39results about How to "Overcome Mode Aliasing Phenomenon" patented technology

The invention discloses a rolling bearing fault diagnosing method and equipment based on CEEMDAN and CFSFDP, belonging to the field of fault diagnosis of rotary machines. The method comprises the steps of acquiring vibration signals of a bearing in a normal state and different fault mode states so as to obtain sample points of the vibration signals of different states, decomposing by using CEEMDANto obtain time and frequency domains characteristics of bearing diagnosis and screening out bearing state representation parameters along with the time domain and frequency domain characteristics, dividing the representation parameters into training samples and a test samples, then using a table CFSFDP algorithm as a bearing fault diagnosis model, inputting the training samples into the bearing fault diagnosis model, clustering output results, obtaining clustering amount, clustering center points of each type, and state types corresponding to the clustering center points, and inspecting the trained diagnosis model by using test samples. The method and equipment can identify different bearing fault types and fault degrees accurately and effectively.

The invention relates to a vibration signal feature extraction method and device, a storage medium and computer equipment. The vibration signal feature extraction method comprises the following steps:collecting a vibration signal of target equipment; carrying out noise reduction on the vibration signal, so as to obtain a noise reduction signal; carrying out empirical mode decomposition on the noise reduction signal, so as to obtain an original intrinsic mode function corresponding to the noise reduction signal; acquiring a superposed signal according to the original intrinsic mode function corresponding to the noise reduction signal, and carrying out empirical mode decomposition and superposition removal on the superposed signal, so as to obtain a final intrinsic mode function corresponding to the noise reduction signal; and carrying out Hilbert-Huang transformation on the final intrinsic mode function, so as to obtain a spectrum signature of the vibration signal. According to the vibration signal feature extraction method, the feature extraction accuracy can be increased, and extracted features can well represent the operation states of the target equipment.

The invention discloses a VMD-GRU-based short-term wind speed prediction method. The method is characterized is characterized in that it comprises, acquiring data of wind speeds at the current momentand n moments closest to the current moment to form a time sequence; and preprocessing the time sequence to obtain a plurality of sub-sequences and residual components, respectively inputting each sub-sequence and each residual component into respective corresponding trained GRU models, outputting predicted values by the trained GRU models, and performing post-processing on all the predicted values to obtain future wind speed prediction data at the next moment. According to the method, non-stationary wind speed data is decomposed into sub-sequences and residual components with different frequencies by adopting a variational mode decomposition method, the stability of the sub-sequences and the residual components is good, and better prediction is facilitated; the method has good predictionprecision for the wind speed with strong volatility, randomness and uncertainty, and the operation state of the wind power generation device can be adjusted more reasonably.

The invention discloses a vibration signal frequency characteristic extraction method based on an EEMD technology, a CMF technology and a WPT technology. The method comprises the following steps: A. adopting an EEMD method to decompose a signal to obtain IMF components; B. selecting the IMF components according to a set critical value to prevent the IMF components from being removed; C. combining the relevant IMF components together by a CMF method to maintain the integrity of the signal; D. carrying out wavelet transform and wavelet packet decomposition to obtain the needed frequency information about the IMF components. The signal has continuity at various scales by means of adding Gaussian white noise having amplitude limitation, so that a modal aliasing phenomenon is effectively overcome, and signal extraction accuracy is improved.

The invention relates to an adaptive decoupling method for a modal-aliasing problem in empirical mode decomposition. The adaptive decoupling method comprises the following steps: adding noise to the signal to be decomposed and acquiring a noisy signal; extracting local extreme points from the noisy signal; selecting window extreme points from the local extreme points; using the window extreme points to establish an upper and a lower envelope lines; accumulating current envelope mean values according to the upper and the lower envelope lines; acquiring current residual signal according to the noisy signal and the current envelope mean values; judging whether the number of the current iterative envelope mean values is less than the first threshold value or not, wherein the current residual signal is used as a first intrinsic modal component; judging whether the number of the window extreme points acquired on the current iteration is less than or equal to the stated second threshold value or not, and if so, acquiring the first intrinsic modal component and a trend term, wherein the trend term is acquired through the noisy signal acquired through adding the noise in the signal to be decomposed subtracting the first intrinsic modal component.

The invention relates to an NPC three-level inverter fault diagnosis method based on improved-treelet transformation. The NPC three-level inverter fault diagnosis method comprises the steps of: constructing an NPC three-level inverter circuit simulation model, simulating a fault process, measuring a voltage waveform of a bridge arm and using the voltage waveform as a fault signal; decomposing thefault signal into a plurality of IMFs components; carrying out Hilbert transformation on each IMFs component to obtain time frequency distribution and an amplitude of the IMFs component, and selectingthe first eight IMFs components; fitting out an envelope signal by using envelope analysis, and screening out fault characteristic parameters; carrying out optimization on treelet transformation by aGaussian kernel function, and generating characteristic vector samples which are independent of each other; and dividing sample data into a training set and a test set according to a ratio of 3:7, wherein the training set is used for constructing an SVM classifier model, and the test set is used for actually diagnosing a circuit fault. According to the invention, EEMD is adopted to decompose thefault signal, then Hilbert transformation is combined, and more frequency domain characteristics are collected; and the NPC three-level inverter fault diagnosis method is more suitable for processinga nonlinear and non-stationary signal generated by the three-level inverter circuit fault.0

The embodiment of the invention discloses a method for predicting output power of a photovoltaic power generation system, variation mode decomposition is performed on historical output power data of the photovoltaic power generation system in a preset time period, an extreme learning machine prediction model is built according to a plurality of decomposition components obtained by decomposition and corresponding meteorological data, a prediction result of each decomposition component is calculated according to the extreme learning machine prediction model, and a sum of the prediction results is used as a prediction result of the output power of the photovoltaic power generation system. A variation mode decomposition algorithm has good noise robustness and non-recursiveness, and selection of reasonable parameters can effectively avoid a mode aliasing phenomenon, thereby obtaining a high-accuracy decomposition signal, and facilitating improvement of prediction accuracy; and the characteristics of good generalization performance and fast learning speed of an extreme learning machine can further improve prediction precision and prediction efficiency. In addition, the embodiment of the invention also provides a corresponding realization device, which further enables the method to have practicability, and the device has corresponding advantages.

The invention discloses a method of calculating the travelling wave front of a single-phase grounding fault in a distribution network. A line data collector uploads two voltage cycles before a fault and two voltage cycles after the fault, performs Karrenbauer transform on the voltage travelling wave to get a line mode voltage, performs Hilbert transform on the line mode voltage, carries out derivation, and calculates the modulus. The obtained modulus is de-noised, an effective interval is taken, and then, the arrival time of the travelling wave front is obtained. The method has the advantage of quick and accurate calculation. Through the method, the arrival time of the travelling wave front can be calculated accurately, and conditions can be created for fault location. The method has a very good application prospect.

The invention relates to the technical field of new energy source power generation prediction, in particular to a short-term photovoltaic power prediction method based on EWT-KMPMR (empirical wavelet transform and kernel minimax probability machine classification), comprising the steps of I, screening to obtain a training sample highly related to a day to be predicted; II decomposing to obtain empirical dimensional component and experience wavelet component; III, building a corresponding prediction for each component, wherein output of the component prediction models as prediction results for each component; IV, stacking the component prediction results to obtain a predicted value of solar photovoltaic output power to be predicted; by predicting short-term photographic powder with the method, calculating quantity is reduced, modal mixing is avoided, high prediction precision can be acquired, and the method has certain value and great significance to reasonably and economically scheduling electricity of the power grid.

The invention discloses a power quality disturbance identification method based on an improved HHT algorithm, which comprises the steps of firstly, collecting signals, and segmenting the signals by using an SAX algorithm; secondly, performing HHT analysis on the different segments of segmented signals, and extracting feature values of the different segments of signals; then identifying the power quality disturbance condition by using a decision-making tree according to the extracted feature values; and finally performing iterative HHT analysis on identified normal power signals so as to further judge whether there is minor disturbance or not, and thus achieving complete identification for the power quality disturbance. The power quality disturbance identification method disclosed by the invention based on the improved HHT algorithm solves a problem that it is impossible to identity non-stationary signals the prior art. The signals are segmented by adopting the SAX algorithm, and then different segments of signals are analyzed to further judge whether there is disturbance or not, thereby well solving a problem of mode aliasing occurred with high frequency and small amplitude.

The invention discloses an underground water signal decomposition method capable of eliminating mode aliasing. The method comprises the following steps: firstly, solving median points in all adjacent extreme points in the underground water signal, and directly utilizing the median points to fit a medina point envelope curve; then, solving time intervals among all adjacent extreme points, arranging the time intervals according to a small to large sequence, and taking one value before transform as a current requirement to take a maximum time interval of the mode when a transform rate of the time intervals exceeds a minimum mode dividing point; then, taking the maximum time interval as a boundary to replace a fitting curve part which is greater than the boundary and is contained between a maximum value and a minimum value with an original signal curve, judging whether a mode decomposition terminal condition is met or not ,if the mode decomposition terminal condition is met, finishing the decomposition of one mode signal, and otherwise, returning to the first step to carry out a repeated iterative operation. The underground water signal decomposition method favorably eliminates a mode aliasing phenomenon of EMD (Empirical Mode Decomposition) and EEMD (Ensemble Empirical Mode Decomposition) and can carry out effective and accurate analysis on a detected underground water signal.

The invention discloses a photovoltaic power distribution network reactive voltage prediction method and system based on a recurrent neural network. According to the technical scheme, the method comprises: establishing a voltage prediction framework based on the high-proportion photovoltaic power distribution network; analyzing and processing the reactive voltage historical data of the high-proportion photovoltaic power distribution network, namely analyzing key factors influencing global reactive voltage characteristics, and preprocessing reactive voltage in combination with the existing reactive voltage historical data of the high-proportion photovoltaic power distribution network; and establishing a reactive voltage prediction strategy containing the high-proportion photovoltaic power distribution network: carrying out variational mode decomposition on the processed voltage sequence, decomposing the voltage sequence into a plurality of components with different characteristics, thenrespectively inputting each component into a recurrent neural network, and superposing prediction results of each component to obtain a final prediction value. According to the invention, the electric energy quality is improved, the safety and stability of power grid operation are improved, energy conservation and loss reduction can be realized through reactive compensation and other modes, and the operation economy and reliability are improved.

The invention discloses an AMD-HHT-based time-frequency domain fluid identification method which comprises: performing empirical mode decomposition (EMD); performing Hilbert transform through an IMF component to obtain Hilbert spectrum time frequency; successively extracting the signals of respective frequency components by an AMD method for AMD decomposition; successively subjecting the channel numbers of a seismic signal to loop operation so as to obtain N frequency signals decomposed by AMD. The method solves modal aliasing caused by intermittent signals in the process of EMD, obtains accurate frequency-domain information, solves a problem that HHT cannot recognize non-stationary signals with similar frequency, can detect all characteristic frequencies, accurately extracts low-frequencyinformation, improves the fidelity of seismic signals, and accurately identifies reservoir fluids.

The invention relates to a voltage flicker parameter detection method based on extreme point symmetry mode decomposition. Based on the extreme symmetry mode decomposition theory and algorithm, a voltage flicker parameter detection step based on ESMD extreme point symmetry mode decomposition is given. Firstly, a voltage flicker signal is decomposed into single-frequency amplitude-modulated waves bythe ESMD method, and then the instantaneous amplitude and instantaneous frequency information of each single-frequency amplitude-modulated wave is obtained by the direct interpolation DI method. Theover-fitting and under-fitting of an envelope of the EMD method is overcome, and the modal aliasing phenomenon of the EMD method can be effectively avoided in the detection without adding noise; the defect that the detection result of the EEMD method strongly depends on the selected noise is overcome. The voltage flicker parameter detection method has strong self-decomposing ability in the detection of voltage flicker parameters, high detection precision, good real-time performance, less false mode, small end-end distortion of an amplitude-frequency curve, and small fluctuation.

The invention relates to a magnetocardiogram signal denoising method for improving empirical mode decomposition permutation entropy, which comprises the following steps of: (1) carrying out improved CEEMDAN on a magnetocardiogram signal obtained by measurement based on a superconducting quantum interferometer to obtain a series of intrinsic mode components IMF from high frequency to low frequency; (2) calculating the permutation entropy of each IMF, and quantitatively detecting and retaining effective IMF components containing magnetocardiogram signal features by using the permutation entropy; (3) according to the permutation entropy, determining an IMF component of the noise and an effective IMF component containing magnetocardiogram signal features; and (4) processing the noise IMF component by using a windowing function, removing the noise amount, and retaining the effective amount. According to the method, effective components and noise components of the magnetocardiogram signals can be efficiently separated, and the modal aliasing phenomenon can be effectively overcome. White noise is effectively removed through permutation entropy detection and windowing noise reduction processing, so that interference is reduced, and white noise removal in human body magnetocardiogram signals is achieved.

The invention discloses a CNG (compressed natural gas) compressor rolling bearing fault feature extraction method, which comprises the following steps of: constructing a frequency spectrum trend according to a frequency spectrum of an acquired complex bearing signal, and dividing the frequency spectrum by taking a minimum value point of the trend as a boundary so as to realize self-adaptive decomposition of the signal to obtain sub-bands of the signal. According to the method, the modal aliasing phenomenon is avoided, and it is ensured that excessive invalid components do not appear in the decomposition result. Besides, singular value decomposition is carried out on the signal sub-bands to obtain singular values of all the sub-bands, and then the singular values are selected by using amplitude filtering characteristics of singular value decomposition and combining a time domain negentropy index, so that noise reduction processing is realized. And carrying out envelope demodulation on the denoised sub-bands, extracting a fault characteristic frequency, and finally realizing fault diagnosis of the CNG compressor rolling bearing.

The invention provides a radar radiation source signal separation method based on improved variational mode decomposition. The method comprises the following steps: establishing a radar radiation source signal library of multiple modulation modes; constructing a variational model required by a variational mode decomposition algorithm; extracting the Renyi entropy of the additive hybrid radar signal as a fitness value; calculating optimal parameters of the variational mode decomposition algorithm by applying an artificial bee colony algorithm; decomposing the mixed signal into a virtual multi-channel observation signal through variational mode decomposition; signal reconstruction is realized by means of a singular value decomposition and quick independent component analysis method; extracting a time-frequency domain Renyi entropy of the separated signal as a distinguishing feature; and verifying the signal separation effect by using a support vector machine. According to the method, the additive hybrid radar radiation source signals are separated and recognized, the improved variational mode decomposition method is provided for solving the problems that the number of signals detected by a receiver is large, priori information is little and the recognition effect is poor, quick separation and accurate recognition of the hybrid radar signals are achieved, and a brand new thought is provided for follow-up processing of the hybrid signals.

The invention relates to an accurate signal sorting method and belongs to the field of signal sorting technology. The accurate signal sorting method has good stability and flexibility, and comprises the steps of 1), obtaining a maximum value point and a minimal value point of signal x (T); 2, conducting the interpolation respectively for the sequences of the maximum value point and the sequences of the minimal value point by using a three-spline function method to obtain an upper envelope line e(t)+ and a lower envelope line e(t)- of signals; 3), calculating the average envelope line and extracting the details d(t)=x(t)-m(t) of the signals; 4), determining whether d(t) meets the two conditions of IMF or not, if not, recording as x(t)=d(t), repeating the steps from 1) to 4) until d(t) meets the two conditions of IMF, wherein d(t) is an IMF and denoted as 1imf.

The invention discloses an ionized layer forecasting method based on VMD and an Elman neural network. A variational mode decomposition method and the Elman neural network are combined to carry out ionized layer TEC forecasting modeling. According to the method, the variation modal decomposition (VMD) and the Elman neural network are combined to perform ionospheric TEC prediction modeling. Considering the characteristics of nonlinearity, non-stationarity and the like of an ionized layer TEC sequence, the VMD algorithm can effectively reduce the complexity of an original sequence and does not generate a modal aliasing phenomenon when TEC original sequence preprocessing is carried out. Compared with a common EMD algorithm, the VMD can effectively avoid the modal aliasing phenomenon in the data preprocessing process, and the filtering and noise reduction performance is more excellent. According to the method, a VMD method is introduced to decompose an ionized layer TEC sequence to obtain acorresponding intrinsic mode component (IMF), and an input value with relatively high quality is provided for a subsequent prediction model.