24 results about "Type graph" patented technology

The invention discloses an injection molding equipment anomaly detection method based on a graph neural network, and the method comprises the following steps: collecting multi-dimensional monitoring data in the operation process of injection molding equipment, and carrying out the normalization, denoising and missing value filling, and obtaining structured monitoring sequence data; dividing the injection molding production process into a plurality of process stages based on timestamps and process stage labels in the structured monitoring sequence data, and constructing a corresponding stage sub-graph set; based on the stage sub-graph set, the energy transfer path, the material flow path and the historical abnormal propagation path, constructing a nested graph structure; inputting the nested graph structure into an improved dynamic graph convolutional network for feature extraction to obtain a time-aware graph embedding representation; and performing anomaly detection on the time perception type graph embedded representation to generate an anomaly detection result of the injection molding equipment. According to the method, nested graph structure modeling and the dynamic graph convolutional network are adopted, and high-precision detection and rapid positioning of the abnormal state of the injection molding equipment are achieved.

The invention discloses a chart visual question-answering method based on dynamic routing and low-rank mixing and a related device, and relates to the technical field of visual question-answering, and the method comprises the steps: obtaining original data containing to-be-processed charts and text questions, and generating corresponding answers through the processing of a pre-trained dual-channel fusion model. In the model, a visual encoder is provided to extract semantic features of complex layout of the chart, and a text encoder accurately converts text problem vectors, so that the feature matching degree of the chart and the problem is improved; the chart mixed connection structure dynamically fuses visual features of the data graph and the data table, and the cross-type chart adaptability is enhanced; according to the chart low-rank mixed structure, related vectors are weighted and fused through a LoRA routing network, low-rank adaptation and dynamic fusion are achieved in combination with output of a large language model layer, training parameters are reduced, and the zero sample question and answer ability is improved; the two-stage training strategy reduces the loss of the video memory. According to the method, the universality, accuracy and zero sample performance of chart visual questions and answers are effectively improved, and the training cost is reduced.

The invention relates to a global graph data unified graph modeling method based on hard hint, which is characterized in that attribute characteristics of nodes, edges and labels are uniformly expressed by using natural languages through a standardized text description template. Besides, semantic coding is performed on the generated text features through the large language model, and a fine tuning strategy based on comparative learning is adopted, so that the representation consistency of the large language model on graph data is enhanced. And finally, introducing a Prompt node and a category node, realizing the unification of the multitask form of graph learning, and guiding model training by using source task key information carried by the Prompt node. According to the method, by constructing the text attribute graph, the graph data in different fields can be modeled under a unified text representation framework, so that the problem that different graph data need to be trained independently from the beginning is solved, and a multi-field and multi-type graph representation learning task is modeled by a single model; unified graph modeling of multi-field graph data can be achieved through a single model, and an important practice normal form is provided for generalization of a graph basic model.

The invention discloses an abnormal monitoring data identification and restoration method based on an improved box type method, and the method comprises the steps: S1, collecting a monitoring data segment with a fixed length, building a data label through a box type graph in combination with the analysis features of a scatter diagram, and obtaining the features of the box type graph; step S2, calculating four statistical parameters of each segment of data according to the box diagram characteristics and the maximum value, the minimum value, the median, the lower quartile and the upper quartile of the display data; s3, calculating a relative difference rate by taking a statistical index of fault-free data in the same structure monitoring time sequence as a standard; s4, identifying fault data segments according to the relative difference rate of the data segments, and classifying the fault data segments; s5, adopting different repair strategies according to different fault data segments; and S6, finally, verifying the repair effect of repairing the fault data segment by adopting the repair strategy. According to the invention, accurate identification and classification of various types of abnormal data are realized.

The invention discloses a wind power plant wake flow rapid optimization method based on a graph neural network, and belongs to the technical field of wind power plant control. According to the method, a wind power plant physical information enhancement graph structure is constructed, fans are abstracted as nodes, and geographic coordinates, wind direction adjustment coordinates and incoming flow wind speed of the fans are subjected to feature fusion; abstracting a wake flow influence relationship into edges, and fusing features with relative spacing and flow direction; based on the graph structure, respectively training a description type graph neural network model and a decision type graph neural network model by utilizing power data generated by a traditional wake flow model and optimal yaw angle data generated by an optimization algorithm; and finally, constructing the real-time wind field state into a graph, inputting the graph into the trained decision-making model, and directly outputting a full-field optimal yaw angle instruction set at a millisecond level. Through feature fusion graph construction and a decision-making type GNN training normal form, the problem that consumed time of traditional yaw optimization calculation is increased along with scale indexes is solved, and a feasible scheme is provided for real-time collaborative optimization of the wind power plant.

The invention discloses a power-economic mapping matching method and system based on an improved hierarchical neural network model, and the method comprises the steps: inputting power data and economic data into a feature coding layer for feature extraction, and obtaining context semantic features of the power-economic data; inputting the context semantic features into an event relation modeling layer, constructing a multi-type graph structure by taking time nodes as event nodes in the graph structure, and modeling the multi-type graph structure to obtain a power-economic state sequence; performing feature matching on the electric power-economic state sequence through a preset electric power-economic feature matching mechanism of a feature matching layer to obtain a semantic similarity between the current electric power economic state and a target economic structure; and the output prediction layer constructs a task output structure according to different task requirements on the basis of semantic similarity, and obtains a final conclusion of power-economy mapping in combination with structural features and a matching result. According to the invention, the expression ability and decision-making assistance ability of the model to a complex coupling system are significantly improved.

The application discloses a kind of big model remote sensing small sample change detection methods and systems combined with graph attention, it is related to artificial intelligence technical field, including: first, the original remote sensing image is carried out radiation correction and atmospheric correction, generates standardization remote sensing image;By geographic registration, the standardization image of different time is aligned to unified geographic coordinate system;Reference phase and change phase image after alignment are obtained;Two phase images are input into the target remote sensing small sample change detection model of pre-training and contain graph attention feature encoder, and the change type graph of target area is output.The application captures the topological relationship of geospatial by graph attention mechanism, solves the problem that traditional method relies on a large number of labeled data, and improves the change detection accuracy in small sample scene.

The embodiment of the invention provides a house type image processing method and system, electronic equipment and a program product, and the method comprises the steps: firstly carrying out the instance segmentation of a target two-dimensional house type image, and obtaining a mask pixel region of each target instance; then, the minimum enclosing rectangle is adopted to replace a traditional axis alignment enclosing frame, the fitting precision of the irregular or inclined wall contour is improved, and the robustness of geometric expression is enhanced; the center lines are further extracted to form a line segment set, and an undirected graph structure is constructed according to the line segment set, so that the connection relation between the walls and the overall topological logic are reflected more accurately; and finally, geometric information such as the center point and the length of the door and window and the wall thickness is fused in the graph structure, a unified target undirected graph structure is generated and serves as structured vector expression of the house type graph, topological reconstruction of special structures such as an inclined wall and an arc-shaped glass curtain wall can be supported, and the accuracy and the applicability of house type graph recognition and three-dimensional reconstruction are remarkably improved.

The invention discloses an agent-based document image-text generation method and system, and the method comprises the steps: calling a large model service according to the input of a user to generate the preliminary content of a document, and adding a special mark and description information into a part related to a chart; positioning description information of the chart through a special mark, and analyzing the description information to obtain a chart type and chart detailed information; performing flow arrangement according to the analysis content, and calling a corresponding type chart generation tool according to the chart type to generate chart content; replacing the generated chart content to the special mark and the chart description position in the document to obtain a replaced document; verifying the consistency and accuracy of the whole replaced document, and repeatedly analyzing and generating the chart when the consistency verification fails until the verification is passed; and finally, outputting the verified final document. According to the method, user requirements can be efficiently met, and the quality and speciality of document generation are ensured.

The invention provides a three-dimensional indoor scene layout generation method and system based on scene graph control, and relates to the technical field of three-dimensional scene generation, and the method comprises the steps: obtaining the natural demand description of a user for a scene layout; inputting the natural demand description into an LLM for analysis, and generating a structured sparse scene graph according to a spatial logic relationship; a multi-conditional constraint scene generation method based on a diffusion model is utilized, a sparse scene graph, a house type graph and spatial constraints are jointly used as conditional information, the generation process of the diffusion model is guided in a mixed injection mode, and a lightweight adaptive weight learning mechanism is introduced to enhance physical constraints on layout. Weight distribution of different guiding conditions in the generation process is balanced in a self-adaptive mode, and finally layout data are output and obtained; and according to the layout data, retrieving and matching a corresponding three-dimensional model from a set 3D-FUTURE model library, and placing the three-dimensional model into a three-dimensional space to construct a complete three-dimensional indoor scene. According to the invention, generalization and accuracy of controllable scene generation are improved.

The application discloses a chart visual question answering method based on dynamic routing and low rank mixing and a related device, relates to the technical field of visual question answering, and comprises the following steps: acquiring original data containing a chart to be processed and a text question, and generating a corresponding answer by processing through a pre-trained double-channel fusion model. In the model, a visual encoder is provided to extract semantic features of a complex layout of the chart, a text encoder is used to accurately convert a text question vector, and the two improve the feature matching degree of the chart and the question; a chart mixed connection structure dynamically fuses visual features of a data graph and a data table, and enhances the cross-type chart adaptability; a chart low rank mixing structure fuses related vectors by means of a LoRA routing network, realizes low rank adaptation and dynamic fusion in combination with a large language model layer output, reduces training parameters, and improves zero sample question answering capability; and a two-stage training strategy reduces the memory loss. The method effectively improves the generality, accuracy and zero sample performance of chart visual question answering, and reduces the training cost.

The application discloses a graph data backdoor defense method and device based on spectral domain transformation and edge weight learning, equipment, medium and program product, relates to the technical field of network security, and includes: obtaining and constructing a feature vector matrix and an eigenvalue matrix of an original graph data set; mapping each feature vector matrix from a spatial domain to a spectral space through spectral transformation to obtain a plurality of spectral domain node features, so as to construct a normal data distribution of the original graph data set, and determine whether each to-be-tested graph data is abnormal graph data; if it is determined that the graph data is abnormal, learning abnormal node features and abnormal adjacency relationships of the to-be-tested graph data through a multilayer perceptron to obtain abnormal edge weights; based on the abnormal edge weights, performing adaptive clipping on abnormal spectral components in the abnormal graph data through a Gaussian mixture model to obtain modified spectral domain node features; and performing inverse spectral transformation on each modified spectral domain node feature to reconstruct a target graph data set. The application can effectively resist multiple types of graph backdoor attacks.

The invention discloses a class model extraction method and device, computer equipment and a storage medium. The method comprises the following steps: receiving a parent class extraction instruction of a user for extracting a first type of primitives, displaying a parent class extraction window, selecting a first subclass list in the parent class window, selecting parent class members to be extracted from first member information, and creating a new parent class model object based on parent class basic information and the parent class members; receiving an interface class extraction instruction of the user for extracting the second type of primitives, displaying an interface class extraction window, selecting a second subclass list in the interface class window, selecting to-be-extracted interface class members from the second member information, and newly establishing an interface class model object based on the interface class information and the interface class members; and taking the parent class model object and the interface class model object as target extraction results, and according to the technical scheme, on the basis of more convenient operation, the extraction efficiency can be improved, and efficient, intelligent, cooperative and friendly class model reconstruction can be realized.

The application relates to the technical field of computer vision, and discloses an image and chart recognition system and method based on computer vision, which comprises the following steps: acquiring a to-be-processed file image set of a to-be-processed file, performing resolution standardization processing to obtain a standard image sequence; determining a visual saliency map according to the standard image sequence, and generating a chart detection frame set; constructing a chart region association graph based on the spatial topological relation and visual feature similarity of the chart detection frame set, and determining a chart type identifier; extracting a coordinate axis structure element, a legend semantic element and a data visualization element, and generating a structured data record; and fusing and encoding the structured data record, the chart type identifier and the corresponding legend semantic element of each chart region to generate a chart knowledge vector, and establishing a chart mapping relation of the to-be-processed file, so that accurate classification and structured data extraction of multiple types of charts are realized, and cross-file semantic association and batch processing performance are optimized.

1.1. shows a display screen with a graphical user interface having a bar type graph; 2.1 shows a portion of a display screen with a graphical user interface having multiple bar type graphs; 3.1 shows a display screen with a graphical user interface having a gauge type graph; 1.2, 2.2, and 3.2 are reference views showing examples of use and do not form part of the claimed designs; any broken lines shown in the drawings are directed to environment and are for illustrative purposes only; the broken lines form no part of the claimed design; and the number signs (“#”) shown in the drawings represent placeholders and indicate that numbers are displayed where these number signs are shown.

The invention discloses a graph data backdoor defense method and device based on spectral domain transformation and edge weight learning, equipment, a medium and a program product, and relates to the technical field of network security, and the method comprises the steps: obtaining and constructing a feature vector matrix and a feature value matrix of an original graph data set; mapping each feature vector matrix from a spatial domain to a spectral space through spectral transformation to obtain a plurality of spectral domain node features so as to construct normal data distribution of the original graph data set, and judging whether each piece of to-be-detected graph data is abnormal graph data or not; if the to-be-detected graph data is judged to be abnormal graph data, abnormal node features and abnormal adjacency relations of the to-be-detected graph data are learned through a multi-layer perceptron, and abnormal edge weights are obtained; on the basis of each abnormal edge weight, performing adaptive cutting on abnormal spectrum components in the abnormal graph data through a Gaussian mixture model to obtain corrected spectral domain node features; and performing inverse spectral transformation on each corrected spectral domain node feature, and reconstructing to obtain a target graph data set. According to the method, multi-type graph backdoor attacks can be effectively resisted.

The invention discloses an information disclosure monitoring analysis method based on artificial intelligence, particularly relates to the technical field of data processing and analysis, and comprises six steps of information data preprocessing, information structured analysis, information quality intelligent evaluation, information disclosure quantitative analysis, abnormal information intelligent identification and monitoring result visual generation. According to the method, deep semantic analysis is conducted on information content through a natural language processing technology, word segmentation, named entity recognition and dependency syntactic analysis are conducted through a Bi-LSTM-CRF model and a BERT model, and a knowledge graph is constructed to achieve structured representation of information; according to the method, the functions of multi-dimensional data aggregation, multi-type chart generation, interactive drilling analysis and multi-format output are provided, the problems that in the prior art, the evaluation dimension is single, the anomaly recognition capacity is weak, and the display form is single are solved, comprehensive analysis from surface features to deep semantics is achieved, and the monitoring accuracy and practical value are improved.

The application discloses a kind of multimodal end-to-end house type design method, system, computer equipment and storage medium, comprising: prompting word and the blank house pixel graph of to-be-generated house type graph are input into Instruct-Pix2Pix multimodal big model, and new house pixel graph containing multiple regions satisfying prompting word description is generated;The new house pixel graph obtained is divided into regions, and each individual region graph is obtained;For each individual region graph, the sample set of corresponding region type is used, and the corresponding home arrangement is generated for it using WFC algorithm, to obtain the region graph with furniture arranged;All furniture-arranged region graphs are fused to generate a house type graph;The house type graph is an image with an outer contour conforming to the blank house pixel graph, region division satisfying the prompting word description and furniture arranged in each region.

The invention provides an irrigation area water distribution method based on topology and hierarchical perception graph similarity calculation, and relates to the field of intelligent water distribution, the method comprises the following steps: respectively modeling a historical water distribution scheme and a current water demand sub-graph into directed unweighted graphs; multi-dimensional feature coding is carried out on each node, and node enhancement representation is obtained after fusion; performing multi-scale feature extraction on the node enhanced representation by using a relational graph convolutional network, and introducing a gating mechanism to filter invalid node information to obtain final node embedding; on the basis of final node embedding, hierarchical constraint weighting attention aggregation double-path features are adopted, and graph-level representation is generated; calculating a similarity score between the two graphs, and performing weighted fusion to obtain a prediction graph editing distance; and recommending an optimal water distribution scheme from a historical scheme library according to the prediction map editing distance. According to the technical scheme, the matching task of the historical water distribution scheme graph and the current demand graph is converted into a scene adaptive graph similarity modeling problem, and a core basis is provided for intelligent recommendation of the irrigation area water distribution scheme.

The invention discloses an automatic generation method for an image knowledge graph based on a graph convolutional network. The method comprises the following steps: carrying out key feature extraction on an input image by using a transfer learning method containing a cross attention mechanism; and inputting the extracted key features as node features into a graph convolutional network, modeling a relationship and a structure between the node features through a message passing and updating mechanism, and generating an image knowledge graph. The method has the advantages that the transfer learning method is adopted, and the adaptive capacity to different types of images is improved by utilizing the universal features learned on multiple data sets by the pre-training model.

The application relates to the technical field of stacking control, and provides a stacking type graph generation method and device, a stacking robot and a robot stacking method. The method comprises the following steps: obtaining an original stacking type graph based on current frame depth information after each stacking operation is completed; performing difference value processing on the original stacking type graph of a previous frame and the original stacking type graph of a current frame to obtain a difference value graph; performing contour detection on the difference value graph to obtain position information of a newly added stacking object in the difference value graph; updating height values in an output stacking type graph of the previous frame according to the position information, and outputting the updated stacking type graph as an output stacking type graph of the current frame. The method can obtain a more detailed stacking type graph by using the information of the previous frame and the current frame, is beneficial to stacking of small objects, and can further improve the volume rate of stacking.

The invention discloses a chart question-answer pair generation method and device, equipment, a medium and a product, and relates to the field of natural language processing, and the method comprises the steps: obtaining a domain knowledge document; the domain knowledge document is a PDF document containing a statistical chart; adopting an optical character recognition method and a visual language model to recognize a statistical chart in the domain knowledge document, and determining a chart position, a chart type and a chart description; constructing a global context and a local context according to the domain knowledge document and the chart position; and according to the chart type, the chart description, the global context, the local context and a pre-constructed chart knowledge base, generating question and answer pairs by adopting a large language model. According to the method, the statistical chart in the domain knowledge document can be accurately identified, the accuracy of chart identification is improved, meanwhile, contexts with different granularities are generated according to the global and local contents of the chart to serve as the basis of question and answer pair generation, and the generation quality of question and answer pairs is improved.

The invention discloses a large model remote sensing small sample change detection method and system combined with graph attention, and relates to the technical field of artificial intelligence, and the method comprises the steps: firstly carrying out the radiation correction and atmospheric correction of an original remote sensing image, and generating a standardized remote sensing image; the standardized images of different time phases are aligned to a unified geographic coordinate system through geographic registration; obtaining the aligned reference time phase and change time phase images; and inputting the two time phase images into a pre-trained target remote sensing small sample change detection model containing a graph attention feature encoder, and outputting a change type graph of the target area. According to the method, the ground feature space-time topological relation is captured through a graph attention mechanism, the problem that a traditional method depends on a large amount of labeled data is solved, and the change detection accuracy in a small sample scene is improved.

The large language model consumes an example query expression, including a data access function, a data analysis function, or a data enhancement function. The large language model receives a centrally managed ontology. The large language model uses centrally managed ontologies and identifies skill ontology types from example query expressions. The skill ontology type is a standardized (in accordance with a centrally managed ontology) input parameter type or structured output. The large language model receives a context for survey and identifies a context ontology type. The large language model receives a received skill based on a correlation between a skill ontology type and a contextual ontology type, the skill ontology type having a connection with the received skill in the graph. The large language model generates and provides an indication of suggested skills for the survey.