65results about How to "Implement features" patented technology

This application discloses a method and apparatus for constructing a multi-scale adaptive gating MambaPlus network, belonging to the field of artificial intelligence and machine learning technology. The method includes: initializing the network configuration and constructing the basic structure; preprocessing the input data to generate a standard dataset; mapping the input data to the hidden space via an input mapping layer, and extracting backbone features from the Mamba backbone; constructing at least two parallel scale branches in the hidden space to obtain a multi-scale feature set; inputting the backbone features and multi-scale features into an adaptive gating module, dynamically allocating weights and adaptively fusing them through a hierarchical gating mechanism to generate fused features; further enhancing the features through cross-scale attention and feedforward enhancement, and then superimposing the residuals to generate the final discriminative features; finally, completing category prediction and model training evaluation. This application, while retaining the advantages of Mamba's long-range dependency modeling, addresses the problems of insufficient utilization of multi-scale information, poor adaptive feature fusion, and low robustness in complex scenarios.

The invention discloses an electric vehicle charging load prediction method, system and device and a storage medium, and relates to the technical field of electric power, and the method comprises the steps: generating graph structure charging load data based on multi-station charging load data; based on the graph structure charging load data, executing a graph convolutional network-gating time convolutional network adopting an adaptive adjacency matrix, and generating hidden layer representation containing spatial adjacency information; the hidden layer representation and the external features are spliced and input to a time domain fusion Transform, and an electric vehicle charging load prediction result is output; according to the method, the spatial correlation extraction and time-dependent modeling of the charging load are realized by constructing the graph convolutional network-gating time convolutional network based on the adaptive adjacency matrix and the time domain fusion Transform model, and the precision and interpretability of multi-station load prediction are improved in combination with variable selection and an interpretable multi-head attention mechanism.

This invention belongs to the field of nonlinear data classification technology and discloses a nonlinear data classification method based on deep transfer learning and visual recognition capabilities. This invention transforms high-dimensional numerical features into structured images, enabling convolutional neural networks to leverage their powerful visual recognition capabilities to handle tabular data classification tasks. This image-based representation naturally reveals the implicit nonlinear relationships within the data. By further integrating deep transfer learning, this method effectively alleviates the data sparsity problem in small-sample scenarios and improves generalization performance through a series of training optimization strategies. The framework proposed in this invention achieves an effective connection between the numerical feature space and the image-based deep learning architecture, providing theoretical innovation and practical value for complex nonlinear data classification tasks.

The present application relates to a kind of based on multimodal feature fusion distribution network disturbance source identification method, belong to the fine disturbance monitoring and diagnosis technical field of smart grid.The method includes: the disturbance class current traveling wave data extracted is carried out time-frequency conversion, generates traveling wave panorama waveform chart and two-dimensional time-frequency chart and is fused into three-channel time-frequency image, then image feature vector is extracted;The disturbance class current traveling wave data is carried out Prony modal parameter fitting, extracts modal parameter feature vector and is projected to high-dimensional feature space;The disturbance class current traveling wave data is carried out Clark transformation, calculates zero mode energy proportion feature and is projected to high-dimensional feature space;Image feature vector, projected physical feature vector and projected modulus energy feature vector are fused;Fusion feature vector is input into classifier, and the class identification result of distribution network disturbance source is output.The present application aims to solve the technical problems that the prior art has single feature representation, weak physical interpretability and difficulty in forming stable power fingerprint.

The present application relates to the technical field of computer vision, deep learning and graphics rendering, in particular to a 3D portrait reconstruction method and system based on cascade diffusion prior and UV space physical constraint, which adopts a parameter efficient adaptation paradigm of freezing a unified DiT backbone and a cascade LoRA adapter, and sequentially completes UV space texture completion, illumination homogenization and intrinsic material decomposition; introduces a cross-intrinsic attention mechanism to realize multi-material pixel-level spatial alignment, and executes differential physical rendering supervision to constrain material physical consistency in the UV domain. The present application realizes 100,000-level field image generalization under less than 100 3D scanning samples, and the generated 4K PBR material assets have no baking light and shadow residues, and the physical behavior is consistent under re-illumination, which is suitable for high-fidelity 3D portrait asset production in the fields of digital human creation, film special effects and virtual reality.

The invention relates to the technical field of medical rehabilitation instruments, and discloses an exoskeleton robot for fracture auxiliary rehabilitation. The robot comprises a shear deformation sensing assembly, a shape memory alloy locking assembly and a joint driving module. The shear deformation sensing assembly comprises a fixed shell, a flexible liner arranged on the inner side of the shell and a sensing beam connected between the shell and the liner. The shape memory alloy locking assembly comprises a locking pin arranged in the connecting rod structure, a shape memory alloy driving piece in linkage with the locking pin and a locking part located on the connecting rod structure and corresponding to the locking pin in position. And the strain gauge is electrically connected with the shape memory alloy driving piece. Transverse micro-slippage representing shearing risks is directly sensed through a mechanical structure, rigid mechanical locking is triggered in a linkage mode, and therefore rapid and reliable physical protection is provided for a fracture part, and secondary injury in rehabilitation training is effectively prevented.

The application provides a medicine cold chain logistics transportation whole-process tracking method and system, and relates to the technical field of logistics transportation monitoring. First, the original sensor tracking data stream output by the internal sensor cluster of a transportation carrier in a continuous collection time window is acquired. Then, the transportation carrier operation trajectory line is generated according to the geographic positioning parameters, and the multi-dimensional environment state marker set is generated by combining the physical parameters at the same time of each geographic position coordinate point. Meanwhile, the logistics transfer node topology connection network containing the logistics transfer node unit and the directed transfer path edge is constructed. The operation trajectory line, the multi-dimensional environment state marker set and the topology connection network are input into the pre-constructed space-time state feature fusion tracking model to generate the whole-process state tracking feature tensor. The abnormal state type identification and the space-time distribution feature are obtained by performing the abnormal state mode matching on the whole-process state tracking feature tensor, and the abnormal early warning instruction set is generated. The application can comprehensively and accurately track the whole process of medicine cold chain transportation, intelligently identify the abnormality and timely generate the early warning.

The application discloses a smart community police management system and method based on Internet of Things security, and belongs to the technical field of smart communities, and solves the problem that existing system data processing cannot utilize edge computing nodes for low-delay real-time analysis, and cannot meet timely early warning requirements, the method comprising preprocessing of multi-source heterogeneous community monitoring information, analysis and processing of the community monitoring information based on edge computing, and generation of a risk propagation path in combination with front-end analysis results and heterogeneity characteristics; in the application, analysis and processing of the community monitoring information based on edge computing avoids the delay of centralized processing, can quickly trigger early warning in the early stage of a risk event, thereby gaining golden disposal time for risk event disposal, and a risk analysis model realizes a leap from single-node anomaly detection to composite risk quantification through multi-dimensional feature fusion and dynamic space-time reasoning, and realizes multi-dimensional feature fusion and dynamic reasoning capability of the risk event.

The application discloses a kind of chronic stress dynamic evaluation method, system, equipment and storage medium based on psychology-physiology multimodal fusion, comprising the following processes: obtaining the dynamic data of multiple time points of evaluation object and clinical baseline characteristic data;Dynamic data is numerized and dimension standardization is handled, and psychological characteristic vector is obtained;Clinical baseline characteristic data is discretely coded and handled, and clinical characteristic vector is constructed;Based on dynamic data, extract heart rate variability features and serological features;Psychological characteristic vector, clinical characteristic vector, heart rate variability features and serological features are spliced according to time sequence, and multi-time point sequence characteristics are constructed;Multi-time point sequence characteristics are input into the machine learning classification model completed by pre-training for calculation, and output chronic stress risk continuous quantitative probability value, and according to chronic stress risk continuous quantitative probability value, output chronic stress judgment result.

This invention discloses a method and system for detecting abrupt temperature changes in lake water. The method includes: acquiring daily time-series data of lake water temperature and calculating the diurnal temperature variation range to obtain a maximum temperature rise sequence and a maximum temperature fall sequence; calculating the quantile thresholds corresponding to the maximum temperature rise and maximum temperature fall sequences to obtain a temperature rise threshold sequence and a temperature fall threshold sequence; and detecting two types of abrupt temperature change events—cold-to-warm and hot-to-cold—based on these sequences, calculating the frequency, duration, intensity, and speed of occurrence of each event, thereby achieving the detection of abrupt temperature changes in lake water. This invention can characterize the rapid change process of lake water temperature on a daily scale, enabling systematic detection and multi-dimensional feature quantification of abrupt temperature changes in lake water. As a method and system for detecting abrupt temperature changes in lake water, this invention can be widely applied in the field of water body monitoring technology.

An aero-engine oil-gas heat exchanger high-altitude performance test device, comprising a test cabin, an oil system, and an air system; the oil system comprises an oil tank, an oil supply pump, and an oil heater; the oil tank is connected with the oil supply pump at an oil outlet, the oil supply pump is connected with the oil heater, and the oil heater is connected with the test cabin at an oil inlet; the test cabin is connected with the oil tank at an oil return port through a pipeline; the air system comprises an air dryer, a turbine expander, a mixing tank, and a vacuum pump; the air dryer is connected with a gas source at an air inlet, the air dryer is connected with the turbine expander at an air outlet, the turbine expander is connected with the mixing tank through a second one-way valve, and the mixing tank is connected with the test cabin at an air inlet through a pipeline; the test cabin is connected with a radiator at an air outlet, and the vacuum pump is connected with the outlet end of the radiator.

The application discloses a kind of cable tunnel edge side multimodal fusion acquisition and collaborative early warning method, belong to power cable tunnel fire safety monitoring technical field.The present application is aimed at the problems of existing cable fire early warning scheme, such as early warning lag, high false alarm and missed alarm rate of single mode detection, cloud processing response delay, edge computing resource conflict, constructs multi-thread parallel acquisition architecture, synchronously collects visual, olfactory and electrical signal mode data and completes timing alignment;Extracting fire precursor features to construct a multi-modal fusion feature vector, calculate the comprehensive risk index;Match the cable fire development stage to execute four-level collaborative early warning judgment;Adaptive sliding window and hierarchical cooling mechanism are used to realize early warning state management;Video acquisition push and intelligent detection unit are fully decoupled, and parallel operation without interference is realized.The present application can realize early warning of cable fire in whole cycle, greatly advance disposal window period, significantly reduce false alarm rate and compress early warning response time.

The application relates to an endoscope image polyp segmentation method and device, computer equipment and a storage medium. A trained neural network model comprises a feature extraction layer, an adaptive perception module, a comprehensive feature extraction module and a segmentation head. The method comprises the following steps: acquiring an endoscope image to be processed; inputting the endoscope image to be processed into the trained neural network model, performing feature extraction on the endoscope image by using the feature extraction layer, and obtaining a plurality of feature maps of different scales; inputting the plurality of feature maps of different scales into the adaptive perception module, performing flexible feature processing on the plurality of feature maps of different scales by using a deformable convolution structure in the adaptive perception module, and obtaining an optimized feature map corresponding to each feature map; and inputting the plurality of optimized feature maps after fusion processing into the segmentation head, and obtaining a polyp segmentation image output by the segmentation head. The application can improve the accuracy of endoscope image polyp segmentation.

The invention discloses a hyperspectral image classification method based on a self-supervised contrast learning network. The method comprises the following steps: constructing a basic network; a parallel rotation prediction and mask reconstruction dual-task self-supervision module carries out feature learning under the condition of insufficient labels; the double-branch spatial spectrum feature extraction fusion module introduces a double attention mechanism to reduce the model complexity and enhance feature representation; the self-supervised contrast feature learning module guides the network to learn feature information with strong discrimination through rotation, mask and contrast learning of multi-task joint loss. And finally, a pre-trained model is utilized to carry out fine adjustment under a small number of labeled samples and send the labeled samples into a classifier to obtain a classification result. Through the double-task self-supervision module, the double-branch spatial-spectral feature extraction and fusion module and the self-supervision comparison feature learning module, the dependency of the model on labeled data is effectively reduced, discriminative features are learned, and the classification precision and the model robustness in a few-sample scene are improved.

The invention provides a wind turbine generator model fault ride-through structured modeling method, which comprises the following steps: (1) selecting a wind turbine generator, carrying out high-low ride-through test on the wind turbine generator, and analyzing transient response characteristics of different wind turbine generators; (2) obtaining fault ride-through response characteristics of the wind turbine generator; (3) detecting the grid-connected point voltage of the wind turbine generator, and judging and selecting a high-low crossing mode; (4) dividing the whole process of the fault period of the wind turbine generator, and carrying out fault switching logic setting; (5) constructing a control instruction of the wind turbine generator; (6) constructing an active and reactive power control instruction during the fault period; (7) performing parameter identification on the physical parameters of the wind turbine generator and the parameters of the stator and rotor controller to obtain a complete model; and (8) comparing the obtained model with a manufacturer packaging model, and checking the consistency of the model and the manufacturer packaging model. According to the method, on the basis of a large number of actual wind turbine generator set actual measurement curves and fault time segment modeling, fitting and response characteristic analysis of response characteristics of different actual wind turbine generator sets can be achieved.

The invention discloses a data model hybrid driven pine wood nematode disease transmission risk prediction method, and belongs to the technical field of pine wood nematode disease monitoring. The objective of the invention is to improve pine wood nematode disease transmission risk prediction precision. The method comprises the following steps: constructing a multi-source factor data matrix; according to the propagation law of the pine wood nematode disease, a cellular automaton model is constructed, and the spatial diffusion process of the pine wood nematode disease in each county-level administrative division geographic unit in the forest is calculated and simulated; constructing a TabNet neural network model, performing feature learning on the multi-source factor data matrix, and identifying key driving factors of pine wood nematode disease transmission; a PINN physical information neural network framework is constructed, a cellular automaton model is used as a physical constraint module, a TabNet neural network model is used as a data driving module, and the physical constraint module and the data driving module are linked through a composite loss function; and using the trained PINN physical information neural network framework to predict the propagation and diffusion trend of the pine wood nematode disease in the future and realize visualization.

The present application relates to the technical field of image processing, in particular to a real-time train oil unloading staff falling recognition method, device, equipment and storage medium, wherein the real-time train oil unloading falling recognition method comprises: acquiring a current frame picture shot by a camera, detecting a person position frame based on a person detection model; performing abnormal recognition on the detected person position frame; performing feature extraction and classification on the target person of the person position frame based on a key point model to obtain key point data and staff uniform classification data; judging whether the target person corresponding to the staff uniform classification data is a staff; performing falling recognition on the person position frame corresponding to all staff based on the key point data; and outputting a staff falling recognition result. Through target detection and key point fusion, the present application realizes person detection and feature extraction on the working staff working scene picture, recognizes whether falling occurs, and outputs a warning prompt information, so that the security personnel do not need to check the camera at all times, and the safety is high.

The application discloses a kind of pancreatitis time sequence image dynamic evaluation methods based on self-attention mechanism, it is related to medical image intelligent analysis technical field, including, to the multi-time point pancreas enhancement CT image after pre-processing is carried out spatial registration, and using Swin-UNet medical image segmentation network extracts the multiscale self-attention feature of pancreas and lesion, generates pancreatitis time sequence image feature sequence;Pancreatitis time sequence image feature sequence is carried out lesion identification and cross-time association, establishes lesion trajectory index table, and calculates trajectory level attention weight in the self-attention layer of Swin-UNet medical image segmentation network, aggregation obtains lesion trajectory weighted representation matrix.The application is dynamically adjusted trajectory level attention weight by using time sequence weighted fusion method, realizes the adaptive fusion of image feature and clinical index in time dimension, improves the consistency and clinical explainability of severity evaluation time sequence.

This invention discloses a monolithic cadmium ion-selective electrode based on highly dispersible lipophilic molybdenum disulfide and its preparation method. The main steps include: preparing a mixture of molybdenum and sulfur sources → mixing the mixture with long-chain alkyl groups to obtain product A → adding product A to an ion-selective membrane solution → casting onto an electrode, followed by ventilation and evaporation to obtain the monolithic cadmium ion-selective electrode. By controlling the capacitance and surface wettability of molybdenum disulfide through surface / interlayer modulation, the superhydrophobic and high capacitance characteristics of the highly lipophilic molybdenum disulfide material are achieved, while exhibiting high dispersibility in the key flow components of the membrane. The highly dispersible membrane component material designed in this invention provides an alternative method for the simplified manufacturing of all-solid-state ion-selective electrodes and offers a new approach for mass production.

The invention discloses a GNSS-based dam seepage area micro-deformation monitoring method and system, and relates to the technical field of dam safety monitoring, and the method comprises the steps: collecting GNSS observation data of a reference monitoring point and a working monitoring point, carrying out the data processing, and calculating a three-dimensional coordinate time sequence through a multi-constellation multi-frequency combined calculation method; forming a space displacement time sequence through a three-dimensional coordinate difference value of the reference monitoring point and the working monitoring point, performing direction projection on the space displacement time sequence according to a dam axis direction unit vector to obtain a dam axis direction displacement sequence, and extracting a residual signal time sequence; and performing time-delay consistency analysis on the displacement sequence in the dam axis direction and the residual signal time sequence, calculating a deformation consistency threshold value and a deformation consistency threshold value change rate, identifying a seepage response state and performing monitoring operation. Through a deformation consistency threshold value and change rate calculation method, time sequence characteristic quantification and continuous monitoring of deformation response of the seepage area are realized.

The application discloses a large flexible space deployable structure modeling analysis method based on discrete geometry and multi-body dynamics, comprising the following steps: based on a three-dimensional geometric model of a space deployable structure, a discrete grid topology composed of one-dimensional rod members and two-dimensional plate members is constructed; based on the discrete geometry, a strain energy analytical model and a dynamic equation of the one-dimensional rod members and the two-dimensional plate members are respectively established; based on the assembly connection relationship of the physical structure of the three-dimensional geometric model, an interface for constructing the dynamic coupling relationship between heterogeneous units is determined; according to the strain energy of the one-dimensional rod members and the two-dimensional plate members and the potential energy of the interface, a coupled global differential algebraic equation set is established; and a domain decomposition algorithm based on interface synthesis is adopted to perform static condensation and ill-condition alleviation on the coupled global differential algebraic equation set, so as to solve the dynamic response. The application is suitable for digital twinning and on-orbit control evaluation of a new generation of space large deployable structure with the characteristics of super-large span, extremely low fundamental frequency and strong geometric nonlinearity.

The application relates to the technical field of visual tracking, and discloses an AI vision-based bamboo-plastic packaging box tracking and positioning identification method, which comprises the following steps: performing scene light field analysis on an original image sequence of a bamboo-plastic packaging box to obtain environmental light field parameters; based on the material reflection characteristics of the bamboo-plastic packaging box, performing optical attribute extraction to obtain reflection characteristic parameters; based on the environmental light field parameters, performing optical response simulation on the reflection characteristic parameters of the bamboo-plastic packaging box to obtain visual samples; performing cross-domain feature alignment on the original image sequence and the visual samples to obtain joint feature expression; performing illumination interference decoupling on the joint feature expression to obtain illumination robust features; based on the illumination robust features, performing pose state evaluation on the spatial position and the attitude of the bamboo-plastic packaging box to obtain six-degree-of-freedom tracking and positioning data; and the application can improve the tracking and positioning identification efficiency of the bamboo-plastic packaging box.

The utility model discloses a live broadcast intelligent switching device which comprises a picture special effect switching table, an electric control box is installed above the picture special effect switching table in a butt joint mode, and a power supply module and a programmable logic controller are fixedly installed in the electric control box. The lower bottom surface of the electric control box is vertically connected with a plurality of first electric telescopic rods, a plurality of second electric telescopic rods and a plurality of stepping motors; the first electric telescopic rods, the stepping motors and the second electric telescopic rods are all connected with the power supply module and the programmable logic controller through wires; according to the utility model, the key is accurately pressed through the first electric telescopic rod and the pressure sensor, the push rod is accurately controlled by the second electric telescopic rod and the linear displacement sensor to move linearly, and the knob is accurately rotated by the stepping motor, so that the fine operation of the picture special effect switching table is realized; meanwhile, a voice recognition and synthesis module is arranged, voice instruction operation and operation feedback are supported, and a scene triggering automation process can be set on the touch screen.

The invention discloses a performance testing device and testing method for an air filter, and belongs to the technical field of testing devices. Comprising a detection mechanism, a material storage assembly comprises a material box, two sets of partition plates are installed in the material box, the material box communicates with an air inlet pipe of the detection mechanism through a discharging pipe, and the partition plates divide the material box and the discharging pipe into three material areas; the adjusting assembly comprises a motor, a supporting shaft is installed at the output end of the motor, a feeding wheel is installed on the supporting shaft and rotationally connected with the discharging pipe, a plurality of material grooves and a plurality of adjusting holes are formed in the feeding wheel, and the material grooves and the adjusting holes which are adjacent in the axial direction are separated through partition plates. The feeding wheel is provided with two feeding channels including the material groove and the adjusting hole, the feeding amount of different channels can be controlled by matching with radial movement of the adjusting frame, mixed input of single-particle-size dust and dust of various particle sizes according to different proportions can be achieved, the feeding concentration of the dust of the same particle size can be adjusted, and the feeding efficiency is improved. Therefore, various actual use environments such as urban road construction sites, deserts and the like are simulated.

The application provides a fluorescent imaging probe based on a covalent organic framework, a preparation method and application. The fluorescent imaging probe is successfully prepared by the preparation method, the fluorescent intensity of the fluorescent imaging probe has a good linear relationship with the concentration of iron ions, the fluorescent imaging probe can be used for detecting the concentration of iron ions, and the fluorescent imaging probe has excellent biocompatibility and can be used for in vivo fluorescent imaging.

The invention particularly relates to a miniaturized ultra-wideband high-efficiency phased-array antenna loaded with parallel plate couplers. The miniaturized ultra-wideband high-efficiency phased-array antenna comprises a plurality of columns of antenna sub-arrays, a combining network, a metal bottom plate and a radio frequency connector, each column of antenna sub-arrays comprises a plurality of metal oscillators; the metal oscillator is composed of a pair of index radiators, the inner side and the outer side of each index radiator are provided with index gradual change curves, and parallel plate couplers are loaded on the upper half portions of the left side and the right side of each index radiator in a mode of being perpendicular to the index radiators. The combining network is embedded between the antenna sub-array and the metal bottom plate; and the radio frequency connector is embedded in the metal bottom plate and is used for feeding the antenna sub-array. By loading the parallel plate coupler, the bandwidth of the antenna is expanded, miniaturization design is realized, cross decoupling grounding loading is adopted, balanced feed and deresonance design and antenna unit pure metal design are realized, integrated machining can be realized, secondary assembly is avoided, and the antenna has the characteristic of high-efficiency radiation due to ultra wide band, high power and high efficiency.

The application relates to the technical field of intelligent water monitoring, and discloses a flow measurement data verification method based on abnormal water use analysis, which comprises the following steps: deploying an intelligent sensor network comprising a flow meter, a pressure gauge and a thermometer, collecting multi-source data of abnormal water flow state and performing phase-amplitude decoupling processing to generate a feature matrix containing flow state correlation information; establishing a multi-layer probability field model, converting error components into measurement error probability clouds, constructing an error back propagation path and designing cascade compensation parameters; correcting the probability clouds by using a gradient optimization algorithm, combining deep learning technology to enhance the probability convergence field and determining a standard flow value; tracking and analyzing the error propagation link and the attenuation coefficient spectrum to establish an error convergence boundary and output a flow correction result; and finally combining environmental state detection to construct an abnormal verification feature spectrum, completing comprehensive verification of the flow measurement data, effectively improving the detection accuracy of the abnormal water flow state, and improving the accuracy and reliability of flow measurement.

The present application relates to the technical field of electronic countermeasure, in particular to a method and system for dynamic interception of UAV group based on distributed jamming array cooperation. The method deploys a cluster of jamming nodes with autonomous logical association in the defense domain dynamically, forms a distributed jamming field with elastic reconfiguration through software-defined radio architecture; extracts time-varying spectral features through adversarial neural network, and constructs a dynamic threat fingerprint library; generates a jamming vector with frequency band following and power self-adaptation based on the dynamic threat fingerprint library, and updates the feature weight of the dynamic threat fingerprint library in real time according to the jamming effectiveness feedback; generates a hierarchical jamming wave sequence based on the multi-agent consensus protocol according to the trend of the situation evolution of the dynamic threat fingerprint library, and forms a time-space coupled interception chain. The present application realizes the whole process dynamic countermeasure capability of UAV group, maintains high efficient interception performance in complex electromagnetic environment, and reduces the interference risk to legal communication.

The present application relates to the technical field of virtual clothing recommendation, and specifically discloses a virtual clothing fitting recommendation method and system based on artificial intelligence, which establishes a cross-domain feature decoupling and alignment mechanism by fusing commodity standard images and user real-shot images, realizes the extraction and unified representation of the essential visual attributes of clothing, and fundamentally solves the problem of inconsistent clothing appearance caused by differences in shooting conditions. By constructing an environment-independent feature model and a domain-invariant feature space, the system can accurately measure the visual similarity of clothing in images from different sources, significantly improving the reliability and user trust of the recommendation results. The system realizes precise matching of the recommendation results and high-fidelity presentation of the virtual fitting effect, can accurately capture the fitting effect of clothing and individual form before the user makes a purchase decision, and realizes early identification and visual satisfaction of personalized needs.

The application discloses a vibration signal denoising and feature extraction method based on wavelet analysis and belongs to the technical field of cable monitoring. The method comprises the following steps: acquiring a cable vibration signal and performing double-tree complex wavelet decomposition to obtain a detail coefficient; a layered adaptive threshold model is established according to the statistical characteristics of environmental noise, and the detail coefficient is subjected to threshold processing to filter out noise; the vibration signal is filtered according to the interval determined according to the impact pulse index and the frequency band energy ratio of historical external damage transient impact; the energy of the filtered detail coefficient is enhanced; the time-frequency entropy and the instantaneous frequency are calculated to extract the time-frequency combined features, and the spatial correlation coefficient of multiple sensors is calculated to extract the spatial features; the energy features, the entropy features and the statistical features based on the full sub-band are extracted, and the features are spliced with the aforementioned features to form a detail-enhanced feature vector; and the feature vector is input into an LSTM network model to output a denoised and optimized feature vector. The application can inhibit environmental noise and enhance fault features.