72 results about "Wavelet entropy" patented technology

The present invention discloses post-wavelet analysis treating method and device for electric power transient signal. The treating method for electric power transient signal after wavelet analysis and before feeding to the electric power monitoring center includes the following treatment on wavelet coefficient: the extraction of module maximum and the detection of irregularity; the statistics and cluster analysis of wavelet coefficient; neural network classification; energy analysis; and wavelet entropy calculation. The present invention can extract the characteristic of electric power transient signal for the application in traveling wave ranging, fault recognition, electric energy quality analysis and equipment fault diagnosis in the transmission line of power system.

The invention relates to a method for extracting and fusing time, frequency and space domain multi-parameter electroencephalogram characters, which comprises the following steps: 1) collecting an electroencephalogram signal; 2) performing data pre-processing on the electroencephalogram signal; 3) extracting Kc complexity, approximate entropy and wavelet entropy from the pre-processed data; 4) on the basis of AMUSE algorithm, acquiring an electroencephalogram singular value decomposition matrix parameter; 5) performing character selection on the time, frequency and space domain character parameters for the extracted Kc complexity, approximate entropy, wavelet entropy and electroencephalogram singular value decomposition matrix parameters; 6) utilizing a SVM classifier to fuse and classify the four parameters of the time, frequency and space domains after the character selection. According to the method provided by the invention, the Kc complexity, the approximate entropy, the wavelet entropy and the electroencephalogram singular value decomposition matrix parameter can be selected for comprehensively presenting electroencephalogram character information, and then subsequent effective fusion is performed, so that effective support and help can be supplied to early diagnosis assessment for the brain functional disordered diseases, such as, Alzheimer disease, mild cognitive impairment, and the like.

The invention relates to a partial discharge signal denoising method based on lifting wavelet transform, which includes the following steps: (1) a partial discharge signal to be denoised is inputted; (2) lifting wavelet decomposition is carried out on the partial discharge signal, so that high-frequency coefficient components of different decomposition scales and a low-frequency coefficient component of the highest scale are obtained; (3) wavelet entropy-based layered thresholds and a soft threshold function are adopted to quantify the high-frequency coefficient components in order to remove noise components, and the high-frequency coefficient components are stored as new high-frequency coefficient components; (4) the new high-frequency coefficient components and the low-frequency coefficient component of the highest scale obtained in step (3) are utilized to compose a coefficient component for signal reconstruction, signal reconstruction is carried out on the coefficient, and thereby a denoised partial discharge signal is obtained. Lifting wavelets are completely transformed in a time (space) domain, and high-pass and low-pass filters are turned into a series of relatively simple prediction and update steps. Therefore the denoising speed of lifting wavelet transform is high, the design is flexible and simple, and the partial discharge signal denoising method is easy to put into practice.

The invention belongs to the field of laser radar signal processing, relates to a laser radar signal processing method based on a wavelet entropy threshold value and a modulus maximum value method, and relates to wavelet transform. The laser radar signal processing method based on the wavelet entropy threshold value and the modulus maximum value method is characterized in that at first a signal is transformed through multi-scale binary wavelet, a signal catastrophe point is removed based on a three sigma standard, the modulus maximum value is adopted, a low scale signal feature is searched and acquired through a selected maximum point on the highest scale based on the theory of scale modulus maximum tracking. Simultaneously, in allusion to halfway situation of denoising effect of a first layer detail information noise, the theory of wavelet entropy threshold is introduced to carry out threshold de-noising, and self-adaption threshold value selection is achieved. The signal noise in a multispectral laser radar system can be effectively removed, and the signal detail information can be retained as much as possible. Therefore, the detectivity of the laser radar is improved, and the laser radar signal processing method based on the wavelet entropy threshold value and the modulus maximum value method has a very good application prospect and development potential.

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 heart rate variability feature classification method based on generalized scale wavelet entropy, and belongs to the field of electrocardiosignal processing. The method comprises the steps that after an original electrocardiosignal to be processed is preprocessed through interference and baseline drift removal, R wave positioning is performed, and an HRV sequence is obtained by calculating the interval between every two R waves; discrete wavelet transform is performed on the HRV sequence to obtain discrete wavelet coefficients, and then alpha-order generalized wavelet entropy of all layers of the wavelet coefficients is calculated by selecting appropriate alpha values as needed; scales with the statistical difference are screened out to serve as feature layers according to the obtained entropy values, and the alpha-order generalized wavelet entropy values of the feature layers are utilized to construct feature vectors to perform classification identifying on the electrocardiosignal.

The invention relates to a low-power consumption portable electrocardiograph monitor and a control method thereof, which belongs to the field of medical appliances. The portable electrocardiograph monitor comprises an electrocardiosignal collection unit, an electrocardiosignal analysis and storage unit and an upper computer. The control method comprises the following steps of: step 1: carrying out digital filtering to collected signals by using a wavelet entropy optimal threshold denoising method; step 2: carrying out rejection of baseline drift to the signals processed by step 1 by using a morphological filtering method; and step 3: carrying out disease classification to the signals processed by step 2 by using a support vector machine algorithm. The invention has the advantage of high integrity, and has the characteristics of portability, compactness and economy.

The invention discloses a multi-load series arc fault detection and positioning method. The method comprises the steps of: collecting a current signal in a main loop, performing omnibearing feature extraction on the current by using time domain, frequency domain and energy analysis, wherein the feature extraction comprises a variance, a correlation coefficient and a peak factor index based on timedomain analysis; based on the harmonic amplitude of discrete Fourier transform and the frequency band energy and wavelet entropy of discrete wavelet transform. calculating the average Gini inpurity reduction amount of different characteristics by adopting a random forest, performing screening to obtain feature combinations for different types of loads, inputting the feature combinations into a deep neural network for training, and establishing an arc detection and positioning model for different types of loads for determining whether an arc occurs or not and determining the position of the arc and outputting fault information.

The invention discloses a method of X-ray fluorescence spectrum background rejection. Iterative wavelet transform is used to analyze the X-ray fluorescence spectrum, which overcomes the defect that spectrum distortion is easily caused when traditional wavelet transform is used to reject the spectrum background. When the characteristic peak and the frequency band of the background coincide, the method provided by the invention can still effectively extract the pure characteristic peak. The concept of wavelet approximation energy is presented in the invention and is used to evaluate the distribution situation of the background energy at a low frequency band. Compared with wavelet energy, the wavelet approximation energy can be used for more directly and accurately evaluating the spectrum background at a low frequency band. According to the method, wavelet entropy is used to select an optimal wavelet basis; after dilation and translation, the optimal wavelet basis can better match the whole or partial spectrum, and the spectrum sparsity after wavelet transform can be increased, and the operational efficiency is improved.

The invention provides an electrocardiosignal denoising method adopting a wavelet entropy threshold value based on EEMD. The electrocardiosignal denoising method comprises the following steps: SA, selecting the added noise times M and the added white noise sequence evaluation coefficient k, and carrying out EEMD decomposition on electrocardiosignals containing noises, thus obtaining a series of intrinsic mode function IMF components with the frequencies being from high to low; SB, carrying out wavelet decomposition on each IMF component, calculating the wavelet entropy threshold value, and carrying out wavelet entropy threshold value de-noising on each IMF component; and C, reconstructing the IMF components after the wavelet entropy threshold value de-noising, thus obtaining electrocardiosignals with the noises removed. With the method provided by the invention, the original signals and the noises can be effectively distinguished, and useful signals are well reserved while the noises are removed.

The invention discloses a method for eliminating nonuniform noise in speech collected by radar. The method comprises the steps that data windowing is carried out on a noisy radar speech signal y(n); according to a given decomposition scale, discrete wavelet transform (DWT) decomposition is carried out on the speech signal; wavelet entropy (WE) calculation is carried out; noise estimation is carried out based on dynamically-updated wavelet entropy (WE); and according to the updated noise estimation, Wiener filtering is used to enhance the speech signal. According to the invention, dynamic advantages of wavelet entropy noise estimation and advantages of Wiener filtering are combined to form a wavelet entropy Wiener filtering speech enhancement algorithm, thus nonuniform colored noise of radar speech can be effectively filtered out.

The invention relates to a method for extracting lightning strike signals and transient harmonic signals in a power system, which overcomes the problem of inaccurate extraction of the method for extracting the lightning strike signals and the transient harmonic signals in the power system in the prior art. The method comprises the steps of firstly converting the collected current signals to digital signals, further carrying out fast Fourier transform on the digital signals, determining the sampling frequency and the decomposition scale of wavelet decomposition according to the transform result, then carrying out wavelet decomposition on the digital signals, sequentially carrying out modulus maxima extraction, singularity detection and generalized wavelet entropy calculation on wavelet coefficients or single-branch wavelet reconstruction signals after the decomposition, and finally extracting the lightning strike signals and the transient harmonic signals according to the calculation result. The method can overcome the deficiencies of the prior art and be applicable to representing information features of the lightning signals and the transient harmonic signals in the time-frequencydomain.

The invention discloses a power distribution fault rapid positioning method based on discrete wavelet transform (DWT), which combines discrete wavelet transform (DWT) and a BP neural network (BPNN) inpower distribution system fault positioning, and wavelet energy spectrum entropy and per unit energy calculated according to detail coefficients of the DWT can effectively reflect fault characteristics. Besides, due to DWT decomposition, noise is filtered out, wavelet entropy (EPU) serves as input of a trained BPNN model, the fault position can be found rapidly and accurately, and the effects ofhigh sensitivity and high reliability are achieved.

The invention discloses a non-reference image quality evaluating method based on wavelet and structural self-similarity analysis, which is used for image quality evaluation of grayscale images and belongs to the technical field of image processing. The method comprises the steps of: firstly, recombining an original image to be evaluated into a new image comprising 4<n>self-similarity sub-blocks; then, carrying out wavelet decomposition on the new image; computing a wavelet entropy; and finally evaluating quality according to a wavelet entropy value, wherein the smaller the wavelet entropy value is, the higher the quality of the images is and the better a visual effect is. Compared with other similar methods, in the method of the invention, a wavelet decomposition algorithm and the self-similarity of the images are combined for the first time; the entropy of a wavelet coefficient is computed for the first time; and a computing result (the wavelet entropy) is used for evaluating the quality of the images. Meanwhile, in the method, any reference image is not needed.

The invention belongs to the technical field of human body distinguishing and discloses a human body heat source feature extracting and distinguishing method based on infrared pyroelectric information. The method aims to achieve 360-degree wide-range and remote detection and detection of static infrared heat sources and effectively solves the problems of confliction between comfort and energy saving of an intelligent air-conditioner and high false alarm rate of a safety supervision system. According to the technical scheme, the method comprises the steps that a stepping motor is adopted to drive a single pyroelectric infrared detector to rotate at a uniform speed, and then remote, 360-degree wide-range and static object detection is achieved; the detection range is a disk shape, wavelet packet analysis is conducted on collected human body heat source samples and non-human-body heat source samples, a signal wavelet entropy is taken as the features of the signal, 5-fold cross-validation is conducted by means of the BP neural network, and then human body heat resources are distinguished from non-human-body heat resources. The method is mainly applied to human body distinguishing.

The invention discloses an intelligent on-line diagnosis and location method of power transformer winding deformation. A winding deformation is called that after the power transformer is lashed by short-circuit or impacted by transportation, characteristics such as distortion and bulging of a local winding can occur under an electrodynamic force or a mechanical force, and buries a huge hidden danger to safe operation of a power network. The common methods of the winding deformation diagnosis are off-line diagnosis, and has the shortcomings of need of transformer shutdown and high requirementsfor professional skills of operators. The invention provides an intelligent on-line diagnosis method of the winding deformation combining information entropy and a support vector machine, the characteristics of current and voltage signals are extracted by using permutation entropy and wavelet entropy, variations of each monitoring index of the power transformer in the aspects of complexity, time-frequency domain and so on are synthesized, and automatic learning of diagnostic logic from fault features is achieved by a machine learning algorithm, the intelligent diagnosis of the winding deformation is achieved, so that the manpower cost is lowered, and the diagnostic efficiency is improved.

The invention discloses a double-layer spectral clustering method of a power load curve considering wavelet entropy dimensionality reduction. The method comprises the following steps: obtaining dailyload data of power load, and forming a data set; segmenting the power load data in the data set into q intervals, calculating a wavelet entropy Sq of an original data set within the interval q, and comparing and evaluating a degree of fluctuation of the data according to the calculated wavelet entropy value and a wavelet entropy threshold, wherein the degree of fluctuation greater than the specified threshold is large, and on the contrary, the degree of fluctuation is relatively small; performing statistics on the load number in which the wavelet entropy value is greater than the wavelet entropy threshold within the interval q, and calculating the proportion of the load data to the total load of the power load; dividing the interval with the proportion being greater than the threshold intotwo segments, calculating the wavelet entropy value again, and comparing and evaluating the degree of fluctuation of the data until the proportion of the load with large degree of fluctuation withinthe interval in all load is less than the threshold or the points in the interval cannot be divided equally, and obtaining load curve data with a variable time resolution; and performing double-layerspectral clustering to obtain refined load clusters with similar forms.

Disclosed is an intelligent on-line diagnosis method for winding deformation of power transformer. When a transformer is subjected to short-circuit shock or transportation collision, transformer windings may undergo local twisting, swelling or the like under the action of an electric power or mechanical force, which is called winding deformation and will cause a huge hidden danger to the safe operation of the power network. Commonly used diagnosis methods for winding deformation are all off-line diagnosis methods, which have the disadvantages that transformers need to be shut down and highly skilled operators are required. The present invention provide an intelligent on-line diagnosis method for winding deformation on the basis of combination of information entropy and support vector machine. By carrying out feature extraction of current and voltage signals based on permutation entropy and wavelet entropy, integrating the variation of the monitoring indicators of the power transformers in complexity, time-frequency domain and the like and automatically learning the diagnostic logic from fault features through the machine learning algorithm, intelligent diagnosis of winding deformation is realized, thereby reducing labor costs and improving diagnosis efficiency.

The invention relates to a real-time fault detection method applied to voltage source converter. Continuous discrete separation is carried out on a power electronic model, linear and non-linear models are extracted, and a multi-speed coordinative simulation system is utilized to carry out real-time simulation test analysis on a high-frequency switch device. Transient disturbance and stable disturbance are distinguished by high frequency wavelet decomposition coefficients obtained after frequency domain segmentation by a Mallat decomposition method. Fault beginning and ending time is determined by transient signals according to a model maximum value point, and the model maximum value in the fault time is used as a threshold for judging whether a fault occurs. Wavelet entropy transient characteristic extraction is carried out the high frequency decomposition coefficients, and when the frequency domain calculation result is inaccurate, the wavelet decomposition coefficient entropy is calculated to determine the phase having the fault.

Disclosed is an intelligent on-line diagnosis method for winding deformation of power transformer. When a transformer is subjected to short-circuit shock or transportation collision, transformer windings may undergo local twisting, swelling or the like under the action of an electric power or mechanical force, which is called winding deformation and will cause a huge hidden danger to the safe operation of the power network. Commonly used diagnosis methods for winding deformation are all off-line diagnosis methods, which have the disadvantages that transformers need to be shut down and highly skilled operators are required. The present invention provide an intelligent on-line diagnosis method for winding deformation on the basis of combination of information entropy and support vector machine. By carrying out feature extraction of current and voltage signals based on permutation entropy and wavelet entropy, integrating the variation of the monitoring indicators of the power transformers in complexity, time-frequency domain and the like and automatically learning the diagnostic logic from fault features through the machine learning algorithm, intelligent diagnosis of winding deformation is realized, thereby reducing labor costs and improving diagnosis efficiency.

The invention discloses an emotion recognition method and system. The method comprises the steps of acquiring an electroencephalogram signal to be recognized; extracting to-be-identified time-frequency domain characteristics, to-be-identified nonlinear characteristics and to-be-identified brain network attribute characteristics; wherein the to-be-identified time-frequency domain feature is a wavelet entropy calculated according to a wavelet decomposition coefficient of the to-be-identified electroencephalogram signal; wherein the to-be-identified nonlinear feature comprises a power spectrum density obtained by performing discrete Fourier transform on the to-be-identified electroencephalogram signal and a Hurst index obtained by performing average error calculation on the to-be-identified electroencephalogram signal; wherein the attribute characteristics of the to-be-identified brain network reflect the correlation among the to-be-identified electroencephalogram signals; and carrying out emotion recognition on the to-be-recognized electroencephalogram signal features by adopting a trained support vector classifier. The method and the system provided by the invention have the advantage that the emotion recognition accuracy can be improved.

The invention provides an autonomous underwater vehicle propeller fault detecting method based on wavelet single branch reconstruction. The method includes: performing multilayer discrete wavelet decomposition on an autonomous underwater vehicle control signal to obtain the fault description, at multiple frequency bands, of a propeller fault; respectively calculating to obtain the wavelet entropy of multi-frequency-band fault information and the optimal wavelet single branch reconstruction dimension; performing wavelet single branch reconstruction by using the obtained optimal wavelet single branch reconstruction dimension, wherein the time sequence position corresponding o the model maximum value point of the reconstructed signal is the fault occurrence moment of a propeller, and the autonomous underwater vehicle propeller fault detecting result is obtained. By the method applicable to the fields such as autonomous underwater vehicle propeller fault detecting, the problems that fault signal features are easily submerged by external interference signals after wavelet decomposition due to the influence of external interference, and fault detecting accuracy is low are solved, and the accuracy of the autonomous underwater vehicle propeller fault detecting is increased.

The invention discloses a method and a device for detecting fault of a power supply system of a spacecraft. The method is that the fault time is accurately detected through a fiber temperature sensor arranged on a primary bus of the power supply system by the wavelet theory based fault detection method according to the characteristic of temperature change of the bus in case of fault. The method comprises the steps of performing wavelet conversion for the original temperature signal containing noise; filtering a noise-containing multi-layer signal through wavelet entropy to remove most of the noise; performing the adjacent multi-scale product method to further inhibit the non-filtered noise interference; detecting the maximum value of a die so as to determine the accurate fault occurrence time. With the adoption of the method and the device, the mis-judgment caused by noise influence can be avoided, and the fault detection accuracy can be increased.

The invention relates to a micro-electromechanical gyro information fusion system based on wavelet entropies and a micro-electromechanical gyro fusion method. The system comprises a micro-electromechanical gyro array module (1), a data preprocessing and statistical analysis module (2), a wavelet entropy analyzing and computing module (3) and a signal weighting fusion module (4) which are orderly connected, wherein the signal weighting fusion module (4) is further connected to the data preprocessing and statistical analysis module (2). The fusion method comprises the steps of: performing data preprocessing and statistical analysis on the signal of each micro-electromechanical gyro, and then performing wavelet entropy analysis and calculation, and finally performing signal weighting fusion so as to obtain a fused composite micro-electromechanical gyro signal. Compared with the common variance weighting method, the information fusion method provided by the invention has the advantages that the accuracy of the gyro output signal is improved, and the stability problem existing in a single system can be solved, so that the accuracy and the robustness of the micro-electromechanical gyro are improved effectively.

The invention discloses a wavelet entropy and sparsity-based cable partial discharge signal adaptive wavelet denoising method, which comprises the steps of collecting a to-be-denoised cable partial discharge signal s (n), and establishing a wavelet library; selecting an optimal base wavelet and an optimal decomposition scale for wavelet decomposition of s (n) according to the wavelet entropy and sparsity of s (n); performing wavelet decomposition on s (n), and obtaining an approximation coefficient alpha J and detail coefficients dj at all decomposition scales j; calculating a wavelet threshold thrj of the wavelet coefficient of the jth layer; carrying out threshold quantification processing on the detail coefficient dj of the jth layer, filtering wavelet coefficients with absolute valuessmaller than a wavelet threshold thrj and weakening wavelet coefficients with absolute values larger than the wavelet threshold thrj to obtain a thresholded detail coefficient dj'of the jth layer; andreconstructing the approximation coefficient alpha J and the thresholding detail coefficient dj'by using discrete wavelet inverse transformation to obtain a denoised cable partial discharge signal s(n) '.

The invention discloses a subsurface crack size measurement method based on a surface wave and a BP neural network. The method comprises the following steps of 1, carrying out ultrasonic reflection and transmission signal detection on a subsurface crack with known buried depth and length; 2, extracting the time domain features and frequency domain features of a reflected surface wave signal and a transmitted surface wave signal; and 3, performing k-weight wavelet transformation on the time domain features and the frequency domain features obtained in the step 2, further calculating to obtain 2k+2 wavelet entropies, and taking the 2k+2 wavelet entropies as time-frequency domain feature parameters; and 4, training the BP neural network which takes the time-frequency domain characteristic parameters as input and takes the buried depth and length of the subsurface crack as output by utilizing the time-frequency domain characteristic parameters obtained in the step 3. According to the method, the damage-free and contact-free laser ultrasonic technology is adopted, the waveform data characteristic parameters generated by the interaction of the crack and the surface wave are analyzed in combination with the wavelet entropy theory and the neural network, and the burial depth and length of the crack can be obtained at the same time.

The invention discloses a manifold SVM analog circuit fault diagnosis method based on particle swarm optimization. The method comprises the main steps: software is used to simulate faults of a diagnosis object; for each fault in a circuit, Monte Carlo analysis is used to detect a feature signal of the fault, a wavelet packet is used to decompose the fault signal, signal decomposition has the maximum law based on the optimal wavelet entropy principle, and an optimal energy value of each group of signals is extracted to serve as a feature value of the fault; and during fault classification, a particle swarm algorithm is used, parameter optimization is carried out on a weight parameter and a penalty parameter in support vectors considering sample data class intervals, and thus, the optimal hyperplane of an SVM has better classification effects, and the fault diagnosis accuracy is improved.

The invention discloses a hierarchical fault diagnosis method based on transient current information and wavelet entropy. Topological structure analysis is firstly carried out on a distribution network; three-phase current information at a bus outlet is then acquired, a fault component is extracted, and a happening short circuit fault type is judged; the corresponding transient current information is then extracted to judge the general range of the happening fault, wavelet packet decomposition calculation is then carried out on the fault current in the range, a wavelet entropy value of signals is obtained, the value is then uploaded to a main station, and the main station judges the specific line of a fault point according to the wavelet entropy value of each detection point. The hierarchical fault diagnosis method based on transient current information and wavelet entropy has the advantages of few switching action times, quick fault point recognition, and comprehensive recognized fault types.

The invention discloses a machined part surface chatter mark defect detection method which comprises the following steps: measuring acceleration response signals of a cutter along the horizontal and vertical directions of a machine tool in the current machining process of a part, filtering cutter-pass harmonic components in the acceleration response signals, extracting wavelet entropy characteristics reflecting machining instability intensity in the acceleration response signals, and calculating the chatter mark defect of the machined part, recording the wavelet entropy characteristics as a vibration characteristic; inputting the vibration characteristics into a pre-trained flutter detection model, and judging the flutter state of a current weak-rigidity machining system for machining thepart; and if the part is in the flutter state, the current machining surface of the part has the chatter mark defect. After each part is machined, according to the accuracy of the current flutter detection model, on the basis of an existing flutter detection model, an incremental learning mode is adopted, continuously-accumulated actual measurement vibration information is utilized, some information which can have adverse effects on judgment precision is eliminated step by step, and therefore the accuracy of the flutter detection model is improved, and the accuracy of the flutter detection model is improved. And the detection precision of the machined surface chatter mark defects is high.

The invention discloses an emotion recognition method and system based on a three-dimensional feature map and a convolutional neural network. The method comprises the following steps: acquiring a to-be-recognized electroencephalogram signal; extracting a basic emotional state-free electroencephalogram signal from the to-be-recognized electroencephalogram signal; adopting wavelet packet transformation to decompose and reconstruct the electroencephalogram signals without the basic emotional state, obtaining multiple frequency band signals, and obtaining the wavelet energy ratio and the wavelet entropy of each frequency band signal; obtaining the complexity of each channel electroencephalogram signal in the electroencephalogram signals without the basic emotional state, and forming electroencephalogram features by the complexity of each channel electroencephalogram signal, the wavelet energy ratio of each frequency band signal and the wavelet entropy of each frequency band signal; arranging the electroencephalogram features to form a feature cube; and inputting the feature cube into a trained CNN model for emotion recognition. The accuracy of emotion recognition is improved.