70results about How to "Implement adaptive selection" patented technology

The invention relates a self-adaptive semi-supervised network traffic classification method, system and equipment. The method comprises the following steps: acquiring a network stream, extracting thestream feature with the preset fixed quantity in each network stream to obtain a network stream feature vector; computing the centroid of the network stream feature vector in each type according to the marked network stream feature vector, thereby obtaining a vector set M; performing self-adaptive semi-supervised k-means clustering by taking the vector set M as an initial center point; mapping theobtained network stream in each type of cluster to the belonged traffic type according to the maximum posterior probability; taking the traffic cluster of the known type as the training data to trainan online traffic classifier. The invention further relates to a system, the system comprises an acquisition module, a vector set processing module, a clustering module, a classification module, andan output module. The invention relates to the equipment. The equipment comprises a processor, a memorizer, and a computer program stored on the memorizer and capable of being run on the memorizer.

The invention relates to a timely four-wheel drive (4WD) electric wheel automobile adaptive steering system and a control method thereof, and belongs to a 4WD electric wheel automobile differential power-assisted steering system. A steering clutch and a steering servo motor are arranged between a main steering pipe column and a rear steering pipe column, and a planetary gear mechanism is arranged between the steering clutch and the steering servo motor, so that the timely 4WD electric wheel automobile adaptive steering system can realize adaptive selection of steering modes and a timely 4WD adaptive transmission-ratio-variable four-wheel differential power-assisted steering function on the basis of finishing the conventional electric wheel automobile differential power-assisted steering, and realizes transmission-ratio-variable control and active steering intervention.

The invention discloses a forward-looking radar imaging sea-surface target key point detection and recognition method. The method includes the steps that radar echo data are quantized into a gray scale image; a region of interest is extracted from the gray scale image, partition is performed, and a target region partition image is obtained; the radar echo data and the region of interest of the image are used, and a target region peak point information matrix is obtained; information merging is conducted on the target region partition image and the target region peak point information matrix, the number K of peak points in a target region in a merging result is obtained through counting, the previous K peak points are selected and serve as target effective peak points and form a target effective peak point image in a binaryzation mode, target axial features are extracted, and a target position is determined; the center of gravity of target energy is calculated and serves as the target key point. According to forward-looking radar target features, multiple mode identification methods are used comprehensively, the inherent features of a target can be reserved, meanwhile interference factors such as artifacts and side lobes can be restrained, and the recognition accuracy and the positioning precision of radar imaging sea surface target key points are improved.

The present invention provides a locomotive and vehicle abnormal axle temperature diagnostic method and system. Temperature time-domain features of a plurality of associated measurement points at a plurality of time windows are combined to a feature set, the feature set is subjected to k-means clustering, the feature position difference is determined through the k-means clustering to determine whether there is an isolated measurement point or not, if yes, the maximum radius of a cluster where normal measurement points belong to is taken as a radius of neighborhood, the number of the temperature time-domain features corresponding to a single associated measurement pint is taken as the minimum neighborhood density, the feature set is subjected to DBSCAN clustering according to the radius ofneighborhood and the minimum neighborhood density to determine whether the distribution density has obvious difference or not. If the results of the k-means clustering and the DBSCAN clustering of thetemperature time-domain features corresponding to a certain associated measurement point have the position difference and distribution density difference at the same time, it is determined that the associated measurement points have temperature anomaly. The method and the system provided by the invention effectively improve the locomotive and vehicle abnormal axle temperature diagnosis accuracy and correspondingly reduce the diagnosis misjudgment rate.

The invention belongs to the technical field of image processing and in particular relates to a self-adaptive image segmentation method based on a fuzzy threshold value. The self-adaptive image segmentation method comprises the following steps: step 1, pre-processing a histogram to acquire an image histogram with double-peak properties; step 2, carrying out gradient detection on the pre-processed image histogram and determining the position of a wave trough; step 3, determining the position of a wave peak according to the position of the wave trough; step 4, determining the distance between two adjacent wave peaks according to peak values of the wave peaks; calculating window width sizes of membership functions of different images according to the distances between the different wave peaks; and step 5, determining a segmentation threshold value. The self-adaptive selection of the window width is realized, and the disadvantage that the segmentation of images on the histogram with inconspicuous double peaks cannot be easily realized by the fuzzy threshold value is effectively improved, an applicable range of the fuzzy threshold value image segmentation method is expanded, and the segmentation effect of the fuzzy threshold value segmentation method is improved.

The invention discloses a mechanical impact type fault diagnosis method based on a variable space-scale frame, which is characterized by comprising the following steps of: firstly, constructing frame functions with different space-scale characteristics on the basis of different control parameters; carrying out multiscale analysis on the dynamic response signal of electromechanical equipment by the constructed functions; calculating the global characteristic kurtosis of each time-scale frame function; selecting the optimal frame function of the dynamic signal by taking global characteristic kurtosis maximization as a target; analyzing the dynamic signal by the optimal frame function; calculating the time frequency joint kurtosis distribution map of each single reconstruction signal; and carrying out fault diagnosis by taking a sub-frequency band to which the maximum value of the time frequency joint kurtosis belongs in a picture as an optimal analysis sub-frequency band. With the method disclosed by the invention, an analysis parameter can be optimized, a key fault characteristic is self-adaptively extracted, the impact type fault diagnosis of the complex electromechanical equipment is realized, the phenomenon that the equipment operation quality is lowered is found as early as possible, and major machine accidents are avoided.

The invention discloses an adaptive image target enhancement method based on a difference of Gaussian model and belongs to the digital image processing field. In the method, based on a lateral inhibition phenomenon of a biological visual receptive field, the difference of Gaussian (DOG) model of image space domain filtering is established. Target and noise gray level distribution prior knowledge in a water surface image and a constraint relation of excitability and inhibition effect offset are used to select a model parameter so as to reach a local optimal enhancement effect. By using the method in the invention, comprehensive performance of target enhancement, background inhibition and noise filtering is better than the comprehensive performance of a traditional spatial-domain high pass filter. The enhanced image possesses a good visual sense and simultaneously a demand of subsequent motion vector estimation to a correlation operation signal to noise ratio is satisfied.

The invention provides a self-adaptive main color extracting method oriented to natural images and belongs to the field of image main color extraction. The problem is solved that an existing main color extracting method cannot achieve automatic determination of main color number, accordingly colors which do not exist in original images serve as main colors or important color characteristics in theimages are missed, and accurate main color extraction is not facilitated. The self-adaptive main color extracting method comprises the steps that the color spaces of images are converted into Lab spaces; then size compression, image enhancement, denoising and the like are conducted on the images; finally optimal main color sets in natural images are extracted in a self-adaptive mode by organically combining a Silhouette outline coefficient and a traditional K mean value clustering algorithm, a large number of natural images shot in various natural environments are tested, experimental resultsshow that main colors in the natural images can be automatically and effectively extracted out by adopting the algorithm, and the representativeness of color characteristics in original images is better shown.

An image segmentation method based on multi-level region synthesis includes the steps of firstly, obtaining multi-level segmentation results by using an existing multi-level image segmentation algorithm; secondly, performing synthesis in a global level range: selecting multiple segmentation results of an image from a low level to a high level, respectively calculating image region characteristicsof each level, performing unified quantitative description of the multiple characteristics, establishing a synthesized model of multi-level image segmentation, and performing the optimal combination of the segmentation regions by using a multi-label graph cutting method; thirdly, according to the global-level synthesized results, selecting a local level range and performing a second synthesis by using a multi-label graph cutting model; and fourthly, subjecting the level labels by the second synthesis to regional mapping to obtain a final image segmentation result. The image segmentation methodbased on the multi-level region synthesis in the invention selects the region with high target segmentation quality from the multiple segmentation levels and realizes adaptive selection; uses less regional features to calculate the segmentation quality and reduces the number of regional nodes involved in the calculation, with the used optimized combination model having a better optimization effect.

The invention discloses a self-adaptive stochastic resonance denoising method for a silicon single crystal growth image under a low signal-to-noise ratio. For the silicon single crystal growth image under the low signal-to-noise ratio, the bistable stochastic resonance is combined with the PSO optimization algorithm, and the self-adaptive stochastic resonance image denoising algorithm based on PSOis designed. According to the method disclosed by the invention, the stochastic resonance is used for detecting weak signals in a lossless mode, so thunder the silicon single crystal growth image under the low signal-to-noise ratio is denoised and enhanced. The quality of the image is improved and the PSO optimization algorithm is utilized. The Donoho noise standard deviation is used as the fitness function of the optimization algorithm. The system parameters of the stochastic resonance are adjusted in real time so as to obtain the optimal resonance output effect, and the image denoising effect is realized. After the silicon single crystal growth image is processed by the method, the noise can be effectively removed. The quality of the image can be improved. Therefore, the meniscus of thesilicon single crystal image under the low signal-to-noise ratio can be accurately detected. The foundation is laid for the accurate detection of the diameter of crystals.

The invention discloses a method for segmenting a uterine fibroid ultrasound image in HIFU treatment. The method comprises the implementation procedures that a tumor contour is manually initialized by a user; an image Gaussian pyramid is established to obtain a coarse scale image; shape constraint energy is established, and a local area active contour model after shape constraint is used for conducting coarse segmentation on the coarse scale image; the obtained coarse scale segmentation result is used as an initial contour for segmenting an original image, and the local area active contour model after shape constraint is used again for conducting fine segmentation on the original image; the fine segmentation result is the final tumor segmentation contour. According to the method, one new shape constraint is introduced, so that the precision for segmenting the uterine fibroid ultrasound image is improved, the problem of boundary leakage or excessive contraction caused by ultrasound image quality is solved, and the segmenting efficiency is greatly improved through implementation of a multiscale algorithm.

The invention discloses a broadband frequency hopping clustering multistage self-organizing network waveform design method. Distributed centerless multi-hop frequency hopping synchronization, initialnetwork establishment, dynamic gateway selection, physical layer modulation-demodulation and coding-decoding and adaptive routing selection based on link learning are realized. Multi-hop frequency hopping synchronization does not need to preset a synchronization center node and a special frequency hopping frequency set and pattern; wherein the synchronization time is related to the number of frequency points in the frequency hopping frequency set, the frequency hopping rate, the relay hop count and the like; each node transmits a frequency hopping synchronization frame probabilistically according to a local clock and a preset frequency hopping pattern, the nodes are all in a receiving state in other time, after a frame header is captured, the current frequency is kept to receive complete frame data, TOD information and node number information of the received data are obtained, and then synchronization is completed. Due to node movement or external environment change, a multi-hop end-to-end link exists and the link quality also changes, the performance of each link is comprehensively evaluated through a learning strategy, and finally, adaptive routing selection is realized.

The invention provides a hydrodynamics flood routing analogy method based on a double-deck asynchronous iteration strategy. Grid data with the same resolution ratio as a digital elevation model are used for representing flood data, and the actual water flow process is simulated by adopting a double-deck asynchronous iterative algorithm; a flood routing moment is determined by an outer layer iteration process according to the given iterative step length, the magnitude relationship between the time used for flowing from one grid to an adjacent grid and the grid iterative step length is compared by an inner layer iteration process to determine the respective inner layer iteration step length of grids, the self-adaptive selection of iterations is thus realized, the fracture surface water yield of each grid is subjected to iterative computations, and a flood routing graph is dynamically drawn by setting a time interval of outputting a flood form. FloodArea software, a homogenization iterative algorithm and the method provided by the invention are adopted for respectively simulating storm flood historical data of Fujian province Wanan drainage basin, a result indicates that the simulative result provided by the method is highest in goodness of fit with the actual disastrous situation, and the flood routing simulation precision is effectively promoted.

The invention discloses a partial discharge wavelet denoising method based on a hybrid particle swarm. The method comprises the steps that step S1, wavelet decomposition is performed on noised partial discharge signals so that the wavelet coefficient under each wavelet scale is obtained; step S2, self-adaptive threshold selection and processing are performed on the wavelet coefficients obtained in the step S1 by using a second-order differentiable class Sigmoid threshold function and a hybrid particle swarm self-adaptive threshold selection method so that effective values are maintained and noise components are eliminated, and the results after threshold processing are saved as new wavelet components; and step S3, signal reconstruction is performed by using the obtained wavelet coefficients so that the denoised partial discharge signals are obtained. The applied threshold function is the class Sigmoid function. Compared with the common soft threshold function, the function is second-order differentiable and adaptive to an iterative algorithm based on gradient descent so that wavelet coefficient threshold self-adaptive selection based on the minimum mean square error can be realized.

The invention relates to an adaptive time-frequency peak filtering and de-noising method based on improved Hough transform, which belongs to the field of geophysical technology. A multi-layer fractional Fourier Hough approach is presented for linear event seismic record. Event envelope identification is carried out on seismic data. A filter trajectory equation similar to the trend of event is built. An optimal trajectory of time-frequency peak filtering is extracted adaptively. Thus, filter bias is fully reduced. The detection result can be obtained quickly under the condition of strong noise. The method has high accuracy and strong anti-interference ability.

The invention discloses an image segmentation level selection method based on scale perception, and aims to improve the segmentation quality of an overall image. The method comprises the following main steps: firstly, obtaining a segmentation result represented by a tree form by using a multi-level image segmentation algorithm; secondly, calculating feature vectors of all segmented regions in eachsegmentation layer, and performing quantitative description on region segmentation quality according to a visual format tower principle; thirdly, constructing a graph model of the multi-level segmented image by taking the hierarchical region with the finest segmentation granularity as a node, and finally, mapping a label to a region corresponding to the original hierarchy, and combining to obtaina final image segmentation result. According to the method, based on the scale selection principle of segmentation area quality, the problem of hierarchical selection of multi-level image segmentation is solved, and the limitation of traditional single hierarchical selection and the uncertainty of threshold parameters on segmentation hierarchical selection are overcome. As a post-processing means, the output quality of the multi-level image segmentation algorithm in a visual processing task can be improved.

The invention provides a substation fire-fighting continuous operation system and a method thereof. The system comprises a fire-fighting robot and fire-fighting medium supply equipment, wherein the fire-fighting medium supply equipment is provided with a water supply mechanism and a foam supply mechanism. A fire-fighting mode combining the fire-fighting robot with the fire-fighting medium supply equipment is provided, and continuous operation can be carried out. Meanwhile, spraying mediums and spraying angles can be switched according to fire situations, it is ensured that fire behaviors are controlled most rapidly, and guarantee is provided for reliable and safe operation of a substation.

The invention provides a method for selecting an adaptive beam-forming mode under a large-scale MIMO (Multiple-Input Multiple-Output) system. The method comprises the following steps: selecting a channel model and acquiring base station antenna array parameters, including the quantity of antennas and the arrangement way of an antenna array; representing the system average throughputs of two users in a single beam-forming mode and a dual beam-forming mode as the functions of the azimuth angles of the two users respectively in combination with the base station antenna array parameters and the channel model; calculating the difference between the azimuth angles of the two users when the system average throughputs in the single beam-forming mode and the dual beam-forming mode are equal to each other, and marking the difference as an azimuth angle threshold value; and adopting the dual beam-forming mode for the two users whose difference between the azimuth angles is greater than the azimuth angle threshold value, and adopting the single beam-forming mode for the two users whose difference between the azimuth angles is smaller than the azimuth angle threshold value. Adaptive selection of the beam-forming mode according to the user position can be realized in order to maximize the system throughput.

The invention provides a video coding reference frame selection method and apparatus. The method includes respectively acquiring the time domain distance absolute value sequence and quantization parameter sequence between the current image and a previous image; respectively constructing a time domain distance weighting table and a quantification parameter weighting table; overlapping the time domain distance weighting table and the quantification parameter weighting table, sequencing from big to small, and selecting the image corresponding to the maximum value as the reference image of the current image. The video coding reference frame selection method and apparatus realize the self-adaptive selection of the reference image in the video image coding process, thereby improving the coding efficiency and the compression efficiency, and saving the bandwidth source.

The invention discloses an urban financial invasion case prediction system and prediction method based on a space-time scale adaptive model. The system comprises a model calculation task management module, a model definition module, a model training module, a model prediction module, a prediction result display module and a model registration module. The method comprises the following steps: (1) carrying out pyramid hierarchical grid statistics on urban financial invasion cases; (2) analyzing spatial and temporal distribution characteristics of urban financial invasion cases; (3) adaptively determining the space-time scale of the urban financial invasion case prediction model; and (4) constructing an ST-ResNet-based urban financial invasion case prediction model as a demonstration. The method is helpful for revealing crime geographical differentiation characteristics, can meet the requirements of different crime prediction precisions, and achieves the optimal arrangement of police patrol work.

The invention provides a bearing fault diagnosis method based on multi-island inheritance and variational mode decomposition, and the method comprises the following steps: collecting a vibration signal of a rolling bearing, employing a variational mode decomposition method optimized by a multi-island inheritance algorithm to decompose the vibration signal of the rolling bearing, and obtaining an intrinsic mode component; calculating a sensitive parameter SP of each modal component, selecting two intrinsic modal components with the maximum value of the sensitive parameter SP for reconstruction,and demodulating the reconstructed signal to obtain an envelope spectrum; carrying out signal enhancement on the envelope spectrum by adopting a minimum entropy convolution MOMEDA algorithm to obtainan enhanced envelope spectrum; and identifying rolling bearing fault frequency characteristics according to the enhanced envelope spectrum, and carrying out rolling bearing fault diagnosis. Accordingto the invention, the fault information and noise information of the bearing signal can be effectively separated, the influence of noise on the fault feature extraction precision of the rolling bearing is suppressed, and the fault feature extraction precision of the rolling bearing is greatly improved.

The invention provides a wind power curve fitting method based on sparse heteroscedasticity multi-strip regression, and the method comprises the steps: automatically detecting an abnormal point through employing a fuzzy C-means algorithm, and obtaining the data of which the abnormal point is removed for original wind power data; constructing a sparse heteroscedasticity multi-strip regression modelaccording to the obtained data; optimizing the constructed sparse heteroscedasticity multi-strip regression model by adopting a variational Bayesian method to obtain posteriori distribution conditions and parameter formulas of all parameters in the model; and initializing model parameters, and solving estimated values of the parameters by utilizing an iterative method according to posteriori distribution conditions and parameter formulas of all the parameters in the model. According to the wind power curve fitting method based on sparse heteroscedasticity multi-spline regression provided by the invention, a plurality of spline basis functions is integrated, the nonlinear fitting capability of the model is improved, and the influence of redundant information on a final regression result isavoided.

The invention provides a tunnel seismic wave data denoising method and system based on time-frequency domain spectral subtraction. The method comprises the following steps: acquiring sampling frequency information and an acquisition signal of noisy seismic record data; converting time domain data into a time-frequency domain; establishing a relationship between a filtering window length and a signal main frequency, and solving the optimal filtering window length; adaptively selecting a peak position and a variance of a Gaussian filter coefficient; respectively selecting wavelet coefficient matrixes of the last plurality of time windows in each frequency band to solve an average value as a noise spectrum estimation matrix; subtracting the noise spectrum estimation matrix in different frequency bands by using the Gaussian filter window added with the coefficient; and converting the wavelet coefficient matrix into a time domain to realize a time-frequency domain spectral subtraction denoising process. According to the invention, tunnel seismic wave data denoising is carried out through a time-frequency domain spectral subtraction method, the distinguishing capacity of a traditional spectral subtraction method in the frequency direction is optimized, and noise reduction and removal in the tunnel environment are achieved.

The invention discloses a rolling bearing fault diagnosis method based on a clustering K-SVD algorithm, and the method mainly comprises the following steps: firstly carrying out adaptive decompositionon an original signal by employing a time-varying filtering empirical mode decomposition algorithm based on particle swarm optimization to obtain each intrinsic mode component, and calculating the related kurtosis index (Kcr) value of each component; then, selecting an intrinsic mode component with the maximum Kcr index value as an input sample of the clustering K-SVD algorithm to perform dictionary learning to obtain an over-complete dictionary DNew; and finally, carrying out sparse feature extraction on the original signal of the rolling bearing by utilizing the over-complete dictionary DNew and combining an orthogonal matching pursuit algorithm, and carrying out envelope spectrum analysis on a sparse representation result to extract the fault frequency feature of the rolling bearing. The method effectively solves the problem that a classical K-SVD algorithm is low in learning precision of the fault impact characteristics of the rolling bearing, and is of great significance for achieving weak fault diagnosis of the rolling bearing.

The invention relates to Doppler signal processing, and aims to track the Doppler signal frequency in real time, effectively filters out base signals, realizes a large speed measurement range, and hashigh efficiency and low cost. The invention discloses an adaptive selection and test method for laser Doppler signal filtering bands, and comprises the steps of: dividing a passband of a filter intofour filtering bands according to the bandwidth of the Doppler frequency, wherein each filtering band is implemented by means of a fixed-bandwidth finite-length unit impulse response filter FIR; performing judgment selection on the four filtering bands in sequence, namely, executing the judgment selection according to a sequence of the filtering band 1, the filtering band 2, the filtering band 3 and the filtering band 4; and calculating a difference value between two moments, that is time spent on signal processing, calculating a maximum value of running speed variations of a tested object within time delta t according to the maximum acceleration of the tested object, calculating a maximum value delta fd of Doppler frequency variations at the next moment, and determining overlapping portions among the four filtering bands according to the maximum value of Doppler frequency variations. The adaptive selection and test method is mainly applied to speed measurement occasions.

The invention provides a self-adaptive registration method and a self-adaptive registration system for biological imaging. The method comprises the steps of extracting the SIFT feature points of images and matching the feature points; according to the SIFT feature points, generating deformation field initial parameters for different deformation images in the self-adaptive manner; fusing the spacecorrelation information of the images with the image registration based on the gray level; and adjusting the step size in the self-adaptive manner in the iteration process. According to the invention,the self-adaptive registration of high-resolution biological imaging is realized through the automatic and intelligent means. The operation is simple and convenient, and the detection efficiency is high. The reliability is high, and the universality is high.

The invention discloses a method applied to UMTS AKA to resist replay attack, and relates to a method for resisting replay attack on basis of an early warning mechanism self-adaptive selection protocol. The method is based on the early warning mechanism self-adaptive selection protocol, thereby effectively resisting replay attack from attackers. On the basis of the defect that the SQN in the UMTS AKA is used for resisting replay attack, the scheme capable of effectively resisting replay attack is provided; by means of the scheme, replay attack is resisted, the network traffic is reduced at the same time, and the execution efficiency of the whole network is improved.

The invention discloses a single-ended adaptive traveling wave ultra-high-speed protection system and method for coping with a high-resistance fault. The method comprises the following steps: measuring the current and voltage of a protection installation part of a direct-current transmission line, calculating a first traveling wave of a zero-mode fault current, and taking an absolute value of thefirst traveling wave; carrying out wavelet transform modulus maximum detection on the absolute value of the first traveling wave of the zero-mode fault current, and adaptively changing the length of adata window of protection calculation; constructing a basic fitting function to fit the absolute value of the first traveling wave of the zero-mode fault current according to the expression form of the first traveling wave of the zero-mode fault current to obtain a fitting coefficient; forming a fault data set containing different fault distances and transition resistance faults, and when a faultactually occurs, obtaining a transition resistance estimation interval of the fault according to the position of (bf, af) obtained by fitting the first traveling wave of the zero-mode current in thefault data set; and according to the obtained transition resistance estimation interval, adaptively changing a set value of traveling wave protection, and realizing high-sensitivity discrimination offaults inside and outside the region.

The invention discloses a natural language theme classification method and device. A training stage in the method comprises the following steps: acquiring a natural language text segment with a knowntheme as a sample set; extracting a plurality of words with the highest occurrence frequency in the sample set to obtain a plurality of feature words; representing each feature word as a vector to obtain a plurality of feature vectors; calculating the similarity degree between any two feature vectors to obtain a similarity degree set; inputting the similarity degree, the topics and the feature words corresponding to the topics into a preset neural network structure for training to obtain a feature spectrum and a model for expressing a relationship between the feature spectrum and a classification result. The classification stage comprises the steps of obtaining a to-be-classified natural language text segment; extracting feature words belonging to the feature spectrum in the to-be-classified natural language text segments to obtain input feature parameters; and inputting the input characteristic parameters into a model for expressing the relationship between the characteristic spectrumand the classification result to obtain the classification result. The method can achieve the adaptive selection of features, and improves the classification accuracy.

The invention provides a method of micro-doppler signal analysis based on combination of linear canonical transform and S-transform. According to the invention, based on echo signal analysis of moving objects performed through the S-transform and linear canonical transform, the optimal time-frequency analysis result frequency resolution is obtained. The method comprises the following steps: a linear regular transformation model of echo signals is constructed; the time-shift characteristic of the echo signals after the linear canonical transform is obtained; an S-transform model of the echo signals is built, and the echo signals after the linear canonical transform is substituted into the S-transform model; based on the time-shift characteristic of the linear canonical transform, the S-transform model is subjected to integration and linear coordinate transformation; a parameter matrix in a linear canonical transform model is obtained; the parameter matrix is adjusted; and the optimal time-frequency analysis result frequency resolution is obtained. The method makes the size of a window function change with the frequency of the analyzed signals, and has a stronger time-frequency analysis function.