39 results about "Global topology" patented technology

A distributed inference communication method of a super-large-scale graph neural network comprises the following steps: dividing a global topology to be processed into a plurality of local sub-domains matched with the number of GPUs, and determining a local master node and a boundary master node allocated to each GPU; calculating coloring priorities of nodes, and performing parallel graph coloring based on the coloring priorities to construct a global independent set sequence; mapping the global independent set sequence to the local sub-domains of the GPUs to generate a local task slice sequence, which comprises boundary node slices and internal node slices, evaluating communication load and calculation load of the slices, and taking the internal node slices as calculation fillers to be executed in the communication waiting period of the boundary node slices; and each GPU executes an inference task by using a multi-stream asynchronous pipeline constructed locally according to the local task slice sequence. The method can improve the throughput of distributed inference.

The application relates to an outdoor geographical semantic retrieval enhanced navigation and search method and system, which comprises the following steps: receiving a natural language instruction, and extracting navigation target semantic elements and search target semantic elements; under the constraint of a structured index, performing beam search, and outputting the latitude and longitude and boundary information of the entity; decomposing the natural language instruction into a subtask sequence, and under the constraint of a predefined executable action and condition space, generating a structured behavior tree; calling a routing engine, generating a global topology path and a waypoint sequence based on the nearest reachable boundary point of the entity; based on the scheduling of the structured behavior tree, performing real-time navigation along the waypoint sequence; after reaching the vicinity of the target, searching for the target related to the search target semantic elements and performing clustering positioning under the constraint of a polygon boundary, obtaining a final target point and converging navigation. Compared with the prior art, the application has the advantages of high positioning accuracy and wide application scenarios.

This invention provides a stylizable and deformation-resistant QR code encoding and recognition method based on topological coding, comprising: converting the binary string to be encoded into natural numbers using a bijective function and recursively decomposing it to generate a topological tree structure; determining the spatial allocation ratio based on the tile traces of each node in the topological tree; performing inward shrinking processing on the polygonal region carrying the QR code, recursively dividing it into non-overlapping closed sub-regions according to the ratio, mapping the hierarchical structure of the topological tree to a QR code with a nested structure and stylizing it; scanning the QR code image and performing preprocessing, extracting the boundary contours of the closed regions and constructing a global topological tree, enumerating multiple candidate topological trees and decoding them one by one, and filtering out the valid binary information as the recognition result after CRC verification. This invention achieves the deformation resistance of the QR code through the mapping of the topological tree structure and nested regions, and improves its visual effect and application flexibility through stylization processing.

This invention discloses a gene regulation inference method guided by topological data analysis and gene network embedding. It combines TDA and GNN to enhance the inference capability of gene regulation networks. By capturing the topological structure of the gene regulation network graph through TDA features, the model's ability to model gene expression is enhanced. The TDA features and GAT embedding representations are effectively integrated through gating fusion. This fusion mechanism enables the model to adaptively adjust node embeddings based on global topological characteristics, which not only improves the accuracy of gene interaction representation but may also enhance the accuracy of regulatory relationship prediction. The traditional GAT architecture is extended through a four-layer graph attention mechanism. Each layer uses residual connections to alleviate the gradient vanishing problem and improve training stability. In addition, independent multilayer perceptron branches are designed for transcription factors and target gene embeddings. This deep architecture can achieve more expressive feature transformations and capture subtle patterns in gene regulation networks.

The application relates to the technical field of power grids and discloses a power grid operation and maintenance multi-task planning method and system based on space-time characteristics. The method comprises the following steps: based on equipment time sequence dynamic characteristics and equipment space correlation characteristics, performing power grid operation monitoring tasks and equipment state monitoring tasks to obtain fault detection statistics and fault positioning statistics; performing a risk research and judgment process on the fault detection statistics and the fault positioning statistics and time sequence qualitative trend characteristics to obtain an operation risk assessment result of a target power grid; inputting the equipment space correlation characteristics into each graph neural network for learning to obtain global topology aggregation characteristics of the target power grid and node space characteristics corresponding to each equipment; and inputting the global topology aggregation characteristics and each node space characteristic into a corresponding strategy network for processing to output an operation and maintenance coordination planning strategy of the target power grid. The application not only improves the overall execution efficiency of power grid operation and maintenance work, but also reduces the comprehensive cost of operation and maintenance scheduling.

The application provides a kind of underground pipeline pipe gallery resource interconnection method and device based on large language model, belongs to computer technology field, specifically includes obtaining the pipeline detection data of all underground pipelines in target area, constructs global topology knowledge graph;Extract local topology knowledge graph from global topology knowledge graph;The structured text is obtained by converting local topology knowledge graph, and the pipeline gallery resources in structured text are evaluated and filtered using a large language model, to obtain structured text with compatibility evaluation label;Using a large language model, execute logical reasoning based on the local topological relationship represented by the structured text with compatibility evaluation label, generate reasoning results that meet pipeline planning and complete path chain from start to end;Based on the reasoning result, a candidate interconnection scheme is generated. Through the processing scheme of the present application, an interconnection scheme is generated that takes into account the reuse of existing resources, cost control and safety compliance, improving the intensive utilization efficiency of urban underground space.

The present application relates to the field of satellite computing power network task scheduling, and provides a satellite computing power network-oriented dependent task scheduling method, device and readable storage medium, the satellite computing power network includes a plurality of interconnected low earth orbit satellite nodes, the task scheduling method includes the following steps: S100: receive the original adjacency matrix of DAG task and the initial feature vector of subtask, perform topological symmetry, extended neighborhood definition and structure position coding, and construct a bidirectional information flow embedding space;S200: based on the symmetric adjacency matrix and the initial feature of the subtask, the DAG task is processed through the graph neural network containing multi-head attention and GRU gate, and the advanced embedding feature containing global topological information is extracted;S300: based on the extracted advanced embedding feature, a state space, an action space and a reward mechanism are constructed to drive the agent to generate a scheduling strategy. Therefore, the joint scheduling for the global topology can be realized, and the satellite environment of dynamic and resource-limited can be better adapted.

This invention discloses a comprehensive collaborative energy-saving optimization method and system for industrial pumps, encompassing the equipment, system, and operating conditions. This method abandons traditional single-point or offline optimization, constructing a real-time dynamic optimization system with a closed-loop "perception-analysis-decision-execution" mechanism. The core of this method lies in: deploying intelligent edge nodes at the equipment level to achieve microsecond-level adaptive adjustment based on real-time operating conditions; building a digital twin model of the pump group at the system level to perform global topology and scheduling optimization; and analyzing process requirements at the operating condition level to generate a dynamic "optimal operating condition curve." Using a unique "system energy efficiency entropy" model as a unified optimization target, and leveraging a cloud-based collaborative optimization engine, it coordinates decisions across the three levels in real time, achieving optimal energy efficiency across the entire chain, from single pump efficiency and pipeline losses to process matching. This invention achieves real-time, adaptive, and globally optimal energy-saving effects.

This invention relates to the field of intelligent processing of chemical process diagrams and industrial image recognition technology, specifically to an intelligent parsing and change management method for chemical process diagrams (P&ID) based on image processing. The method includes: acquiring a two-dimensional degraded document image containing unstructured modification traces; extracting underlying image connectivity features and character features; inputting a graph neural network to reconstruct an initial logical topology network; dividing local regions and detecting logical conflicts using deterministic topology rules, quantifying the probability value of topological logical conflicts; comparing with a preset danger threshold, actively severing logical connections in high-risk areas and marking them as isolated sub-graph regions awaiting manual review, while retaining connectivity paths in low-risk areas to generate a global topology network; updating the global topology network based on manual review feedback, and outputting the final document drawing change parsing version. This invention significantly improves the recognition stability of complex on-site drawings.

This invention discloses a method, system, terminal device, and storage medium for task allocation in directed acyclic graphs (DAGs), relating to the field of graph representation learning technology. The invention includes: performing topological analysis on the input DAG graph, constructing directed path distance relationships between nodes and generating topological distance codes; introducing critical path structural biases, embedding critical path lengths and node structural importance into the attention calculation process to enhance the model's ability to perceive important structural information; explicitly introducing directional constraints into the DAG within the self-attention mechanism by constructing a directional consistency attention mask, thereby achieving unified modeling of complex topological structures and long-range dependencies; constructing a graph-level representation learning model, and then optimizing task scheduling based on the current server operating state. This invention can more fully capture hierarchical relationships, critical path information, and global topological features in DAG graphs, significantly improving the structural representation capability of complex workflows and reducing the overall system scheduling cost.

This invention discloses a method and system for lane line vectorization pre-annotation based on multi-frame LiDAR point clouds. It constructs a global dense point cloud map by performing spatial coarse filtering and spatiotemporal synchronous stitching on the original point cloud data; extracts the main road surface point cloud through elevation fine filtering and plane segmentation; suppresses and enhances non-uniform backgrounds based on morphological top-hat transform; extracts candidate lane line point clouds using adaptive threshold segmentation combining statistical lower bound and recursive maximum inter-class variance method; extracts structured candidate lane lines and geometric features based on weighted iterative RANSAC line fitting guided by both intensity and longitudinal density; and constructs a global topology graph model and a multi-dimensional scoring mechanism to screen and resolve conflicts among candidate lane lines, outputting the final lane line vector map. This invention effectively overcomes uneven intensity and local noise interference, achieving high-purity and high-robustness lane line vectorization reconstruction, and significantly reducing the cost of manual annotation.

This invention relates to the field of distributed energy regulation in power distribution networks, specifically disclosing a photovoltaic cluster collaborative regulation system and method based on digital twins and topology association. The system includes a cloud-based decision-making layer, an edge control layer, and a terminal execution layer. The cloud deploys a global topology database and a digital twin engine to perform day-ahead optimization scheduling and generate regional power targets. The edge layer uses an improved NSGA-III algorithm incorporating topology constraint factors for multi-objective optimization, generating localized regulation commands. The terminal layer possesses four-quadrant enhanced control and communication interruption autonomy functions. The system constructs a three-level topology model of "distribution transformer-branch line-equipment," quantifies electrical distance and topology association, and establishes a dynamic adjacency matrix. This invention achieves topology-driven precise power allocation and dynamic clustering, effectively improving photovoltaic absorption capacity, regulation response speed, and power distribution network operation safety through digital twin pre-simulation and cloud-edge-terminal closed-loop collaboration.

This invention relates to the field of computer-aided engineering technology and discloses a method for generating unstructured meshes based on graph convolutional networks (GCNs) and high-order size fields. It aims to solve the problems of striping artifacts easily generated when generating size fields from STL background meshes, and the excessive computational burden caused by global encryption. The method first generates an initial size field based on an STL mesh; then, it constructs a GCN model to extract features and classifies edges to identify striped edges; next, it employs a "detection-selective order-upgrading" strategy, inserting high-order nodes only on striped mesh edges to construct a high-order background mesh; finally, it uses a hybrid interpolation strategy, employing linear interpolation for non-striped mesh edges and cubic spline interpolation for high-order edges to calculate size values. This invention efficiently eliminates artifacts without changing the global topology, enabling the generation of high-quality unstructured meshes.

The application provides a deep earth target monitoring method and device for spatiotemporal continuous modeling. The method embeds sensor discrete sampling data into a four-dimensional spatiotemporal coordinate domain and establishes a global topological graph structure through a dynamic graph attention mechanism; an implicit neural spatiotemporal continuous field based on coordinate mapping is constructed, deep earth signal features are extracted by using a multi-scale position coding and a periodic activation architecture; a geomechanics partial differential equation is introduced in parameter optimization to construct a physical constraint operator, realizing physical consistency deduction of non-observed coordinate points; a Fourier neural operator is used to perform operator mapping in the frequency domain, reconstructing a cross-scale global continuous feature field; and a companion gradient inversion algorithm is used to realize reverse source location and microcrack trajectory tracking. The application realizes conversion from discrete observation to high-fidelity continuous physical field, significantly improving monitoring accuracy and early warning reliability in extreme working conditions such as blind area monitoring, noise interference and local sensor failure.

This invention discloses a method, apparatus, and device for intelligent identification and classification of microplastics based on multi-dimensional collaboration. The method includes: acquiring optical microscopic images and auxiliary modal data of the water sample to be identified; performing multi-modal encoding and fusion processing to obtain a fused feature vector; obtaining global contextual features through global attention sequence encoding; obtaining a microplastic segmentation probability map through network decoding; generating a morphological topology map based on the segmentation probability map; extracting graph structure information to obtain global topological features; calculating geometric statistical features; and fusing global contextual features, global topological features, and geometric statistical features for prototype matching to obtain a morphological classification result. Because this invention utilizes optical and auxiliary modal information uniformly through multi-modal fusion, explicitly models the skeleton and contour relationships based on the topology map to overcome local texture misjudgment defects, and combines multi-feature prototype matching, it effectively improves the accuracy of microplastic identification and classification under complex morphological conditions and uneven sample distribution.

The application discloses a skeleton action recognition method based on dynamic graph convolution fusing topological structure priori, and comprises the following steps: generating a topological barcode and converting the topological barcode into a continuous global topological feature vector through a differentiable mapping mode; constructing a node flow dynamic graph and an edge flow dynamic graph; extracting features of four modes of joints, bones, joint motion and bone motion, calculating topological confidence by using the global topological feature vector, and independently generating affine transformation parameters for each mode according to the confidence to obtain multi-modal joint features fusing topological priori; in the shallow stage of the network, the global topological feature vector is injected with high intensity to constrain the bone geometry structure, and in the deep stage of the network, the injection intensity of the topological information is gradually reduced through an inverse S-shaped continuous scheduling function, so that the network focuses on learning high-dimensional semantic discriminative features; and outputting an action recognition result; and the application can accurately distinguish complex or easily confused human behaviors.

This disclosure provides embodiments of a device link visualization rendering method, apparatus, and computer-readable medium. One specific implementation of the method includes: acquiring a set of device link information; preprocessing the set of device link information to obtain a set of target device link information groups; for each target device link information group in the target device link information group set, generating a local topology based on the target device link information group to generate a local device link graph; performing global topology optimization based on the obtained set of local device link graphs to obtain a global device link graph; performing layered rendering on the global device link graph to obtain a set of rendering layers; and overlaying blended layers on the set of rendering layers to obtain a rendered device link graph. This implementation achieves efficient link graph rendering in large-scale, complex device link visualization rendering scenarios.

The application discloses a power distribution network secondary business real-time simulation method based on a coroutine and a discrete event queue, which comprises the following steps: S1, constructing an entity set; mounting each component corresponding to each entity; constructing a global topology; constructing a business system; S2, establishing a discrete event queue; any observable event in the discrete event queue is in the cycle control on the event scheduler; S3, instantiating each business logic into each corresponding coroutine task; each coroutine task is handed over to the event scheduler for control, and is used for setting each coroutine task into a suspended-waken-up-short processing-re-suspended execution mode, and starting simulation. Through the overall modeling of entities-components-topology-system, the application carries out specific data interaction between secondary business simulation and actual physical simulation, optimizes the object and running logic of the secondary business simulation, guarantees the fixed step execution of the actual physical simulation, and improves stability, reduces delay and resource consumption.

The application provides a three-dimensional road model generation method and system based on a maximum smoothing strategy, comprising: acquiring point data in GIS data; filtering the point data, and analyzing a current optimal solution from a local topological structure to a global topological structure; according to the optimal solution, acquiring a structure with the smallest radius to perform smoothing processing on a current road to complete curve fitting; and combining a spatial structure of the road and a corresponding strategy to further generate a three-dimensional road. The application automatically generates a 3D road semantic structure in a large scale through sparse road network GIS data, which conforms to road national standard specifications and can be used for traffic simulation, improves the degree of automatic generation, and generates a three-dimensional road network road which is continuous in height through cubic spline interpolation of a fixed boundary, thereby improving the visual effect of the road, enhancing the reality of the road, and conforming to highway design standards.

The application relates to the technical field of intelligent logistics equipment scheduling, and discloses an unmanned towing trailer cluster intelligent scheduling management platform based on a cloud edge end collaborative three-layer architecture design, which comprises a cloud platform layer, an edge computing layer and a vehicle-mounted terminal layer, the cloud platform layer is configured with core modules such as task management, global intelligent scheduling and path planning, adopts a scheduling strategy combining a genetic algorithm and reinforcement learning, and realizes accurate matching of transport capacity and tasks by fusing multiple constraint conditions; the path planning module is based on vehicle dynamics constraints, adopts a hierarchical architecture of global topology planning, local behavior decision and smooth trajectory generation, adapts to the physical characteristics of towing trailers and solves multi-trailer conflicts, and is matched with multi-link redundant communication of 5G, Wi-Fi, Beidou short message to guarantee stable data transmission, is suitable for multiple scenes such as factory areas, ports and warehouses, is compatible with unmanned towing trailers of different brands, and realizes intelligent, unmanned and efficient management of cluster operation.

Methods and devices for encoding and decoding 3D point clouds. A learned deep entropy model over octrees is proposed for lossless compression of 3D point cloud data. The self-supervised compression consists of an adaptive entropy coder which operates on a tree-structured conditional entropy model. The information from the local neighborhood as well as the global topology is utilized from the octree structure. In an embodiment, the features from the parent level is up-sampled to bring them to the resolution of the current level before further feature aggregation. For processing dense massive point clouds and to facilitate parallel processing, a block-based compression scheme is proposed to reduce the required computation and time resources.

The application discloses an axial slice-based underground mine pipeline point cloud completion method, relates to the technical field of point cloud data processing, and is used for data completion of pipeline point cloud, and comprises the following steps: slicing a pipeline point cloud data after preprocessing along an axial direction, reducing three-dimensional point cloud to two-dimensional plane processing, calculating related geometric parameters of the pipeline, extracting a pipeline center line based on improved DBSCAN clustering and least square circle fitting, constructing a skeleton model under geometric topological constraints, combining local geometric features and global topological constraints, realizing geometric reconstruction of a missing area through a rotary surface generation and local interpolation algorithm, and realizing robust repair of the missing area. The application extracts the pipeline center line and geometric parameters in combination with a convex hull and a RANSAC method, realizes completion of the pipeline point cloud data through a continuous rotation correction method, and significantly improves the completion quality of the pipeline point cloud.

This application relates to a hierarchical path planning method, apparatus, computer equipment, readable storage medium, and program product for a traction substation robot, belonging to the field of robot path planning technology. This application can improve planning efficiency and address poor environmental adaptability. The method includes: constructing a two-dimensional grid map based on the physical environment information of the traction substation; determining the robot's starting point and a set of task points to be inspected on the two-dimensional grid map; encoding the task point set using a matrix encoding network based on a MatNet architecture to obtain a node pair feature matrix that integrates global topological relationships; generating a global target inspection path based on the node pair feature matrix using a trained generative diffusion model; generating local navigation paths connecting each task point in the global target inspection path; and constructing a third-order Bézier curve to smoothly fit the key inflection points of the local navigation path to generate a final execution path that satisfies the robot's minimum turning radius and kinematic constraints.

This invention discloses a human-machine collaborative warehousing multi-agent path planning method based on intent estimation, comprising: constructing a global topology graph of the warehousing environment; obtaining the current state information of operators and multiple agents in the warehousing environment to generate a current observation vector; continuously probabilistically inferring the operator's work target intent based on the operator's historical movement trajectory and current observation vector; inferring the nodes or edges of the global topology graph of the warehousing environment that the operator may occupy within a preset time range based on the intent estimation results; determining the set of agents affected by conflicts during the multi-agent path finding process; performing path replanning for agents affected by conflicts in a local subgraph, iteratively generating conflict-free local path schemes and merging them into the global path. This invention achieves efficient conflict handling and real-time replanning in dynamic environments, taking into account the safety, near-optimal nature, and computational efficiency of path planning, and is suitable for large-scale human-machine collaborative warehousing scenarios.

A method, system, and medium for predicting latency in low-Earth orbit satellite networks are disclosed. The method includes: acquiring service flow attributes and routing information to construct a topology representation; establishing a flow-link mapping, aggregating flow attributes passing through the same link, and implicitly estimating link states; modeling the global topology and local paths based on link states, wherein the global topology utilizes a graph neural network to extract global context, and local paths aggregate link states along the route to obtain path-level features; and fusing path-level features and global context to predict end-to-end latency. This invention completes latency modeling using only flow-level attributes and routing information without relying on link-level telemetry data, reducing control plane acquisition and transmission overhead, ensuring prediction accuracy through dual-scale modeling, improving system scalability, and saving on-board resources.

This invention discloses an AI computing resource management system and method supporting fine-grained cross-domain scheduling, relating to the field of artificial intelligence technology. It addresses the background issues of propagation latency and jitter in the process of resource status collection, aggregation, and propagation to usage by the scheduler. This leads to inconsistencies across fields in the global view and time asynchrony across nodes, resulting in a sharp increase in communication and performance crashes. It also lacks engineering guarantees for atomic start / stop and parameter consistency of cross-node container groups. The proposed solution includes constructing a global topology resource view: a multi-regional coordinator cluster based on a distributed consensus protocol synchronizes and maintains a global resource view in real time. This invention reduces erroneous scheduling caused by cross-domain status lag, avoids misjudgments caused by cross-field inconsistencies, improves the robustness of cross-domain scheduling, reduces the risk of a sharp increase in communication, and avoids catastrophic communication degradation in cross-domain multi-GPU tasks.

The application discloses a welding robot pre-welding global planning method and system, relates to the technical field of intelligent control of welding robots, and comprises the following steps: generating an initial three-dimensional space structure of a steel member to be welded based on a neural implicit representation technology; constructing a hierarchical topological graph of a plate weld seam corresponding to the target member; calculating a global topological completeness index based on the cognitive confidence of each weld seam to be welded of the target member; if the global topological completeness index does not reach a preset confidence threshold, then planning active supplementary measurement under an information gain mechanism in combination with welding process requirements; if the global topological completeness index reaches the preset confidence threshold, then determining cognitive completeness, and outputting a global welding planning task. Through the cycle of evaluation, decision-making and execution, the method iteratively approaches the cognitive completeness state, and finally generates the global welding planning task, thereby solving the technical problem that the existing passive mode working mode of the welding robot is difficult to meet the demand of flexible automatic production.