138 results about "Feature modeling" patented technology

This application discloses a landslide segmentation method, apparatus, equipment, and medium based on hybrid Mamba and frequency domain calibration, relating to the field of landslide disaster detection technology. The method involves fusing optical satellite imagery with slope and lithological topographic constraints, preprocessing the data, and then inputting it into a model containing a hybrid Mamba encoder, a multi-scale adaptive gating module, and a progressive frequency domain calibration decoder for processing. Through multi-stage hybrid coding, contextual aggregation modulation, and progressive resolution recovery, accurate landslide segmentation is achieved, effectively improving global feature modeling capabilities, enhancing the identification accuracy of multi-scale and boundary-ambiguous landslides, balancing lightweight design and robustness, and better adapting to the actual monitoring needs of highway landslides.

The application discloses an infrared thermal imaging power line circuit breaker fault diagnosis method and system based on YOLOv13, innovatively improves the original YOLOv13 model through module replacement, structure addition and characteristic adaptation, carries out light feature extraction through an infrared multi-scale temperature perception residual module, and outputs an enhanced infrared feature map; the infrared gradient guide deep convolution module is used for amplifying fault edge features, and an edge clear feature map is output; the infrared cross-scale temperature fusion module is used for processing the aligned features to generate a feature correlation graph; the MobileViT module is combined with the above-mentioned infrared special module in depth, a light-weight feature modeling system suitable for infrared fault diagnosis is constructed, a head detection output network is combined with an infrared adaptive multi-task loss function, and the accuracy of fault category and position prediction is ensured. Through the above technical improvement, the application provides a new scheme which is accurate, efficient, light-weight and highly adaptive for infrared thermal imaging power line circuit breaker fault diagnosis.

This invention relates to the field of industrial visual defect detection, and more particularly to a progressive pseudo-label optimization-based weakly supervised detection and segmentation method for industrial defects. The method includes the following steps: acquiring images of industrial surface defects; training a basic network and simultaneously initializing a teacher model and a student model; supervising the training of the student model; simultaneously inputting the industrial surface defect images into the teacher and student models to obtain initial pseudo-labels and feature maps; extracting feature centers and constructing a memory bank; constructing a contrastive learning loss; constructing a Gaussian distribution model to calculate pixel-level feature assignment probabilities; fusing these probabilities with the initial pseudo-labels to generate an enhanced pseudo-mask; and processing this to obtain a clean pseudo-label; training and updating the student model while simultaneously updating the teacher model; and outputting the optimized student and teacher models after multiple rounds of closed-loop optimization. This invention, through pseudo-label optimization and refined feature modeling, reduces the labeling cost of industrial defect detection and significantly improves detection and segmentation accuracy.

This invention relates to the field of intelligent information processing technology and discloses an intelligent assessment system for endometrial receptivity based on biosensor data processing. The system acquires signals of multiple types of biomarkers in bodily fluid samples through multi-channel biosensor detection and performs structured preprocessing and quantitative transformation. Based on this, it establishes object-level associations between biosensor data and multi-source information to construct a unified feature data system. Furthermore, it constructs a weighted heterogeneous assessment graph and introduces a manifold stochastic feature modeling mechanism to generate manifold feature data. The manifold features and original features are jointly input into a shared-parameter cyclic Transformer model. Through cyclic inference and consistency constraints, multi-round feature fusion is achieved, outputting the endometrial receptivity assessment result. This invention improves the multi-source data modeling capability and the stability and consistency of the assessment process.

This invention discloses an individual motion behavior analysis system based on dynamic modal configuration and rule-guided fusion, belonging to the fields of intelligent perception, sports science, and human-computer interaction. It includes a data input module, a dynamic modal organization and configuration module, a multimodal feature modeling and fusion module, an indicator-level evaluation and behavior state prediction module, a statistical rule-guided analysis and result correction module, and a structured result generation and output module. These modules work together to form a complete closed loop for motion behavior analysis technology. This invention aims to address the problems in existing motion behavior analysis technologies, such as limited dimensionality of single-modal information, rigid multimodal fusion methods, insufficient stability and interpretability of analysis results, and strong dependence on large-scale manually labeled data.

The application belongs to the technical field of marine environment expansion, and discloses a shipborne radar sea wave perception domain expansion method and system based on deep learning, which constructs a sea wave perception domain step-by-step expansion model, uses a sea wave perception domain expansion submodule to realize spatial feature modeling and limited range expansion of local observation data, and gradually deduces larger range sea wave field information through a step-by-step expansion and residual connection mode; meanwhile, a sea wave continuity constraint mechanism is introduced to apply continuity constraints at the junction of the expanded sea area and the original observation sea area, so that the overall wave field can maintain high accuracy and spatial continuity in a large range. Through the shipborne radar sea wave perception domain expansion method based on deep learning, the real-time perception ability of the shipborne radar to a large range of sea areas in the local environment can be significantly improved, the independent operation of the ship in the marine environment is supported, and the real-time sea wave perception ability of the ship navigation safety and the marine engineering application is significantly improved.

The application provides a remote intelligent operation monitoring method and system of a smart substation, relates to the technical field of intelligent operation and maintenance of a substation, and relates to multi-source heterogeneous sensing, deep feature modeling, fault prediction evaluation and model self-optimization.The application collects electrical, environmental and meteorological data through heterogeneous sensors, constructs a structured original data set, carries out time series prediction in combination with a convolution-LSTM model, and outputs a fault probability and a confidence interval by using a Transformer fusion network, so that online early warning and response control are realized, and the application has the self-adaptive updating capacity driven by federated learning.

The application discloses a cancer survival prediction method and device based on whole-slice pathological images, and relates to the technical field of medical artificial intelligence and digital pathological image analysis. The method comprises the following steps: acquiring whole-slice pathological images and patient survival information, and performing pretreatment; performing cell detection on the whole-slice pathological images to obtain cell structure data; dividing the whole-slice pathological images into a plurality of image regions, and determining regions of interest; screening the cell structure data in each region of interest to construct a cell subset; sorting the cell structure data according to the distribution relationship of cells in space to construct a cell structure sequence; constructing a cell structure modeling model, performing feature modeling on the cell structure sequence, and obtaining tissue structure features capable of representing the microstructure of tumor tissues; and inputting the tissue structure features into a survival prediction model to predict the survival risk of the patient. The application can improve the accuracy and stability of cancer survival prediction.

The application provides a water conservancy and hydropower construction scheme generation method based on construction object feature modeling, comprising the following steps: step 1) construction object feature identification and modeling; step 2) construction scheme generation constraint construction: combining engineering specifications, historical schemes, expert experience and other data, a set of multiple source heterogeneous constraint conditions is constructed, and the set of constraint conditions is in a structured form to participate in the combination and generation of subsequent construction scheme content; step 3) combination and generation of construction scheme content: the construction scheme is disassembled into multiple controllable content units, and combination and generation are performed according to different constraint conditions in the generation process. The method adopts a content generation mechanism of "constraint driving + content combination", realizes unified support of a set of technical schemes for multiple construction scenes, and significantly improves the application range of the technical scheme.

This invention discloses a visual BEV perception method based on implicit and explicit dual-path height feature collaboration, comprising: real-time acquisition of raw image data from multiple perspectives around the vehicle via an onboard surround-view system, followed by inputting the data into an image encoding network for feature extraction to obtain multi-scale feature images; processing the multi-scale feature images through a vertical height feature modeling unit and a horizontal feature modeling unit to obtain a BEV feature representation containing three-dimensional spatial information; wherein the vertical height feature modeling unit adopts a dual-path parallel architecture with explicit height modeling branches and implicit height modeling branches; and inputting the BEV feature representation into a target detection output module to obtain the target detection result. This invention, without relying on external depth sensors such as LiDAR, effectively improves the three-dimensional spatial perception of pure vision systems through visual input, significantly enhancing the ability to express vertical dimension features, thereby achieving more accurate and reliable three-dimensional target detection in complex scenes.

This invention discloses a multi-level network-type energy storage control and protection system and method, relating to the field of energy storage technology. The method includes: S1, performing feature modeling on the multi-source operating signals of the acquired energy storage system to construct a multi-source operating feature set; S2, performing time series rate of change analysis on the multi-source operating feature set to extract dynamic features of current direction change and power reversal; S3, performing energy disturbance amplitude analysis on the dynamic energy feature set to calculate an adaptive protection threshold set in real time; S4, comparing the real-time sampled signal with the adaptive protection threshold set to identify whether an over-limit state has been triggered; S5, when an over-limit state is detected to be triggered, extracting the real-time dynamic features of the over-limit event, matching the real-time dynamic features of the over-limit event with the energy backlash judgment model, determining the anomaly type of the over-limit event based on the matching result, and outputting a protection action command or dynamic balance control command corresponding to the anomaly type.

The application belongs to the technical field of image processing, and particularly relates to an image missing area repairing method and system based on multi-scale mixed feature modeling. The original image to be repaired and a missing area mask are spliced in the channel dimension, multi-layer normalization and convolution operations are performed on initial features, feature extraction is performed on a content channel branch and a gate channel branch, multi-layer convolution and linearization operations are performed to obtain channel gate features of each layer; the channel gate features of the last layer are split into spatial detail component features and long-range correlation component features, spatial detail components and long-range spatial components are obtained through linear mapping, and fusion output features of each layer are obtained through processing; and the fusion output features of the first layer are used as a repaired complete image. The scheme can guarantee the structural continuity of a large-area missing area, and effectively reduce blurring, artifacts and boundary fracture phenomena.

The application discloses a kind of spatiotemporal modeling ground surface temperature downscaling method and system considering energy constraint, comprising: 1) acquisition and preprocessing of multi-source remote sensing and meteorological data;2) feature grouping and spatiotemporal feature tensor construction;3) time feature and spatial feature extraction;4) spatiotemporal feature modeling and high-resolution ground surface temperature estimation;5) loss construction and cross-scale consistency constraint;6) model iterative training and result output.The application breaks through the modeling limitation of time variation law and spatial detail description in the prior art, realizes the collaborative promotion of time continuity and spatial fine expression of ground surface temperature.On this basis, the overall temperature deviation and the problem of insufficient physical rationality commonly existing in the prior art are also effectively avoided, and the application is significantly superior to the prior art in terms of timing stability, spatial accuracy and physical reliability.

The application discloses a kind of outbound line decision-making method and system based on incoming call display perception, the method includes: determining outbound task and obtaining the number attribute of outbound line, the number attribute and outbound task are modeled to obtain number feature vector and outbound service feature vector;Based on the willingness prediction network of pre-set, according to number feature vector and outbound service feature vector, the willingness score sequence corresponding to outbound line is obtained by prediction;For any outbound line, according to the constraint condition of willingness score and line cost, the target outbound line executed by outbound task is determined by calculating through the pre-set resource allocation algorithm;The feedback sample of target outbound line corresponding outbound task is obtained, and the willingness prediction network and resource allocation algorithm are continuously updated by hybrid learning mechanism.It can be seen that the application can realize the intelligent selection and dynamic optimization of outbound line, so as to effectively improve the reach rate of intelligent outbound system.

This invention provides a multi-dimensional user profiling automated segmentation and precise targeting system and method, aiming to solve the problems of single user feature dimensions, lagging updates, and inaccurate matching in existing advertising targeting. The method includes: collecting multimodal data such as user text, behavior, and images; extracting feature vectors from each data source using a model and fusing them into a unified high-dimensional representation; inputting this data into a multi-label neural network to output user interest, behavior, and intent tags; using an improved clustering algorithm to form a strategic audience package; and combining this with ad placement and resource allocation for ad scheduling, while simultaneously updating the user profile and model through a closed-loop user feedback mechanism. This method features innovative aspects such as multi-source heterogeneous feature modeling, minute-level tag updates, and tag-driven clustering and ad delivery, significantly improving audience identification accuracy and ad ROI performance.

This application discloses an image processing method, apparatus, device, and storage medium, relating to the field of computer vision technology. The method involves inputting image data into a feature modeling network for feature extraction to obtain a feature map; then, target region detection is performed based on the feature map to determine the target location. The feature modeling network includes a local distortion modeling branch and a global state space modeling branch that receive the same input features. The local distortion modeling branch processes the input features based on convolution and deformable convolution to generate a spatial offset prior field and local enhancement features. The global state space modeling branch processes the input features based on a state space model and introduces a spatial offset prior field to dynamically modulate the global state space modeling process, generating global context features. This application can solve the problem in related technologies where traditional convolutional neural networks based on regular receptive fields struggle to stably extract target features from images with distortion or stretching.

The application provides a high-speed railway infrastructure detection data processing method and device, and belongs to the high-speed railway data processing field. The high-speed railway infrastructure detection data processing method comprises the following steps: acquiring detection data, wherein the detection data comprises structured data and unstructured data; performing mile calibration on the structured data by using a dynamic functionalization feature modeling alignment method; performing data cleaning on the detection data; constructing an intermediate data model for mixed storage of the structured data and the unstructured data, and performing mile correction on the unstructured data; constructing a data quality scoring model for scoring the detection data; and storing the detection data through the intermediate data model when the score of the detection data meets the requirements. The method optimizes the mile correction method of the detection data, and improves the mile correction accuracy of the detection data. Meanwhile, the intermediate data model is constructed for storing the detection data, and the problem of multi-source data fusion difficulty can be overcome.

This invention relates to the field of air-ground collaborative path planning technology, specifically to a method, system, and storage medium for air-ground collaborative task allocation and path planning decision-making in unknown environments. It constructs a collaborative decision-making system consisting of a UAV task allocation module, a meta-learning optimization module, and an unmanned vehicle path planning module. During execution, the path is dynamically corrected based on environmental change information continuously transmitted back by the UAV. This invention effectively coordinates conflicting objectives such as path distance, time window satisfaction, and load balancing through a multi-objective decomposition and dynamic attention fusion mechanism. Leveraging the real-time information transmitted back by the UAV and the rapid fine-tuning capability of meta-learning, it significantly enhances the online response speed to sudden obstacles and environmental changes. Simultaneously, this system overcomes the shortcomings of data lag and spatiotemporal alignment error accumulation in heterogeneous platforms through structured feature modeling and closed-loop planning processes, achieving efficient collaboration from perception to execution.

The invention belongs to the technical field of aircraft control, and particularly relates to a simulation test system and method for an aircraft obstacle avoidance algorithm. The invention discloses a simulation test system and method for an aircraft obstacle avoidance algorithm. The system comprises a virtual environment modeling device, a dynamics resolving device, an obstacle avoidance logic processing device, a physiological feature modeling device, a stress parameter mapping device, a comprehensive evaluation feedback device and a data monitoring device. According to the method, kinematic parameters are mapped into physiological stress parameters such as gravity overload and motion sickness, and an obstacle avoidance track is fed back and corrected according to a physiological comfort threshold, so that an obstacle avoidance path conforms to the tolerance limit of a human body on the premise of ensuring safety. According to the method, deep fusion of control engineering and aeronautical medicine is realized, obstacle avoidance assessment extends to the field of physiological tolerance from physical safety, the scientificity of simulation assessment of the manned aircraft is improved, and the risk of personnel disability caused by violent maneuver in real flight is reduced.

The application discloses an industrial vehicle identity recognition and unlocking system based on image recognition and fingerprint comparison, comprising: a multi-modal biological feature modeling module for collecting face images, voice prints and fingerprints to generate identity potential vectors; an identity verification and unlocking control module for performing authorization comparison based on the face, the fingerprint and the identity potential vectors and outputting an unlocking instruction; a running track construction module and a baseline track generation module for constructing a running track set and a baseline track set to obtain a differential feature sequence; a decomposition reward structure construction module and a MaxEnt IRL learning module for constructing a basic reward and a personalized reward to establish an improved MaxEnt IRL model; a time sparse strategy solving module and an instrument control execution module for solving a time sparse instrument control strategy and generating an instruction after identity verification is passed. The application improves the industrial vehicle enabling safety and the effectiveness of alarm decision in the running process.

The application discloses an industrial production anomaly monitoring system based on machine learning, comprising: a data acquisition and preprocessing module for acquiring image and sensor data and constructing a training data set; a knowledge graph construction module for establishing a graph and mapping it into semantic constraint features; an image segmentation network construction module for constructing a network output probability distribution graph based on Swin-Unet; a model training and anomaly mask generation module for designing a hybrid loss function to train a model and generate a mask; a time series feature modeling and state updating module for introducing a Bayesian adaptive factor to dynamically update a state transition matrix of an improved GMM-HMM model; a state evolution identification module for decoding an optimal state sequence to identify an evolution mode; and an anomaly final determination module for combining sensor data to check an output final determination result. The application improves anomaly identification precision and evolution trend early warning capability.

This invention relates to the field of data processing, providing a highly efficient generative anonymization method and system for big data. The method includes: preprocessing and feature modeling the original large dataset to obtain a feature vector set; constructing privacy risk indicators and task utility indicators based on the configuration parameters of the target sharing scenario, respectively, for the privacy leakage degree of the generated data and the performance of downstream tasks, resulting in a privacy utility quantification index system; inputting the feature vector set and the privacy utility quantification index system into a pre-built model, training the pre-built model through a multi-objective optimization loss function to obtain a generative anonymization model; inputting the scenario configuration vector into the generative anonymization model to generate a target anonymized dataset; evaluating the target anonymized dataset according to the privacy utility quantification index system; and outputting the evaluated target anonymized dataset to a data application system. This invention balances privacy security and data utility, improving scenario adaptability.

This invention relates to the field of energy storage anomaly monitoring technology, and more particularly to a deep learning-based energy storage anomaly monitoring method and system. The method involves acquiring operational data from the historical stable operation of an energy storage system to construct a stable operation dataset. Joint feature modeling is then performed on this data to generate operational behavior embedding vectors and form operational behavior sequences. Furthermore, an autoencoder neural network model is used to encode the features of the operational behavior sequences, learning the behavioral change patterns under stable operating conditions of the energy storage system and constructing a normal behavior trajectory vector. Real-time operational data is then input into the model to generate the current behavior embedding vector, and its behavioral offset feature vector relative to the normal behavior trajectory vector is calculated. Anomaly determination is then performed based on the continuous offset of this feature within a continuous time window. This allows for early identification of abnormal states before operating parameters exceed threshold ranges, improving the accuracy and timeliness of energy storage system anomaly monitoring.

The application belongs to the field of CT tomographic reconstruction and the field of artificial intelligence, and discloses an industrial battery CT image super-resolution method based on deep learning. The method takes a low-resolution tomographic image as input, constructs an end-to-end battery CT image super-resolution network, and outputs a high-resolution tomographic image. The network uses multi-scale attention module (MSA) and enhanced grouping module (EGM) for parallel calculation, adopts hybrid channel attention feed-forward module (HCFB) for nonlinear enhancement, and combines a frequency modulation module for frequency domain feature modeling. The method can obtain a battery CT tomographic reconstruction result with clearer boundaries, more coherent layer structures and fewer artifacts, and provides a more stable and reliable data basis for subsequent defect detection, structure segmentation and three-dimensional analysis.

This invention belongs to the field of smart city information technology, specifically relating to an adaptive operation method for industrial robots in elevator installation and maintenance. It covers the entire process of industrial robots, from elevator scene perception and operation planning to safe execution. Employing core algorithms such as multi-dimensional feature modeling of the elevator car / shaft, scenario-based adaptation of installation and maintenance operation trajectories, and dynamic early warning of human-machine-equipment safety interaction, it overcomes the technical bottlenecks of traditional elevator operation robots, which suffer from poor scene adaptability, low operation accuracy, and insufficient safety control. This achieves accurate perception and modeling of elevator scenes, dynamic optimization and adaptation of operation trajectories, and real-time early warning and control of safety risks, significantly improving the efficiency, accuracy, and reliability of elevator installation and maintenance operations. It is suitable for the full-scenario operation needs of various types of elevators, including residential, commercial, and high-speed elevators, for installation, commissioning, inspection, and maintenance.

The application discloses a kind of big data dynamic path intelligent planning system and method for instant distribution, its system includes multi-source real-time sensing module, big data fusion processing module, distribution demand prediction module, multi-objective dynamic optimization scheduling module, path reconstruction and adaptive updating module and feedback learning module.The application realizes the quick identification and path dynamic reconstruction to traffic congestion, order change and emergent event by constructing multi-source real-time sensing and feature deviation trigger mechanism, avoid distribution personnel to follow invalid path, simultaneously, to traffic, meteorology, road risk and driver load multi-source heterogeneous data are unified feature modeling and weight adaptive fusion, and combined with local path repair strategy, only abnormal section is re-planned, significantly reduce computational complexity and improve system stability in large-scale concurrent scene.

The application provides a real-time feature point extraction and matching method based on star operation and GateMLP, which comprises the following steps: S1, preprocessing the input image and inputting the image into a lightweight convolutional network to extract initial features; S2, based on the output feature map, deep feature modeling is performed through the structure of the fusion of a convolution branch and a star operation module, and a dense descriptor and a feature point heat map are output; S3, non-maximum suppression is performed on the heat map, and key point coordinates are screened in combination with a confidence score; S4, cosine similarity and a nearest neighbor strategy are adopted to coarsely match the key points of left and right views, and an initial matching pair is obtained; and S5, the matching descriptor obtained through coarse matching is input into a gated multilayer perceptron for fine-grained matching, and accurate matching point pairs are output. The method has good real-time performance and robustness while ensuring high extraction and matching accuracy, and is suitable for practical scenes such as unmanned aerial vehicle navigation and SLAM that require fast response.

The small sample point cloud target classification method based on LSHW-Mamba network relates to the technical field of 3D point cloud processing and deep learning, and solves the problems that the existing technology focuses on feature mining in the Euclidean space domain, and under the condition of insufficient target samples, the feature representation ability of the target geometric structure is insufficient, and it is difficult to capture the key information for distinguishing similar objects. The present application explicitly analyzes the spectral response of the local geometric structure of the point cloud through the spherical harmonic function, combines the local spherical harmonic wavelet transform to realize the geometric feature extraction of the point cloud, simultaneously introduces the geometric gating mechanism to optimize the feature propagation efficiency of the Mamba model, models the target features in parallel with the Transformer attention branch, considers the relevance of the local and global features of the target, and improves the discrimination ability of the network model for the target under the small sample scene. The present application effectively solves the core problems of insufficient feature discrimination and decreased classification accuracy under the small sample condition.

The application belongs to the technical field of aircraft control, and particularly relates to a simulation test system and method of an aircraft obstacle avoidance algorithm. The system discloses a simulation test system and method of an aircraft obstacle avoidance algorithm, and the system comprises a virtual environment modeling, a dynamics solving, an obstacle avoidance logic processing, a physiological characteristic modeling, a stress parameter mapping, a comprehensive evaluation feedback and a data monitoring device. The system maps kinematic parameters into physiological stress parameters such as gravity overload and motion sickness, and corrects the obstacle avoidance trajectory according to the physiological comfort threshold feedback, so that the obstacle avoidance path meets the human tolerance limit under the premise of safety. The application realizes the deep integration of control engineering and aviation medicine, extends the obstacle avoidance evaluation from physical safety to the physiological tolerance field, improves the scientificity of manned aircraft simulation evaluation, and reduces the risk of personnel disability caused by violent maneuvering in real flight.