal component analysis method to obtain low-dimensional data matrix samples, and classifying the samples into two parts including training samples and test samples; (5) establishing a radial basis neural network, training the radial basis neural network by using the training samples, and storing the trained neural network; and (6) testing the trained radial basis neural network by using thetest samples, and analyzing a fault diagnosis result. The method is high in diagnosis speed and high in diagnosis accuracy.

A forwarding looking ground penetrating mine detection apparatus includes a radiation source for irradiating a sample of ground suspected of containing at least one mine with a plurality of frequency swept ground penetrating radar signals. A detector receives target signals backscattered from the ground responsive to the radar signal. The detector includes a time-frequency analyzer which transforms the target signals into a time-frequency image representation (TFR). In a preferred embodiment, the detector can include a wavelet packet transformer (WPT) for extracting time-frequency localized information from the TFR in the form of feature set constructed from a wavelet table. The apparatus can also include a data dimensionality reducer for selecting features to form a feature subset from the feature set, preferably based on reference to a training data set. A multilayer neural network classifier can be based on the feature subset, and be adaptable to the surrounding environment through learning.

The invention discloses a failure diagnosis and health evaluation method based on wavelet power, manifold dimension reduction and dynamic time warping, and aims to improve the feature separability of bearing failure, impeller failure and the mixed failures of a centrifugal pump and realize diagnosis and health evaluation of various states. The method comprises the following steps: firstly, decomposing collected vibration signals of the centrifugal pump into 8 wavelet components by applying wavelet packet conversion; extracting wavelet energy of each component to be taken as a failure feature to obtain an eight-dimensional failure feature vector; then conducting dimension reduction on the eight-dimensional feature by applying a manifold learning method to obtain a three-dimensional feature vector with better separability, simplicity and stability; finally, based on the feature vector, measuring the distance of test data and training data by applying a dynamic time normalization method so as to determine the current failure state and realize failure diagnosis of a bearing. The distance value can also reflect the health degree of the current state, can realize evaluation of the health state of the centrifugal pump, and has the excellent practical engineering application value.

Based on principles of wavelet transform and energy spectrum, information entropy, and modal analysis, characteristic quantities of electropolar voltage, current, displacement and sound signal in spot welding procedure are calculated out. A neural network model created is trained through characteristic quantities and nugget area. Comparing the nugget area calculated from neural network model with the area actual measured determines error value, from which the neural network model is adjusted, is carried out until the error is within required region. The invention provides measured result with high accuracy and small error, suitable to online testing and controlling spot welding resistances of allautals and other metal plates.

The invention discloses an image denoising method based on shearlet transformation and Wiener filtering. The method comprises the following steps of: (1) source image input; (2) symmetric extension; (3) shear transformation; (4) wavelet packet decomposition; (5) Wiener filtering; (6) inverse wavelet packet transformation; (7) inverse shear transformation; (8) inverse symmetry transformation; (9) image fusion; and (10) denoised image output. By the image denoising method, the defect that the anisotropic information of an image cannot be expressed very well by the wavelet transformation in the prior art is overcome; the problem that a denoising effect is non-ideal because coefficients are subjected to the same processing in different directions by using a single threshold is solved; the advantages that the shearlet transformation has multi-directionality, the output of a filter can be adjusted by the Wiener filtering according to the regional variance of the image and the like are utilized; and therefore, detailed information of the image can be analyzed more accurately in high-frequency coefficients in different directions of the image. Finally, a high-quality denoised image is obtained.

The invention discloses an intelligent fault diagnosis method for a rolling bearing. The method comprises the steps of obtaining a vibration test signal of the rolling bearing by utilizing an acceleration sensor; performing decomposition and reconstruction on the vibration signal by adopting wavelet package transform to obtain reconstructed signals of all frequency bands; performing singular value decomposition on the reconstructed signals by applying a singular value decomposition technology, and combining singular values of all the frequency bands into eigenvectors of the rolling bearing; and forming an eigenmatrix by the eigenvectors of the rolling bearing in different states, inputting the eigenmatrix to a PNN neural network for performing training, performing feature extraction on the vibration signal, and inputting the vibration signal to the trained PNN neural network for performing fault mode identification. According to the method, feature information contained in the vibration signal of the rolling bearing can be accurately extracted and is accurately identified and distinguished through the PNN neural network; and the method is high in diagnosis speed, high in precision and high in reliability.

The invention discloses a bearing fault diagnosis and prediction method based on an extended Kalman filtering algorithm, and the method comprises the following steps: 1) employing a full service life cycle vibration signal of a bearing; 2) constructing an AR model through the vibration signal, carrying out the filtering analysis of the vibration signal, and highlighting a signal correlated with a fault; 3) extracting energy information correlated with a wavelet packet coefficient through employing wavelet packet transformation, and constructing a feature character; 4) carrying out the calculation of a mahalanobis distance, constructing health indexes based on the mahalanobis distance, converting the non-negative and non-Gaussian distribution health indexes into Gaussian distribution data through Box-Cox transformation, and determining a related abnormal threshold range through the features of Gaussian distribution and the inverted Box-Cox transformation; 5) carrying out fitting analysis of health index data in a loss period, constructing a degeneration model and a status space model, updating model parameters through employing current data and the extended Kalman filtering algorithm, and predicting the remaining service life of the bearing. The method is higher in prediction precision, and is shorter in consumed time.

Embodiments of the invention compress ultrasound RF data after the receiver beamformer. An efficient compression algorithm is disclosed that incorporates the use of the Discrete Cosine Transform (DCT) and the Discrete Wavelet Packet (DWP) transform, followed by quantization of the wavelet coefficients. The algorithm first processes ultrasound scanlines from the receiver beamformer using a DCT transform. In a low-power configuration, a Hadamard transform may be used instead of the DCT. The output of the DCT processing is DCT coefficients at different frequencies. An optimized wavelet packet transform is then used for each coefficient line. The wavelet filter kernels and the wavelet packet trees can be jointly optimization using a two-step optimization algorithm.

The invention embodiment discloses a fault diagnostic method and diagnostic device of a sensor. The method includes the following steps: receiving output signals of the sensor; utilizing a wavelet packet to analyze the output signals; screening the conversion coefficient of the wavelet packet with the maximum amplitude, so as to reserve the data capable of representing the fault features; performing feature extraction according to the change rate of the component energies of the frequencies of the sensor, so as to obtain feature vectors; and inputting the feature vectors into a specified neural network, so as to obtain the fault type of the sensor, wherein the node number of the input and output layers of the neural network are determined by the dimension numbers of the feature vectors and the fault type numbers of the sensor respectively, and the network weight and the threshold are determined by training through a training sample. The fault diagnostic method and diagnostic device adopt the wavelet packet to perform refinement partition of the dynamic signals of the sensor, and determine the fault type of the sensor according to the output of the neural network, so as to effectively diagnose the catastrophic fault of the sensor.

The invention discloses an expert fault analytical and diagnostic method of a parallel mixed type power quality regulator. The method comprises the following steps that: a voltage signal is collected at a sampling point of a power quality regulator and is used as a fault sample signal; denoising processing is carried out on the collected fault sample signal; wavelet packet transform is utilized to carry out processing on the denoised fault sample signal so as to extract an energy characteristic vector; the extracted energy characteristic vector used as a target input vector, together with a set fault type output vector, is input into a BP neural network to carry out training, so that a fault classifier is established; and fault signals that are collected in real time are input into the fault classifier that compares the fault signals with various fault characteristics in the fault classifier, so that a fault disgnosis result is output. According to the invention, a fault state can be accurately diagnosed, so that losses caused by faults occur on a core component in the power quality regulator can be avoided or reduced.

The invention relates to the technical field of computers, in particular to a transformer sound anomaly detection method based on an improved wavelet packet and deep learning. The method comprises thefollowing steps: A) collecting audio signals of N transformers in different running states; B) performing wavelet packet transformation\ on each audio signal, determining a threshold lambda of the sample entropy by adopting an improved sample entropy, recalculating a wavelet coefficient eta of each component, reconstructing component signals, and obtaining reconstructed audio signals; C) performing short-time Fourier transform to generate a feature image; D) classifying the extracted feature images according to the running state of the transformer; and E) establishing a convolutional neural network model, using the classified feature images for training, and using the trained convolutional neural network model for transformer sound anomaly detection. The transformer sound anomaly detection method has the advantages that noise signals in collected transformer signals can be effectively eliminated, abnormal fault characteristics of the transformer are extracted, and engineers are assisted in transformer fault diagnosis.

The invention discloses a fault diagnosis method for a PWM inverter of a motor drive system. The fault diagnosis method is characterized in that an inverter fault diagnosis model based on both wavelet packet decomposition and an RBF neural network is designed, wavelet packet transformation is utilized to extract a feature vector of a fault signal of the inverter, and the feature vector is taken as the input quantity of the RBF neutral network; a wolf pack-simulated annealing algorithm is adopted to optimize structural parameters of the RBF neutral network; and 22 groups of learning samples and 6 groups of test samples are utilized to train and examine the RBF neutral network. Simulation experiment analysis shows that when the fault diagnosis method is used for an open-circuit fault of the PWM inverter of a three-phase motor drive system, the fault position of an inverter TGBT power tube can be accurately positioned, fault diagnosis is quick, accurate and efficient, and the fault diagnosis method contributes to improving the running reliability of the motor drive system.

A method of detecting and isolating at least one rhythmic component from a discrete-time input signal, comprises subjecting the input signal to discrete wavelet packet transform multi-resolution analysis; applying wavelet packet basis selection criteria to the result of the analysis to evaluate rhythmic signal features of the input signal; and isolating at least one rhythmic signal component from the input signal based on the evaluation.

The invention discloses a turning chatter detection method, and relates to the technical field of detection. In the turning process, the state of a machine tool can be reflected in dynamic cutting force. The turning chatter detection method includes the steps that firstly, an off-line data training model is used, force signals are decomposed to a sixth layer through wavelet packet transformation, energy of each node is worked out, and a 64-dimension feature vector is obtained; dimensionality reduction is conducted on the feature vector through least squares support vector machine-regression feature elimination (LSSVM-RFE), redundancy features are eliminated continuously, optimal features are selected out, and a least squares support vector machine classifier is trained according to the optimal features; and each selected feature corresponds to one wavelet packet node, in the on-line detection process, only a small wavelet packet matrix is needed to decompose force signals to the small wavelet packet nodes selected in the off-line training process, the feature vector is built and input into the classifier, and a detection result is obtained. By the adoption of the dimensionality reduction method, the turning chatter detection method has the beneficial effects of being high in speed and high in identifying accuracy and effectively guaranteeing the machining safety and the product quality.

The invention relates to a wind turbine drive chain vibration noise suppression and incipient fault feature extraction method. The method includes utilizing the experience mode to analyze single measuring point vibration signals, screening intrinsic mode functions capable of characterizing the fault feature frequency, reconstructing and acquiring fault feature signals to highlight the fault features, removing the noise effect on the constructed signals through autocorrelation analysis, combining with the wavelet packet transform feature frequency extraction method, and implementing the noise suppression of the single measuring point vibration signals; on the basis, utilizing the adaptive resonance theory to combine and analyze a plurality of measuring point vibration signal spectrum after the noise suppression treatment, and implementing the extraction of incipient fault features. The method has the advantages that the effects of background white noise and short interference noise can be removed effectively, the weak incipient fault feature frequency can be extracted, and the method can be directly applied to a wind turbine monitoring and fault diagnosis system and implement the diagnosis of wind turbine drive chain mechanical faults.

The invention relates to a fault diagnosis method for a rocking arm gear of a coal cutter. Multi-class fault feature information of rocking arm vibration signals and a fault classification method are utilized to achieve fault diagnosis of a rocking arm gear. Firstly, vibration signals at different measurement points are obtained by an acceleration sensor, the vibration signals are decomposed using wavelet packet transformation, then the time domain vibration signal of each part is obtained after wavelet packet decomposition, the energy coefficient, the gradient coefficient and the skewness coefficient of the time domain vibration signal of each part are extracted and taken as fault feature parameters of the rocking arm gear, and the fault feature parameters are utilized to train fault classification models of a support vector machine, and thus the optimal fault classification model is obtained, and the fault diagnosis of the rocking arm gear of the coal cutter is achieved. The beneficial effects are that problems that fault information of the vibration signal of the rocking arm gear is difficult to extract and is single, and the fault identification rate is low are solved, and the multi-parameter-based fault diagnosis method of the rocking arm gear is enabled to be advantaged by simple calculation, easy implementation, high identification rate and the like.