22 results about "Global structure" patented technology

The invention relates to the technical field of image analysis, in particular to a hydraulic engineering construction quality monitoring method and system based on image recognition, and the method comprises the following steps: analyzing the pixel intensity and gray gradient based on the image data of a construction region, building the adjacency characteristics of a local region, and generating an adjacency matrix through the spatial connection characteristics. According to the method, the pixel intensity and the gray gradient in the image data are analyzed, the spatial correlation characteristic weight distribution is established, the spatial characteristic analysis capability of the crack and the slope protection deformation is improved, the distribution characteristics of the crack and the deformation are optimized through path connection and edge weight extraction, the method is more accurate, the reliability of crack edge characteristic identification is enhanced, and the identification accuracy of the crack and the slope protection deformation is improved. The crack propagation direction and the slope protection deformation trend are comprehensively mastered by calculating node dynamic offset and connectivity characteristics, the prediction capacity of construction dynamic change is improved, local and global structure change characteristics are integrated, and quality monitoring from crack microscopic characteristics to the overall state of the slope protection is achieved.

The invention relates to the technical field of image recognition, in particular to a construction site-oriented reinforcing steel bar binding normativity visual detection method, which comprises the following steps of: driving a camera lens to shoot a focus stack image sequence covering the depth of field before and after a single binding node, extracting a linear contour, and obtaining a multi-dimensional image set of a binding area. According to the method, the focus stack image sequence covering the depth of field before and after the single binding node is shot by the driving camera, and the mapping of the line segment coordinates and the depth of the reinforcing steel bar is established, so that the problems of mutual shielding and depth confusion caused by staggered reinforcing steel bars in the construction site are solved; on the basis, a steel bar binding topology network diagram describing the connection relation between steel bar intersection points is constructed, discrete steel bar line segments and intersection point information are integrated into an overall network with a global structure relation, detection is not limited to isolated nodes any more, and evaluation can be conducted from systematicness and continuity of a whole steel bar framework.

The invention discloses a reinforcement learning-based global structure perception R-Tree construction method, which comprises the following steps: designing a Markov decision process constructed by an R-Tree, the Markov decision process constructed by the R-Tree comprising state design, action design and reward design; a reinforcement learning agent algorithm framework is applied, and an Actor-Critic framework and a near-end strategy optimization algorithm are selected to construct a reinforcement learning agent; training a reinforcement learning agent by adopting a self-chess-playing mechanism to carry out R-Tree incremental construction; a trained reinforcement learning agent replaces a traditional variant local optimal child selection algorithm to automatically construct an R-Tree in a construction process of inserting data one by one. According to the method, the construction performance and the query efficiency of the R-Tree are remarkably improved, meanwhile, good flexibility and expansibility are achieved, and the real-time updating requirement in a dynamic data scene can be met.

The invention discloses a three-dimensional Gaussian point cloud grid construction method and system based on multi-modal embedding, and the method comprises the steps: obtaining three-dimensional Gaussian point cloud data, and carrying out the sparse sampling and adaptive sampling processing, and obtaining a point cloud embedding vector; introducing an attention mechanism, and carrying out Poisson reconstruction on the three-dimensional Gaussian point cloud data to obtain a grid embedding vector; and carrying out Transform network training and autoregressive sampling based on the point cloud embedded vector and the grid embedded vector, and generating a three-dimensional Gaussian point cloud grid. According to the embodiment of the invention, the global structure and the local detail features of the point cloud can be comprehensively captured, the compactness when the grid is generated by autoregression sampling is improved, and the quality of the generated image or data is improved. The method can be widely applied to the technical field of three-dimensional point cloud reconstruction.

The invention provides a stereoscopic warehouse positioning system and method based on 3D visual perception, and relates to the technical field of visual identification positioning. Determining a calibration element index of each storage location center in the storage location global image, and determining spatial pose descriptions in adjacent storage location bounding boxes on cargo storage locations of the stereoscopic warehouse according to the calibration element indexes of all the storage location centers; verifying and identifying the position identification image to obtain a visual primitive feature corresponding to each cargo in the cargo placement area, and further determining a positioning identification layer sequence of the cargo on a specified storage position in the stereoscopic warehouse according to the visual primitive features corresponding to all the cargoes; and carrying out fusion sensing positioning on the goods in the stereoscopic warehouse according to the space pose description and the positioning identification sequence. According to the invention, collaborative perception can be carried out on the storage location global structure and the cargo visual features in a multi-sensor fusion environment, so that the accuracy of cargo space positioning is improved.

The invention relates to an avionics application configuration method based on message driving and a dynamic loading mechanism, which comprises the following steps of: generating a hierarchical binary configuration file containing port configuration information and mapping configuration information based on an avionics system interface control file in a construction stage, and analyzing and loading the binary configuration file to a global structure body in an operation stage, according to the method and the system, the mapping configuration information is automatically matched according to the message type of the to-be-processed message, and after parameter configuration is carried out on the corresponding equipment port according to the port configuration information, data reading and writing are carried out by utilizing the information in the equipment resident index, so that message driving and dynamic loading of application configuration are realized. According to the method, manual configuration errors can be avoided, multi-protocol adaptation is supported, only binary configuration files need to be updated when protocols are modified, the operation can be carried out offline, port mapping and related settings are automatically matched by the system according to message types, and therefore the configuration management efficiency and reliability of the avionics system are improved, and the integrity of configuration data is guaranteed.

The application belongs to the technical field of building structure health monitoring, and discloses a visual perception driven building structure damage intelligent monitoring and evaluation system and method. The method acquires a structure surface image with rolling shutter distortion and motion blur and a transient kinematics state sequence, and constructs a transient motion state space model. A two-dimensional pixel offset vector field coupled with platform displacement and exposure time difference is deduced in combination with structure depth information. An apparent relative displacement of adjacent image frames is extracted, a pure pixel displacement is obtained by using the vector field to strip non-structure deformation, and a real second-order physical strain tensor field is calculated by back-projection to a three-dimensional physical surface. Based on the tensor field and a preset material mechanics threshold, a spatial integral evolution model containing a non-negative truncation operator is constructed to solve a global structure damage intelligent evaluation index, and a monitoring and early warning result is output. The application can effectively eliminate the space-time distortion error of dynamic visual inspection.

The invention specifically relates to a deep learning coupled layered rock mass structural plane parameter inversion method, and relates to the technical field of geotechnical engineering and artificial intelligence, and the method comprises the steps: deploying an initial system coupling a deep learning proxy model and a forward numerical simulator, cooperatively laying a monitoring network, and determining an engineering digital baseline and a model performance reference through initial inversion. According to the invention, the hybrid network architecture constructs the deep learning agent model, and the efficient coupling inversion engine is formed by combining the forward numerical simulator of secondary development, so that the problems that the traditional inversion method is low in efficiency and easy to fall into local optimum are solved; the model takes similar engineering cases and indoor rock mechanics test data as a training basis, constructs a digital baseline of global structural plane parameters for engineering, and effectively overcomes the defects of indoor test scale effect, limited field investigation range and the like of a traditional method.

The present invention relates to a point cloud data repair method that integrates shape structure recovery and detail compensation. The present invention includes a point cloud repair network consisting of a global structure recovery branch and a local detail compensation branch to repair point cloud data; the global structure recovery branch adopts an encoder-decoder structure to perform a coarse repair on the input missing point cloud data to obtain a coarse repair point cloud with complete structural information; the local detail compensation branch obtains a reshaped point cloud based on learning the geometric details of the input missing point cloud; finally, the outputs of the two branches are spliced and fused and iteratively sampled at the farthest point to generate a shape fine repair result with detail information preserved and uniform sampling point distribution. In view of the limitation that it is difficult to effectively maintain the original shape detail structure information in three-dimensional point cloud shape repair, the present invention adopts a dual-branch point cloud repair network to effectively restore the shape details of the original point cloud while repairing the overall structural information of the original point cloud shape and generate a complete three-dimensional point cloud model with uniform sampling point distribution.

The invention relates to the technical field of laser cladding, in particular to a method and a system for predicting the size of a laser cladding cross section of an inclined base plane. Aiming at the problems of fracture, holes, noise and the like of the cladding layer boundary in a deep learning model prediction result, the invention provides an end point extension repair algorithm based on graph structure analysis, and boundary defects are effectively repaired through the steps of morphological operation, connected domain analysis, path reverse extension and the like. In order to improve the cladding layer section prediction precision, a Pix2Pix model is applied to cladding layer section form prediction for the first time, and perception loss is introduced to enhance the perception ability of the model to boundary details and a global structure. Besides, structural similarity indexes are introduced, molten pool images with high similarity are extracted from molten pool videos under different parameter conditions to serve as an extended training set, and the generalization ability and robustness of the model are effectively improved.

The invention relates to the technical field of computer vision, in particular to a panoramic picture generation method and device based on structured prompt, equipment and a medium. The method comprises the following steps: constructing a training data set; training the pre-training generation model based on the training data set to obtain a trained panoramic picture generation model; and obtaining a single-view-angle image and target prompt information, and generating a panoramic picture based on the single-view-angle image and the target prompt information through the panoramic picture generation model. According to the application, the structural features of the object in the streetscape are provided by introducing the local structure prompt and the global structure prompt, the global panoramic information is integrated through the panoramic branch to ensure the overall consistency between the visual angles, and the global consistency optimization is performed on the multi-visual-angle image through the panoramic alignment module. According to the invention, details and local structures are captured through combination of multi-view branches and local multi-view information, so that the problems of complex structures and inconsistent sizes of objects in panoramic picture generation of outdoor scenes are effectively solved.

The present application relates to the technical field of laser cladding, and particularly relates to a laser cladding cross-section size prediction method and system for an inclined base surface. In view of the problems of boundary fracture, holes and noise in the prediction results of a deep learning model, an endpoint extension repair algorithm based on graph structure analysis is proposed, and through morphological operation, connected domain analysis and path reverse extension and other steps, the boundary defects are effectively repaired. In order to improve the prediction accuracy of the cladding layer cross section, the Pix2Pix model is applied to the cladding layer cross section shape prediction for the first time, and the perception loss is introduced to enhance the perception ability of the model to the boundary details and global structure. In addition, the structural similarity index is introduced, and the high-similarity molten pool images are extracted from the molten pool videos under different parameter conditions as an expanded training set, so that the generalization ability and robustness of the model are effectively improved.

The application discloses a kind of concrete arch dam global structure mechanics parameter partition fast inversion method, it is related to dam operation safety monitoring and management technical field, including according to engineering design and monitoring data, three-dimensional finite element model of dam body and foundation is established using finite element software, according to dam construction material mechanics parameter information, partition is carried out to three-dimensional finite element model and the sensitivity of each regional mechanics parameter is analyzed, to determine the sensitivity mechanics parameter that influences concrete arch dam deformation, according to the result of sensitivity analysis, adaptive intelligent sampling is carried out to sensitive mechanics parameter, the deep learning agent model reflecting the nonlinear relationship between dam sensitive mechanics parameter and each monitoring point deformation is constructed, and the deep learning inversion of dam body elastic modulus and bedrock deformation modulus is carried out.The concrete arch dam global structure mechanics parameter partition fast inversion method of the application can efficiently and accurately determine the inversion actual operation period arch dam structure mechanics parameter, and provides a good basis for dam safety analysis.

The invention discloses a skeleton guide point cloud segmentation method for building structure identification. The method comprises the following steps: carrying out normal estimation on point cloud data of a target building structure to obtain a point cloud with a direction, calculating rotary symmetry axis points in the point cloud with the direction, generating a point cloud skeleton by utilizing all the rotary symmetry axis points, and storing the point cloud skeleton as an undirected graph; the point cloud skeleton is decomposed into a plurality of branches, each branch comprises a group of edges in the undirected graph, and the edges in one branch have similar directions; points in the point cloud with the directions are distributed to all the branches, corresponding subspaces are formed, and the subspaces corresponding to all the branches form a building structure recognition result. According to the embodiment of the invention, the ROSA rotational symmetry axis extraction method is utilized, the skeleton line reflecting the geometric morphology of the target can be stably extracted from the complex point cloud, and global structure guidance is provided for subsequent segmentation; consistent mapping of the spatial structure and the semantic structure is achieved, the structure understanding ability of segmentation is enhanced, and the overall segmentation robustness is improved.

The invention discloses a prestressed beam simulation method based on an incomplete bonding theory and a vector finite element, which comprises the following steps: calculating the mechanical response of the section of a prestressed beam under the action of different loads based on an incomplete bonding theory model so as to construct a bending moment-corner relationship; dispersing a prestressed beam structure where the section of the prestressed beam is located into a plurality of mass points by adopting a vector type finite element method, and connecting the plurality of mass points by using a mass-beam-free unit to construct a global structure analysis model; and taking the bending moment-rotation angle relation as a force-deformation relation of the mass-free beam unit in the global structure analysis model, calculating displacement response of each mass point in the global structure analysis model, updating internal force in the calculation process according to the deformation of the mass-free beam unit and the bending moment-rotation angle relation, and outputting a simulation result of the prestressed beam structure. The method provided by the invention can accurately simulate the complete stress behavior of the beam from elasticity, cracking, yield to limit damage, and provides data support for subsequent research.

The present application relates to the field of building structure safety analysis, and particularly relates to a building structure safety analysis method fusing point cloud data. First, based on the obtained three-dimensional point cloud data, a three-dimensional space grid is constructed, and fixed space anchor points are determined. Then, based on the fixed space anchor points, the three-dimensional point cloud data is mapped into the anchor domain, an improved structure tensor neighborhood difference method is introduced, and a structure change intensity index is obtained. Finally, based on the structure change intensity index, an improved time-symmetry second-order difference structure evolution evaluation method is introduced, and a global structure safety index is obtained. The technical problems of being difficult to depict the dynamic evolution process of the structure over time, lacking the ability to identify gradual damage, being sensitive to point cloud data noise and prone to misjudgment, lacking a unified and stable spatial reference system, being difficult to align cross-frame data and affecting the overall analysis accuracy, and relying on experience rules for overall evaluation indexes and lacking data-driven quantitative expression are solved.

The application discloses a multilevel system model construction method, system, device and medium, and relates to the technical field of system engineering and system modeling. The method comprises the following steps: acquiring information data of a multilevel system to be constructed; determining a structure undirected graph according to the structure relationship and the construction objects between construction objects; performing regional segmentation according to a set path step based on the level corresponding to the construction objects and the structure undirected graph, and obtaining a plurality of regional network structures; for any regional network structure, determining the dependency relationship between the construction objects by using a data structure learning algorithm, and determining a graphical network according to the dependency relationship; determining a structure network model of the multilevel system to be constructed according to the structure undirected graph and all the graphical networks; and the structure network model is used for representing the mechanism characteristic information of the multilevel system in cooperative interaction. The application can perfect the global structure of the system and realize the construction of a global static model.

The invention discloses a building remote sensing image change detection method based on dynamic context aggregation, and belongs to the field of image detection. The method comprises the following steps: constructing a data set and carrying out sample division; a DCA-NET network model is constructed; performing parameter adjustment on the constructed network model; inputting the divided data set into a model for detection to obtain a change result; and carrying out quantitative evaluation on the change result. According to the invention, by introducing a dynamic context attention mechanism, the model can better integrate local and global information, so that the detection capability of a target change area is improved; compared with the problems that the correlation of neighborhood features is not fully considered and the standard convolution operation is limited to a fixed local receptive field in the prior art, the shape and the size of the convolution kernel can be flexibly adjusted, and the capturing ability of the model in a large-scale global structure and small-scale detail features is improved.

The invention discloses a corundum spinel material microstructure intelligent analysis method and device, and the method comprises the steps: obtaining an original microstructure image of a material, carrying out the standardization preprocessing, and inputting the image into a pre-trained material analysis model, so as to obtain a sintering temperature type and a prediction probability; the model comprises an input receiving module, a double-branch feature extraction module, a multi-modal feature fusion module and a classification decision module, wherein the double-branch feature extraction module comprises a local feature extraction branch and a global feature extraction branch which are arranged in parallel; and the model performs classification decision by fusing the local detail features and the global structure features which are extracted in parallel. The method also generates a decision thermodynamic diagram based on a model classification result, and reveals a microstructure basis of a model sintering temperature classification decision. According to the method, the problems of feature extraction limitation and black box decision making of an existing single model are solved, high-precision and interpretable intelligent analysis of the sintering temperature of the corundum spinel material is achieved, and reliable guidance is provided for material process optimization.

The invention discloses a method for realizing automatic generation of a carrying path based on MCS system data modeling, and relates to the technical field of path generation. The method comprises the following steps: obtaining a current carrying path diagram generated based on a material control system, and obtaining global structure information of the path diagram; according to a plurality of topological dimension features in the path diagram, calculating a corresponding path diagram structure complexity index, the topological dimension features including a path redundancy condition and a local area diagram density; comparing and judging the structural complexity index of the path diagram with a preset structural complexity threshold value; and if the complexity index is not less than a preset structure complexity threshold, triggering path diagram structure compression processing, and generating a new conveying path diagram. Through the mode, the global topological structure complexity problem of the current carrying path diagram generated by the material control system can be measured and handled, so that the automatic path construction has efficient initial performance, and meanwhile, the stability and maintainability under large-scale system operation can be supported for a long time.

The invention provides a federated learning encryption system and method based on model structure transformation. The system comprises computing equipment participating in training and a central server, the computing devices participating in training are connected with the central server in a wireless communication mode. According to the method, the global model is converted into the private local model structure before federated learning local training through model conversion, and the private local model structure is converted back to the global structure after the local training is finished, so that the local model structure of the local training is highly confidential in the whole federated learning training process, and the accuracy of the federated learning training is improved. Therefore, the central server or an external attacker cannot reversely deduce the private data information of the local client through the uploaded gradient information; compared with a traditional federated learning encryption method, the method has the advantages that local model information is encrypted through lightweight model conversion, gradient information is encrypted in an implicit mode, the privacy of federated learning is remarkably improved, and the method has very high practical value.

The invention discloses a method for constructing a three-dimensional spherical dynamic model of seabed plateau dive. The method comprises the following steps: solving mass, momentum, energy and component conservation equations under the condition of neglecting the dynamic influence of compressible fluid; applying mantle convection vitality; applying a viscosity jump; an autonomous plate-like construction behavior is shaped; embedding a seabed plateau with multi-parameter attributes; and finally completing the construction of the three-dimensional spherical earth dynamics model after model testing. The constructed three-dimensional numerical model adopts a three-dimensional spherical construction method with an autonomous class plate construction behavior, and fine description is performed on the seabed plateau by introducing different sizes and different rheological state parameters. The model not only can accurately reproduce the geodynamic phenomena such as bending and fracture of the plate in the subduction process of the seabed plateau, but also can reveal the response difference of the seabed plateau with different sizes and rheological characteristics to the local sea ditch structure, and realizes the cross-scale simulation of the influence from the local structure to the global structure.