812 results about "Adaptive weighting" patented technology

The invention discloses an adaptive question-answering system based on knowledge distillation and multi-modal dynamic fusion, and the system comprises a knowledge distillation module which is used for migrating knowledge of a teacher model pre-trained on corpora in the communication field to a lightweight student model, achieving model compression through optimizing a distillation loss function, and obtaining a multi-modal dynamic fusion model; the loss function comprises a soft label output by the teacher model and a KL divergence constraint output by the student model; the multi-modal knowledge fusion module comprises a feature extraction unit, a self-adaptive weighting unit and an attention fusion unit; the self-adaptive inference engine comprises a semantic analysis unit; according to the cross-modal reasoning method and system, semantic alignment of equipment parameters, protocol texts and topological graphs is achieved through the multi-modal dynamic fusion technology, and the cross-modal reasoning accuracy is improved; compared with an original model, the lightweight student model has the advantage that the reasoning speed is increased in a protocol analysis task.

The invention relates to the technical field of man-machine interaction, in particular to a robot behavior mode dynamic adjustment method based on multi-mode perception, which comprises the following steps: S1, collecting a visual image, a voice signal and an environment parameter of a target scene in real time; s2, extracting each modal feature; s3, dynamically allocating weights, and generating a fusion feature vector; s4, identifying a current scene type and a user attribute; s5, matching a corresponding interaction mode from a preset strategy library based on the scene type and the user attribute identified in the step S4; and S6, executing the matched interaction mode in the S5, and updating the weight distribution rule according to the feedback data. According to the method, vision, voice and environment characteristics are fused through a multi-mode perception technology, the behavior mode of the robot is dynamically adjusted based on a self-adaptive weighted fusion algorithm, and the interaction strategy is optimized in combination with user feedback, so that the interaction accuracy and the intelligent level of the robot under different scenes and user attributes are improved.

The invention relates to the technical field of aircraft skin defect detection, and discloses an improved YOLOv11-based aircraft skin defect multi-scale detection system and method. Extracting multi-scale features of the input image through HMSB; the HMSB is composed of a plurality of parallel sub-networks, each sub-network independently processes input images with different resolutions, and a lightweight Transform encoder is adopted to extract multi-scale features; the multi-scale features are adaptively fused through DFFM; the DFFM calculates the attention weight of each scale feature through a gating mechanism to realize self-adaptive weighted fusion of the multi-scale features; the defects are classified and positioned through SDH, and a detection result is output; the SDH comprises detection branches customized for various defect types, and each branch adopts a deformable convolution and a Transform decoder to improve the positioning precision of irregular defects. According to the invention, through hierarchical multi-scale feature processing and a dynamic adaptive fusion mechanism, the performance bottleneck of a traditional method in detection of tiny and irregular defects is solved.

The invention provides a real-time deep sea subsurface buoy monitoring system, which belongs to the technical field of deep sea measurement, and comprises a control chip, a multi-parameter sensor array, a data storage device and a power supply, high-precision monitoring is realized through the following steps: recording ocean data acquired by a multi-parameter sensor; correcting the sound wave propagation model and calculating the relative position of the subsurface buoy; constructing a three-dimensional data matrix and realizing data compression; identifying abnormal data by using a matrix disorder index; processing the time sequence data by adopting a sliding window weighted average method and eliminating jump; carrying out distributed preprocessing on the multi-source heterogeneous data and carrying out parallel processing by applying a matrix partitioning technology; accumulated errors are corrected by combining historical data and applying a Kalman filtering algorithm, and the problems of environment interference and sensor drift are solved through a deep sea environment disturbance compensation network model and a self-adaptive weighted error optimization function.

The invention discloses a black pig image segmentation method based on multi-feature fusion, and relates to the technical field of image processing, and the method comprises the steps: obtaining a black pig image, and carrying out the contrast enhancement processing; inputting the enhanced image into a residual convolutional network to generate a depth feature map, extracting a local shape feature map through curvature threshold segmentation and a double fitting strategy, and fusing the two feature maps to generate a black pig feature map; establishing a spatial position prior probability graph based on the black pig sample library, calculating regional correlation and performing adaptive weighting to obtain a fusion feature graph; boundary segmentation and iterative optimization are carried out on the fused feature map based on the dynamic behavior pattern map and the attitude constraint rule, and an initial segmentation map is generated; and adopting a group behavior model as an optimization criterion, correcting the boundary of the initial segmentation image, and outputting a final segmentation result. According to the method, the segmented enhancement function based on the double-peak characteristic and the local texture feature self-adaptive adjustment strategy are constructed, so that differential enhancement of image preprocessing is realized.

The invention discloses a multi-scale attention aggregation-based target point cloud segmentation method for a battery replacement robot, and the method comprises the steps: 1, collecting an RGB image, a depth image and three-dimensional point cloud data of a target fastener through a binocular structured light depth camera, and carrying out the fusion to generate a FastSeg3D data set; 2, a two-stage preprocessing method is provided, noise points are removed through radius filtering, and background point clusters far away from a target are removed through DBSCAN density clustering; 3, the network encoder uses a local feature aggregation module to extract geometric features, and the calculation complexity is reduced in combination with a random sampling strategy; 4, embedding a multi-scale attention aggregation module into the jump connection of the encoder and the decoder, fusing the features through a channel and a space attention unit, and achieving the self-adaptive weight weighting of the features of each layer of the encoder; and 5, recovering the resolution of the original point cloud by adopting nearest neighbor interpolation up-sampling, outputting a segmentation semantic tag, and obtaining a high-quality point cloud target. According to the method, the operation time of the battery replacement robot is shortened, and the balance problem of large-scale target point cloud segmentation speed and precision is solved.

The invention discloses an improved YOLOv5 protective equipment detection method combining channel selection and an attention mechanism, and the method comprises the steps: replacing an original spatial pyramid pooling fusion module SPPF of a YOLOv5 backbone network part with a channel selection multi-scale fusion module CSMF, and enabling the channel selection multi-scale fusion module CSMF to be based on a feature adaptive weighted fusion strategy of a multi-channel selection mechanism. According to context information and semantic distribution of input image features, fusion weights of feature channels under different receptive fields are dynamically adjusted, so that more discriminative multi-scale feature integration is realized. According to the invention, the adaptive capacity of the model to multi-scale targets is improved through the CSMF module, and the problem of inaccurate recognition of small targets such as gloves by a traditional model is solved; after the CBAM is adopted to enhance the MBconv module, the model can more effectively distinguish a key area and a background area in an image, and the robustness of target shielding and background interference in a complex scene is enhanced.

The invention discloses an on-load tap-changer vibration fault diagnosis algorithm based on tensor feature and adaptive weighted Stacking integration, relates to the technical field of on-load tap-changer fault diagnosis, and is used for improving the fault diagnosis precision. Comprising the following steps: S1, data acquisition; s2, feature extraction; the method comprises the following steps: extracting multi-scale time-frequency characteristics of an on-load tap-changer vibration signal by using wavelet scattering transform WST, and realizing low-rank decomposition and dimensionality reduction characterization of high-dimensional characteristics by combining a non-negative tensor decomposition model NTF; s3, fault diagnosis; a multi-base learner Stacking integration framework is adopted, and a prediction matrix is generated through K-fold cross validation; through a swarm intelligent optimization algorithm SRA, hyper-parameters and fusion weights of all base learners are adjusted, L2 regularization suppression over-fitting is introduced, and finally fault classification is realized by adopting a logic regression element learner with Softmax cross entropy. According to the invention, through fault diagnosis of multi-model adaptive fusion and optimization, the fault identification precision, stability and on-line monitoring capability are improved.

The invention discloses a heterogeneous sensor fusion method and system for dynamic covariance optimization, and the method comprises the steps: constructing a multi-source heterogeneous sensor network which comprises an underwater inertial navigation unit, a Doppler velocity sensor, an ultra-short baseline positioning system, a submarine topography matching device and a geomagnetic gradient detector; the method comprises the following steps: designing a distributed integrated navigation system by adopting a hierarchical fusion architecture, establishing a sensor confidence coefficient dynamic evaluation model based on online estimated error covariance matrix eigenvalue analysis, carrying out state prediction on a local filter level by applying Kalman filtering, and then carrying out measurement updating by adopting a novel variational Bayesian algorithm; a sensor autonomous switching and adaptive weighted dual-mode adjustment mechanism is developed on the main filter level, and self-healing improvement of navigation precision in a non-stationary environment is realized, so that the problems of dynamic reliability evaluation and fusion efficiency optimization in underwater multi-sensor multi-source navigation are solved.

The invention discloses a fragmentation method, system and device based on information entropy and Transform. The method comprises the following steps: calculating a basic amazing degree index and a corresponding attention distribution entropy value of each word position in a text sequence to be fragmented; dynamically adjusting the basic amazing degree index through the attention distribution entropy to obtain a corrected amazing degree, and obtaining a candidate boundary position set; the KL divergence of each candidate boundary position is calculated, and local entropy increase features are obtained through the information entropy variation; performing feature fusion processing on the corrected amazing degree, the local entropy increasing feature and the KL divergence based on an adaptive weighting mode to obtain a boundary score of each candidate boundary position; constructing an initial boundary position set, and carrying out constraint and optimization adjustment on the candidate boundary position set to obtain an optimized boundary position set; and performing segmentation processing on the to-be-fragmented text sequence to obtain text fragments conforming to semantic logic. The fragmentation method provided by the invention shows good adaptability on corpora in different fields.

Disclosed in the present invention is a physics-informed neural network-based thermo-mechanical coupling analysis method for an inertial microsystem, the method comprising: S1, configuring material parameters and boundary conditions of an inertial microsystem, and establishing a thermo-mechanical coupling analysis model; S2, performing electro-thermal coupling analysis to obtain a temperature distribution of the inertial microsystem; S3, performing thermo-mechanical coupling simulation analysis to obtain a thermal stress distribution of the microsystem; S4, predicting temperature fields of the microsystem by means of a physics-informed neural network; S5, using the temperature fields as boundary conditions for mechanical simulation of the microsystem, obtaining mechanical properties such as stress and strain of the microsystem; and S6, performing electromechanical coupling simulation analysis to analyze the impact of structural deformation on various parameters of electrical performance. The present invention improves the solution accuracy of the neural network by means of an improved adaptive weighting strategy, combines a complete polynomial basis function with the neural network, and introduces an expanded basis function to reduce the state dimensionality, thus reducing computational costs and time, achieving accurate prediction of temperature fields of microsystems at multiple moments, and allowing for computation of the performance of microsystems under electro-thermal-mechanical multi-physics coupling.

The present disclosure discloses a multi-modal brain network calculation method, apparatus, device, and storage medium. The method is configured to train a brain disease prediction model. After the brain region structural feature and the brain region functional feature are separately extracted from magnetic resonance diffusion tensor imaging data and brain functional magnetic resonance data, a graph representation diffusion learning network is used to separate the universal feature and the unique feature in the brain region structural feature and the brain region functional feature. And then, multi-modal universal and unique feature fusion is implemented based on an alignment algorithm and adaptive weighting technology. Thus, complementary information between the multi-modal data is fully mining. The model can learn an effective feature of a related disease in a training process, and a finally obtained brain region disease prediction model has higher precision and better prediction effect.

The invention relates to a multi-modal sparse fusion three-dimensional target detection method in a structured environment, and the method comprises the steps: carrying out the cross-domain self-compensation of point cloud geometric features and image spectral features, and carrying out the fusion of multi-view sparse features; the perception capability and the recognition precision of a long-distance target, a complex shielding target and a small-scale target, and the detection performance and the real-time performance of a long-distance dynamic target are improved; a novel ParScaleNet image backbone network is designed, and a parallel multi-branch structure is introduced to enhance the multi-scale feature representation capability and the channel dependency relationship modeling capability; a staged progressive feature fusion strategy is adopted, hierarchical feature interaction is realized on a fine-grained spatial scale, and a cross-channel adaptive weighting mechanism is combined, so that the network can efficiently extract and focus key features, the feature expression ability of the model is improved, the stability and robustness of detection precision are ensured, and the detection accuracy is improved. And high-efficiency and high-precision target detection in a structured road scene is realized.

The invention discloses an ultrasonic endoscope navigation system and method based on deep learning, and relates to the field of medical image analysis, and the system comprises a parallel encoder module which is used for extracting local features of an ultrasonic endoscope image through a CNN branch, and capturing global context information through a Transform branch; the channel attention fusion module is used for carrying out adaptive weighted fusion on the extracted features; a decoder module that generates an anatomical structure segmentation mask based on the fused feature; the time sequence processing module is used for receiving the frame-by-frame segmentation result output by the decoder module and realizing time sequence coherence analysis of the ultrasonic endoscope video through a bidirectional LSTM network; and the multi-modal fusion module is used for carrying out registration and fusion on the ultrasonic endoscope images. According to the scheme, a high-reliability artificial intelligence auxiliary tool can be provided for early screening of pancreatic cancer, meanwhile, the learning threshold and clinical application cost of the ultrasonic endoscope technology are reduced, and popularization of the ultrasonic endoscope technology in basic medical institutions is promoted.

The invention discloses a multi-base-station AOA cooperative low-altitude target rapid positioning system. The system comprises an AOA measurement base station network, an AOA data preprocessing module, an adaptive weighted intersection positioning module, an extended Kalman filtering state estimation module and a data fusion and system integration module. The system synchronously measures the arrival angle of a target signal through multiple base stations, removes noise in combination with smoothing filtering and an anomaly rejection algorithm, and then solves the initial position of a target by using a self-adaptive weighted intersection algorithm based on measurement quality and geometric distribution. And then, fusing the target motion model and the measurement model by adopting an extended Kalman filtering algorithm to realize dynamic estimation and prediction of the position, the speed and the course. The system can realize high-precision and real-time positioning and continuous tracking of targets such as low-altitude unmanned aerial vehicles, small aircrafts and the like in a complex electromagnetic environment and a sight distance limited scene, and has visual display and regional alarm functions.

The invention relates to the technical field of computer vision, in particular to a high-rise facility operation risk monitoring method based on deep learning and point cloud detection, and the method comprises the steps: collecting a three-dimensional point cloud in real time, and extracting a target point cloud through dynamic threshold denoising and template registration; performing joint coding on space geometry and time sequence motion by using a pre-trained space-time diagram network in combination with an attention mechanism; high-reflectivity beacon points are identified, and the change rate of displacement and angular velocity is calculated; constructing a gating fusion model, dynamically weighting and coupling semantic features and measurement data, and generating risk probability distribution through mutual information consistency check; a fuzzy logic classifier with membership degree optimization is used for mapping to four-level early warning, and grading response is triggered; after early warning, a key frame incremental learning fine tuning model is extracted, and preprocessing parameters are reversely optimized to form a closed loop. According to the method, through multi-source heterogeneous data fusion, dynamic adaptive weighting and a self-evolution mechanism, the real-time performance, accuracy and robustness of risk monitoring in a complex construction environment are remarkably improved.

The invention provides a lightweight video anomaly detection method and system based on information gating compression, and the method comprises the steps: carrying out the feature extraction of a video sequence, and obtaining a high-dimensional semantic feature; calculating an L2 norm, an energy feature vector and a mean feature vector by using the high-dimensional semantic features; enhanced features are obtained through the L2 norm, the energy features, the mean value features and the high-dimensional semantic features; utilizing the enhanced feature vector to obtain a dynamic gating weight; obtaining a value judgment matrix by using an L2 norm; based on the video sequence length, obtaining a dynamic mask through calculation; obtaining a final feature vector through the dynamic mask; obtaining a final abnormal score by using the final feature vector; and obtaining a detection result through the final abnormal score. According to the method, by introducing an attention mechanism and an amplitude-based gating strategy, the non-key feature flow is adaptively weighted or cut off, and the model is guided to focus effective information.

The invention relates to an MT-InSAR self-adaptive fusion measurement method and device for long-term surface deformation of a coal mining subsidence area, and the method comprises the steps: obtaining surface deformation monitoring data in a plurality of modes, and carrying out the at least one of time-space alignment, noise removal and error correction of the surface deformation monitoring data, so as to obtain surface deformation data; calculating coherence indexes of the earth surface deformation data; generating a space adjustment coefficient by using a double-peak Gaussian function model constructed based on a coal mining subsidence area deformation rule; constructing an adaptive weighting method based on the coherence index and the spatial adjustment coefficient of the earth surface deformation data to calculate the weight of the multi-source InSAR data; and performing weighted fusion on the surface deformation monitoring data by using the weight of the multi-source InSAR data to obtain a final surface deformation result. Therefore, the problems of insufficient monitoring precision, incomplete coverage, poor adaptability and the like due to the fact that a single InSAR technology is difficult to comprehensively reflect spatio-temporal evolution characteristics of long-term surface deformation of the wide-area coal mining subsidence area are solved.

The invention relates to the technical field of image processing, in particular to an intelligent recognition and positioning system for battery support welding defects. Obtaining defect degrees and edge pixel points of the pixel points according to gradient features and neighborhood gray features of the pixel points in the welding image; obtaining change complexity according to the change characteristics of the local edges of the edge pixel points; edge sharpness is obtained according to the local gradient features and the gray level change features of the edge pixel points; obtaining local uniformity according to the local gray level distribution characteristics of the edge pixel points; obtaining a confidence coefficient according to the change complexity, the edge sharpness and the local uniformity; and obtaining an adjustment factor of the edge pixel point according to the confidence coefficient and the defect degree. The method comprises the following steps: adjusting a weighting coefficient in a high-lift filtering algorithm according to an adjustment factor to obtain a self-adaptive weighting coefficient; performing image enhancement and defect identification according to the adaptive weighting coefficient; and the accuracy of image enhancement and defect identification is improved.

The invention relates to the technical field of industrial mechanism models, in particular to a shock absorber performance optimization control method based on model fusion, which comprises the following steps: extracting a low-frequency disturbance state variable and inputting the variable-topology industrial mechanism model to generate a nominal reference state trajectory; utilizing a depth state observation network fused with energy passivity constraint to calculate a non-linear model mismatch compensation amount and an adaptive weighting parameter; performing dynamic fusion on the nominal reference state trajectory and the compensation amount based on the adaptive weighting parameter to generate a generalized state estimation value; and executing dynamic multi-objective optimization based on the generalized state estimation value, and generating mixed mode control input acting on an execution end. According to the invention, through adaptive fusion of a mechanism model and a data driving method, physical consistency, calculation real-time performance and robustness of a control process are considered.

The invention discloses a high-precision real-time positioning system and method for positioning personnel based on multi-mode fusion. The system comprises a multi-source data acquisition module, an intelligent data processing module, a fusion positioning model construction module and a positioning result application module, provides visual monitoring, trajectory analysis and scene customization functions, and supports flexible expansion through modular design. According to the method, satellite positioning, Bluetooth beacon and inertial navigation technologies are integrated, the weight is dynamically adjusted in combination with an adaptive weighted fusion algorithm, and a fusion positioning model supports outdoor (satellite dominant + EKF calibration, the precision being 2-5 meters) and indoor (beacon dominant + fingerprint matching, the precision being lt); according to the invention, intelligent switching between three modes (inertial navigation + ZUPT correction) and signal loss (1 meter) is realized, data noise is optimized by using Kalman filtering, and high-precision real-time positioning in indoor and outdoor complex scenes is realized. According to the invention, the problems of insufficient precision, poor environmental adaptability and weak expansibility of a single positioning technology are solved, and the comprehensive performance of the positioning system is significantly improved.

The invention provides a multipath fusion phased array radar antenna unit coordinate determination method and system, and the method comprises the steps: obtaining an antenna array surface power distribution diagram, recognizing the boundary and geometric center of an array, and planning a spiral and cross-shaped composite sampling path; on the planned sampling path, collecting phase data at a plurality of inclination angles and frequencies; based on the signal-to-noise ratio and the angle correction factor, carrying out adaptive weighted fusion calculation on the adjacent point phase difference under multiple conditions; establishing an overdetermined equation set by using the fused phase difference, and iteratively solving the accurate coordinates of the antenna unit through two rounds of weighted least squares; and establishing a drift model to perform dynamic compensation, and performing real-time calibration by using a unit with determined coordinates as a reference point. According to the invention, through a multi-angle and multi-frequency redundancy measurement strategy and in combination with a signal-to-noise ratio-based adaptive weighting algorithm, environmental interference can be effectively identified and suppressed. Even in a complex electromagnetic environment, the system can still keep stable measurement precision.

The invention discloses an urban particulate matter migration path identification method based on data fusion of a dynamic diffusion model and multi-source sensing. According to the method, firstly, multi-source pollution related data of a fixed monitoring station, a mobile monitoring device, a meteorological observation node and a traffic sensing system are collected, and standardized input is constructed after time synchronization, coordinate mapping and exception handling. And then establishing a hybrid dynamic diffusion model fusing a two-dimensional Gaussian plume model and a Lagrange particle tracking mechanism, and realizing assimilation of monitoring data and model output in combination with an improved particle filtering algorithm to obtain a correction concentration field. Based on concentration gradient analysis and particle trajectory superposition, a pollution migration path is extracted, then indexes such as a migration intensity index (MII), a path stability factor (PSF) and path average correlation are calculated, and recognition and sorting of a migration direction, a pollution source position and path stability are achieved. Meanwhile, an alpha dynamic weight factor is introduced, adaptive weighting is realized between the Gaussian model and the LPDM model, and the path strength and consistency are considered. Experimental results show that the method is superior to a traditional method in average absolute deviation (MAD) and hot spot overlap ratio (IoU) indexes, diffusion and migration rules of particulate matters under complex urban conditions can be more accurately revealed, and the method has important environmental governance and emergency management application value.

The invention discloses a multi-mode water-air-land cross-medium cooperative sensing system and method based on an MEMS (Micro Electro Mechanical System). The system comprises an underwater robot, a buoy station, an unmanned aerial vehicle and a ground control station, each node is provided with an MEMS sensor integrating acoustics, optics, magnetism, inertia and other modes, and data transmission and relay are achieved through acoustics, optical communication and radio links. And the ground control station carries out fusion processing, target identification and decision control on multi-source data through time synchronization and space registration, and issues a cooperative instruction to each node to realize a perception-decision-re-perception closed loop. The method comprises five steps of node deployment and calibration, multi-modal data acquisition, hierarchical transmission and sharing, adaptive weighted fusion and closed-loop cooperative control. According to the invention, global coverage, high-precision and multi-mode environment and target perception can be provided in a complex water-air-land environment, and the detection reliability and the communication efficiency are significantly improved.

The invention belongs to the technical field of aviation meteorological monitoring, and relates to an airport dominant visibility automatic measurement method and system based on a three-dimensional laser radar, and the method comprises the steps: carrying out the full-space scanning of an airport through the laser radar, so as to collect echo signals; preprocessing and calibrating the echo signal; performing atmospheric extinction coefficient inversion on the preprocessed echo signal based on an inversion algorithm to obtain an atmospheric extinction coefficient, and then converting the atmospheric extinction coefficient into meteorological visibility; carrying out quantile statistics on all meteorological visibility in a period, judging a dominant visibility value, and carrying out adaptive weighted fusion on the dominant visibility value, the visibility value measured by the point-type visibility meter and the visibility value observed manually to obtain a fused dominant visibility value. The system can break through the limitation of traditional manual and point type visibility meter observation, and realizes accurate and automatic measurement of all-region, all-weather and unattended dominant visibility of an airport.

The invention relates to a semi-supervised remote sensing image classification method based on hierarchical crossing, and belongs to the technical field of remote sensing image processing and computer vision. The method comprises the following steps: firstly, constructing a representation learning module, adopting a hierarchical cross pseudo-tag generation mechanism, randomly combining different hierarchical features in a main model and an index moving average (EMA) model to construct a cross model, and generating a high-quality pseudo-tag for a weakly enhanced image; secondly, a self-adaptive weight mechanism is introduced, the label loss-free contribution is dynamically adjusted in combination with the training progress and the pseudo label utilization rate, and the self-adaptability of the model in different training stages is enhanced; and finally, designing a label alignment strategy based on a training stage, adjusting pseudo label category distribution through a time decline function, guiding the model to pay more attention to minority categories at the initial stage, and improving small sample category performance. Through hierarchical cross random combination, an adaptive weighting mechanism and a label alignment module, pseudo label quality, training stability and minority class performance are improved.

The invention discloses an image region analysis method based on entropy driving feature enhancement, and relates to the technical field of image analysis and feature enhancement. According to the method, the local channel information entropy is used as a core feature statistical magnitude, and adaptive weighting and strengthening of different importance region features are realized through explicit quantification of image feature information amount; weight distribution is dynamically adjusted according to the characteristic values, the response of a high-information dense area is remarkably enhanced, and meanwhile low-information and noise interference areas are effectively restrained. On the basis, a cross-layer attention mechanism based on entropy prior is designed, feature statistical information is embedded into a gating and weight generation process, and attention distribution with feature significance as guidance is achieved. According to the method, through an entropy-driven adaptive feature enhancement mechanism, the perception capability, the feature discrimination capability and the analysis precision of the model on the salient region of the image are effectively improved.

The invention relates to the technical field of power system fault diagnosis, and discloses a flexible DC power distribution network fault diagnosis method and system. The method comprises the steps of collecting a fault current time sequence signal of a direct current side of the flexible direct current power distribution network, inputting the fault current time sequence signal into a pre-trained fault diagnosis model, and performing adaptive time frequency feature extraction on the fault current time sequence signal by using an interpretable complex time frequency convolution layer to obtain a time-frequency-amplitude tensor, a time-amplitude trajectory, a time-frequency trajectory and a frequency-amplitude trajectory are constructed, the trajectories are subjected to Gramer difference angular field coding, a three-channel time-frequency feature matrix is generated, the three-channel time-frequency feature matrix is input to a parallel double-branch feature extraction module, and global features and local features are extracted; and carrying out adaptive weighted fusion on the extracted global features and local features, classifying the fused features, and outputting a diagnosis result of a fault category. According to the invention, the accuracy of fault diagnosis of the flexible DC power distribution network is improved.

The invention discloses a distribution network-oriented traveling wave fault location centralized analysis system, which relates to the technical field of power system fault location and comprises a centralized master station and a plurality of monitoring terminals. The monitoring terminals are deployed at different node positions of a power distribution network, and synchronously collect fault traveling wave signals and upload the fault traveling wave signals to the centralized master station. The centralized master station comprises a topology modeling module, a data preprocessing module, a wave head identification module, a candidate branch screening module, a matching evaluation module and an accurate positioning module. According to the invention, through establishing a full-process quality tracking and adaptive weighting mechanism from preprocessing to accurate positioning, a system can adjust a positioning strategy according to a distribution network topology structure and a signal state based on a two-dimensional comprehensive evaluation system of topology sensitivity and signal quality; high fault positioning accuracy and reliability are kept in a complex environment of the distribution network, and the robustness and practicability of the system are improved.

The invention discloses a TCN-SVM rolling bearing fault diagnosis method fusing an SE attention mechanism, and belongs to the technical field of mechanical fault intelligent diagnosis. Aiming at the problems of feature redundancy, noise sensitivity, insufficient Softmax classifier generalization and the like existing in a traditional time sequence convolutional network (TCN), the invention provides a solution for collaborative optimization of a deep network and a support vector machine. The method comprises the following steps: acquiring a vibration signal of the rolling bearing; constructing a multi-fault sample set, and processing an original signal; constructing an SE-TCN feature extraction network, capturing multi-scale time sequence features by adopting expansion causal convolution, and embedding an SE module into a residual module to realize channel adaptive weighting; and a support vector machine (SVM) classifier decision function is constructed, and fault classification is completed through the RBF kernel SVM. Experiments show that the method has high fault recognition accuracy and robustness, and the problem of confusion of composite fault features of the rolling bearing is effectively solved.