114 results about "Permutation entropy" patented technology

The invention discloses a rolling bearing fault diagnosis method based on composite multi-scale permutation entropies, and belongs to the technical field of fault diagnosis. The method comprises the following steps: measuring a vibration signal of a faulty object; extracting composite multi-scale permutation entropies from the vibration signal; reducing the dimension of the composite multi-scale permutation entropies with use of a Laplacian score; taking the first multiple composite multi-scale permutation entropies with low scores after dimension reduction as fault feature vectors and dividing the fault feature vectors into multiple training samples and multiple test samples; inputting the multiple training samples into a multi-fault classifier established based on a support vector machine to perform learning so as to classify the test samples; and identifying the working mode and the fault type of the faulty object according to the classifying result. According to the fault diagnosis method disclosed by the invention, feature extraction is highly innovative, and the degree of identification is high in the process of fault mode identification.

The invention relates to a vibration fault diagnosis method of a hydroelectric generating set. The method comprises the steps of: firstly, carrying out denoising processing on collected original vibration signals by utilizing a random resonance technology; secondly, carrying out characteristic vector extraction on the vibration signals after the denoising by utilizing a multi-dimensional permutation entropy technology; thirdly, establishing fault diagnosis models of optimized support vector machine based on an improved particle swarm algorithm; fourthly, inputting extracted characteristic vectors into the models of the models of optimized support vector machine based on the improved particle swarm algorithm for fault diagnosis. The method is applicable to the vibration fault diagnosis of the hydroelectric generating set, the diagnosis result is high in precision, the fault type of the generating set can be relatively accurately diagnosed, the reliable diagnosis result is provided to operation maintenance personnel of the generation set, the maintenance personnel can process faults timely and rapidly, and the safety and the economic performance of the operation of the generating set are ensured.

The invention provides a rolling bearing fault diagnosis method based on improved multi-scale amplitude perceived permutation entropy, relates to the field of digital signal processing, and aims at solving the problems of low separability of feature extraction, low accuracy of fault recognition and insufficient analysis of fault severity in the existing fault diagnosis method of the rolling bearing vibration signal. The method comprises the following steps: step 1. obtaining vibration signal sample sets of the rolling bearing under different fault types and different fault degrees; step 2. obtaining the optimal PR component for subsequent feature extraction; step 3. obtaining fault feature vectors of different fault types and different fault degrees; step 4. inputting the feature vector into the random forest classifier; and step 5. obtaining the fault type and the fault severity of the rolling bear. The extracted feature vector has good separability and high fault description ability,and the average recognition accuracy rate reaches 99.25%. The method can be widely applied to the field of bearing fault diagnosis.

The invention provides a partial discharge type identification method based on synchrosqueezing wavelet transform, and aims to provide an identification method for the partial discharge signal of a transformer to realize the identification of the partial discharge signal of the transformer. The method comprises first, decomposing the typical partial discharge signal of the transformer by using thesynchrosqueezing wavelet transform; then, using a multi-scale permutation entropy as a feature quantity of the discharge type identification by using the energy and complexity difference of the partial discharge signal at different decomposition scales; and finally, identifying a discharge type by using a support vector machine classifier based on the extracted feature quantity. The method can achieve average partial discharge signal identification accuracy higher than 90%, and is obviously superior to other commonly used transformer partial discharge identification methods.

The invention relates to a seismic signal random noise suppression processing method. The method includes the following steps that: original noisy seismic signal s(t) are decomposed by using an improved noise-adaptive complete set empirical modal decomposition method, so that a finite number of IMF components and residual components are obtained; an energy demarcation point l of effective signalsand noises in the IMF components of each order is determined according to a permutation entropy theory; and singular value decomposition and noise reduction processing is performed on the high-frequency IMF components of the (1-l)-th order with more noises; and the high-frequency IMF components which have been subjected to the secondary noise reduction processing, low-frequency IMF components which are not processed, and the residual components are accumulated and reconstructed, so that de-noised seismic signals can be obtained. According to the seismic signal random noise suppression processing method of the invention, a multi-step joint processing technology is adopted, and therefore, a seismic noise suppression processing effect can be effectively improved, the details of the lineups ofa seismic channel set can be improved, and conditions can be provided for subsequent seismic data processing, interpretation and forward and inverse calculation.

The invention provides a lower limb motion mode identification method integrated with surface electromyography and acceleration signals. The method comprises steps that firstly, the surface electromyography and acceleration signals of a lower limb of a human body are acquired; the surface electromyography signal is decomposed through utilizing a local mean decomposition algorithm to acquire multiple product functions, according to the average Euclidean distance representing separation of different motions, the multi-scale permutation entropy of the first product function after decomposition through the local mean decomposition algorithm is determined, and the multi-scale permutation entropy of the first product function is extracted as surface EMG signal characteristics; importance of theentropy at different scales is calculated, the scale entropy is determined to form a 4-dimensional characteristic vector, and the 4-dimensional characteristic vector and a triaxial acceleration sequence entropy form a 7-dimensional characteristic vector; the 7-dimensional characteristic vector is inputted to a binary tree support vector machine improved according to the intra-class average Euclidean distance and inter-class sample distribution to carry out lower limb motion mode identification. The method is advantaged in that human body lower limb motion intention can be accurately identifiedin real time, and the identification result can be utilized for exoskeleton robot interaction control.

The invention relates to an abnormal public place sound extraction and recognition method for improving extreme-point symmetric mode decomposition (ESMD), abbreviated as D-ESMD. The method is characterized in that random T distribution sequence signals are added to abnormal public place sound, thereby reducing the influence of public place background noise on abnormal sound feature extraction; aiming at the problem of poor decomposition effect of original ESMD when abnormal sound is decomposed, a method for replacing extreme midpoint odd-even interpolation with symmetric midpoint interpolation is put forward to improve the decomposition efficiency and recognition rate of abnormal sound; aiming at the defects of original ESMD on effective decomposition mode selection, a permutation entropy algorithm is put forward to detect the complexity of the mode obtained by ESMD decomposition, so that an effective mode component of abnormal sound is adaptively obtained. By using the method, the feature of abnormal sound can be sufficiently described, good class recognition results can be obtained, and the feature of abnormal sound can be accurately extracted; and the method has good robustness on environmental background noise.

The invention discloses an electric energy quality analysis method based on variational mode decomposition multi-scale permutation entropy. The method comprises the steps of collecting original training data of electric energy quality monitoring points when the electric energy quality monitoring points are disturbed by different types; adopting variational mode decomposition to decompose the components to obtain K IMF components; calculating the multi-scale permutation entropy of each IMF component, and constructing a feature vector of the original training data; selecting R features from thefeature vectors to form an optimized feature vector of the original training data; taking the optimized feature vector of the original training data and the corresponding disturbance type as input data and output data respectively, and training an ELM neural network model to obtain an electric energy quality disturbance classifier; and acquiring optimized feature vectors of voltage signals of to-be-detected power quality monitoring points according to same method; inputting the optimized feature vectors into the power quality disturbance classifier to obtain the disturbance type of the power quality monitoring point to be detected. According to the method, the real-time diagnosis efficiency of the disturbance type of the power quality is greatly improved.

The invention discloses an intelligent rotary machine fault depth network feature identification method. A vibration sensor is arranged at a to-be-detected rotating mechanical part of a train rollingbearing; collecting an original vibration sequence when the rolling bearing works; decomposing and reconstructing the original vibration sequence through a singular spectrum analysis method; extracting a root-mean-square value of the reconstructed vibration sequence; standard deviation, skewness index and peak value; a fault position is judged by using a rotary machine fault position diagnosis model obtained by training of a support vector machine; and then, carrying out ensemble empirical mode decomposition on the reconstructed vibration sequence, calculating the permutation entropy values ofa group of decomposed intrinsic mode components, taking the permutation combination of the permutation entropy values as a detection characteristic, and judging the fault type by using a rotary machine fault type diagnosis model obtained by training of a support vector machine. The fault position and the fault type of the rotary machine can be detected more timely, and the fault diagnosis accuracy and reliability are improved.

The present invention discloses a method and a system for noise reduction of a ground penetrating radar B-scan image based on EEMD and permutation entropy. First, a ground penetrating radar two-dimensional B-scan image signal is obtained, and then each obtained B-scan image signal is subjected to noise reduction to obtain the B-scan image signal of the channel after the noise suppression signal, and finally the signal after each channel is recombined to obtain a noise-suppressed two-dimensional B-scan image signal; for any one B-scan image signal, the method for denoising includes: performing EEMD decomposition on the B-scan image signal of the channel to obtain K IMF components arranged from high frequency to low frequency; calculating arrangement entropy values of each IMF component; and selecting an arrangement entropy value not greater than a preset value IMF components to reconstruct, and obtaining the signal after denoising. The invention solves the signal mode aliasing problem existing in the EMD decomposition and can effectively reduce the noise.

The invention discloses a bearing fault quantitative trend diagnosis method based on morphology and a multi-scale permutation entropy mean value. When the fault size of the inner ring or the outer ring of a bearing is changed, the modulation degree of the vibration signal of the bearing is changed. The change affects the complexity and randomness of the vibration signal. The method, by using the superiority of the multi-scale permutation entropy in the aspect of representing the degree of uncertainty of vibration signals, draws a relational graph of the multi-scale permutation entropy mean values and the fault sizes, and then realizes quantitative trend diagnosis of the rolling bearing faults. The vibration signal acquired by an experiment contains serious noise and a large number of interference signals. In order to remove the noise interference and enhance the impact performance of the vibration signal, the multi-scale morphology is introduced into the method so as to greatly improvethe accuracy of rolling bearing fault quantitative trend.

The invention discloses a multivariable time sequence prediction method based on adaptive noise reduction and integrated LSTM (Long Short Term Memory). The multivariable time sequence prediction method is used for solving the problems of unstable performance and low prediction precision when a multivariable time sequence with non-stationary, non-linear and noisy characteristics is predicted by anexisting method. The method comprises the following steps: decomposing a noise-containing chaotic multivariable time sequence by adopting a complete set empirical mode decomposition method with adaptive noise to obtain a series of intrinsic mode functions with frequencies from high to low; Distinguishing a noise-containing high-frequency intrinsic mode function from a low-frequency noise-free intrinsic mode function by adopting a permutation entropy thought; Constructing a self-adaptive threshold value and a self-adaptive threshold value function to reduce noise of the noisy intrinsic mode function; Constructing a stacked automatic encoder to extract characteristics of the multivariable time sequence after noise reduction; Constructing a multivariable time sequence weak predictor based onthe LSTM neural network; And constructing an integrated algorithm considering the prediction error of the verification set, and combining a plurality of LSTM weak predictors to obtain a strong predictor.

The invention provides a high speed train wheel set bearing fault diagnosis method based on MEEMD permutation entropy for the disadvantages of EMD and EEMD. The high speed train wheel set bearing fault diagnosis method based on the MEEMD permutation entropy comprises the following steps in turn: signal acquisition; filtering and denoising of the original vibration signal; MEEMD decomposition; permutation entropy feature extraction; dividing the high-dimensional feature vectors into two groups; model training; and diagnosis result. In the feature extraction link, the signal features are reflected on multiple dimensions by introducing the MEEMD, and the relative single fault mode identification rate is obviously enhanced in comparison with EMD permutation entropy feature identification rate.The data required for the analysis method based on the MEEMD permutation entropy are short, and the anti-noise and anti-interference capacity is high so that the method can be effectively applied tohigh speed train wheel set bearing fault analysis.

The invention provides a time sequence complexity measurement method based on the image micro-structure frequency analysis. The method comprises the steps of constructing a signal recursive matrix; drawing a gray image according to the recursive matrix at the recursive state of the i moment and at the recursive state of the j moment; filtering the gray image according to Gaussian kernel functions of different scales so as to obtain different Gaussian gray images and form a Gaussian pyramid; subjecting the Gaussian kernel functions of different scales and the gray images to the convolution operation to obtain the scale space of the images; in the above scale space, preliminarily determining the positions and the scales of feature points; conducting the least-squares fitting based on the secondary expansion equation of the Taylor function of a scale-space function, and removing unstable feature points by utilizing the extreme values of a fitting surface; clustering remaining feature points; continuously changing the values of an influence degree and a validity degree through calculating; subjecting an obtained clustering result to information measurement; calculating the complexities of different signals by using an approximate entropy and a permutation entropy; and subjecting the results of microstructure recursive entropies to comparative analysis. The above method provides a beneficial reference for the design of novel entropy methods.

The invention provides a concealed attack detection method for industrial control on-site equipment, and the method can achieve the effective detection of concealed attacks. The method comprises the steps: comparing observation output of an industrial control object with predicted expected output, obtaining a residual error sequence, and carrying out the residual error preprocessing; calculating apermutation entropy of the residual error sequence; determining that the industrial control object encounters the concealed attack if the descend range of the permutation entropy of the residual error sequence is greater than a preset first threshold value in a preset time interval, or else determining that the industrial control object does not encounter the concealed attack. The method is suitable for the concealed attack detection of the industrial control on-site equipment.

The invention discloses a power quality disturbance detection method for a power distribution network based on improved EWT and CMPE, and the method comprises the steps of decomposing a PQ disturbancesignal of an active power distribution network system by adopting improved empirical wavelet decomposition EWT, filtering noise of the PQ disturbance signal, and decomposing to obtain an EWT component containing characteristic information; using the EWT component containing the characteristic information an input signal of a composite multi-scale permutation entropy CMPE algorithm, carrying out permutation entropy calculation on each EWT component containing the characteristic information by utilizing the composite multi-scale permutation entropy CMPE algorithm, and calculating an entropy value matrix of each PQ disturbance signal under each mode function; using a PCA algorithm to perform dimensionality reduction on the calculated entropy value matrix, calculating principal component components, and obtaining characteristic ranges of various PQ disturbance signals; according to the obtained eigenvalue matrix after the dimensionality reduction processing, using the eigenvalue matrix asthe input quantity of a SVM algorithm; and identifying the PQ disturbance signal of the active power distribution network system containing the distributed energy. The power quality disturbance detection method for the power distribution network based on improved EWT and CMPE is simple in steps and accurate in classification, and can improve the reliability of the power distribution network.

The invention discloses a classification system for EEG signals in different anesthesia states. The classification system comprises an EEG signal acquisition module, a front-end signal processing module, a frequency domain and time domain parameter calculation module, an anesthesia depth estimation module and a display module. The characteristics of different anesthesia depths are obtained, and the obtained characteristic values are divided into clear anesthesia, superficial anesthesia, normal anesthesia and deep anesthesia. The invention introduces a brain function index which combines the permutation entropy and the burst suppression ratio to analyze the complex nonlinear random signal of the EEG signal in the frequency domain and the time domain, thereby improving the classification accuracy of the deep anesthesia. The classification method can be used to determine the anesthesia depth of a patient during anesthesia and operation, and provides a reliable basis for the medical staffto perform anesthesia operation on the patient.

The present invention discloses an OLTC mechanical fault diagnosis method based on MPE and SVM. The method comprises the following steps of: 1) performing collection of vibration signals of an on-loadtap-switch (OLTC) in a normal state and vibration signals of the OLTC in fault state through an acceleration sensor to perform preprocessing of the vibration signals; 2) performing multi-scale permutation entropy (MPE) calculation of the collected vibration signals, and constructing feature vectors taken as input of a support vector machine (SVM); and 3) inputting the feature vectors obtained inthe step 2) into the SVM to perform training of the SVM, and inputting test data into the trained SVM to determine a fault mode of the OLTC. The OLTC mechanical fault diagnosis method based on the MPEand the SVM does not need a lot of data for training of the SVM and is higher in diagnosis precision, and is obviously better than a BP neural network in the diagnosis effect of the OLTC.

The invention discloses a wind turbine generator fault diagnosis method, which comprises the following steps: according to the vibration signal characteristics of a wind turbine generator gearbox, carrying out variational mode decomposition on signals under different working conditions to obtain a series of intrinsic mode function components, and respectively solving multi-scale permutation entropies of the intrinsic mode function components; combining the multi-scale permutation entropy and the original signal time domain feature into a feature vector, and inputting the feature vector into atransfer learning algorithm; the covariance of a source domain and a target domain being minimized through a linear transformation matrix, the distribution difference of signal data of the source domain and the target domain being reduced through second-order statistics alignment, and then inputting the feature vectors of the aligned signal data of the source domain and the target domain into a support vector machine for fault classification. According to the method, the problem of poor classification effect caused by different distribution of the vibration signal data under different workingconditions can be solved, and the method has higher accuracy in wind turbine generator fault diagnosis under variable working conditions.

The invention discloses a joint noise reduction method based on variational mode decomposition and permutation entropy, and belongs to the technical field of signal processing. The joint noise reduction method comprises the following steps: firstly, reading a noisy signal x (t); calculating the scale index a value of x (t), and determining the decomposition number K of variational mode decomposition (VMD) according to the a value; secondly, performing K-layer VMD decomposition on the noisy signal x (t) to obtain a series of variational mode components uk; then, calculating the permutation entropy of each variational mode component uk, and further determining the value of K; and finally, removing the noise uk component, and reconstructing the remaining uk component to obtain a noise-reducedand filtered signal. The method has the advantages of adaptively determining the decomposition number, effectively removing noise components and being high in robustness and real-time performance, effective noise reduction and filtering processing can be carried out on signals, and the method has good technical value and application prospects for non-stationary signal noise reduction methods.

A method for separating and identifying composite fault features of a wind power transmission chain includes the following steps of: obtaining a vibration signal by using an acceleration sensor; decomposing the acquired vibration signal in a whole frequency range by using multi-wavelet packet transform; by using a permutation entropy as an evaluation index, selecting single signals satisfying requirements to perform reconstruction to complete signal noise reduction and composite fault separation; and processing the reconstructed signals by using an energy operator demodulation method to identify fault information.

The invention discloses a calculation order tracking method capable of adaptively reducing noise and avoiding order aliasing, which is characterized by comprising the following steps of: defining a margin frequency according to signal rotating speed information and a predicted maximum analysis order; carrying out VMD pre-decomposition on the signal, reserving a mode of which the center frequency is lower than the margin frequency, and abandoning the mode of which the center frequency is higher than the margin frequency so as to filter out high-frequency noise in the signal and high-order components in a non-analysis order bandwidth; calculating the permutation entropy PE of the reconstructed signal: using the PE for representing the random degree of the time sequence, wherein the larger the value of the PE is, the more random the time sequence is; optimizing the VMD parameter by adopting a differential evolution algorithm to obtain a parameter, and adaptively generating a reconstructedsignal; and calculating a resampling order, carrying out calculation order tracking on the obtained reconstructed signal, and carrying out FFT after obtaining the resampling signal to obtain an orderspectrum of the signal. The method is used for processing an original vibration signal so as to adaptively reduce noise interference in the collected vibration signal and highlight fault information.

The invention discloses a personalization anesthesia closed-loop control system. The brain electricity data are obtained by a signal collection end of a brain electricity data acquisition device, then the collected brain electricity data are transmitted to a brain signal pretreatment device for the removal of noises and outliers; then the pretreated brain electricity data are sent by the brain electricity signal pretreatment device to a brain electricity signal analysis device; the brain electricity signal analysis device and the closed-loop control system are connected and the pretreated data are analyzed, the expected anesthesia depth index RPE (Renyi permutation entropy) values are obtained, and the RPE values are transmitted to a closed-loop control system; the closed-loop control system is connected with an injection device and at the same time controls it to work, the next-step anesthetic infusion speed is calculated by the closed-loop control system, after the signals are received by the injection device, and the real time control of the anesthetic infusion speed is completed by the injection device. The system has the advantages of real-time monitoring, automatic control, and high accuracy and the like.