166 results about "Constraint graph" patented technology

The invention relates to the technical field of artificial intelligence, can be applied to the digital medical field and the financial field, and discloses a causal graph model-based data processing method, device and equipment and a storage medium, the method comprises the steps of obtaining target data and a corresponding to-be-explained reasoning result, and constructing a causal graph according to the target data, obtaining a corresponding structured causal graph; performing causal constraint on the structured causal graph by adopting a causal constraint graph network learning mechanism to obtain an optimized causal graph; and performing anti-fact reasoning on the reasoning result based on the optimized causal diagram, and generating an interpretation result of the target data according to an anti-fact reasoning result. By fusing the causal graph model and the graph neural network technology and introducing the anti-fact reasoning module, accurate causal modeling and explanation of the high-dimensional nonlinear data are realized, the limitation of a traditional causal graph model in the aspect of processing the high-dimensional nonlinear data is solved, and the defect of a graph neural network output result in causal explanation is made up.

The invention relates to the technical field of mutual inductor tests, and provides a mutual inductor test abnormal data automatic filtering analysis method and system, and the method comprises the steps: obtaining magnetic field data, electric field data and thermal field data, respectively carrying out the feature extraction and weighted fusion, and obtaining a multi-physical field fusion feature vector; constructing an electromagnetic induction chain type propagation graph, calculating the weight of an edge in the graph, constructing a magnetic flux conservation constraint graph convolutional neural network, learning propagation and evolution modes of transformer test abnormity, and obtaining a coupling abnormity feature vector; the coupling anomaly feature vectors are classified, normal data, single physical field abnormal data and multi-physical field coupling abnormal data are recognized, and corresponding data are judged as abnormal data and filtered; and carrying out physical mechanism analysis on the filtered abnormal data to obtain an abnormal detection report. According to the invention, high-precision automatic identification, filtering analysis and physical mechanism traceability diagnosis of transformer test abnormal data are realized.

The invention provides a mountain power station current collection system state online monitoring method and device, and relates to the field of data processing. According to the method, distributed optical fiber data, environmental meteorological data and electrical measurement data are acquired to construct a multi-mode time sequence matrix, an operation state diagram is formed through graph structure representation and graph signal processing, a working condition mode cluster is obtained by combining dynamic clustering, a constraint diagram optimization framework is input, thermal, electrical and mechanical constraints are introduced, and an operation state track is generated. And constructing and fusing a non-stationary degradation path library and environment exposure history to form a dynamic degradation evolution model, and finally outputting in-service reliability and predicting residual life by a risk inference engine to realize online monitoring of the state of the current collection system of the mountain power station. By implementing the technical scheme provided by the invention, the accuracy of online monitoring of the state of the current collection system of the mountain power station is improved.

The invention discloses a knowledge graph-based software material consistency automatic checking method, which comprises the following steps of: collecting and processing a software declaration material to form a standardized original material set; forming an initial entity set and a candidate relation set based on the normalized original material set; constructing a heterogeneous knowledge graph based on the initial entity set and the candidate relationship set; executing cross-modal semantic anchoring on the heterogeneous knowledge graph to generate a semantic anchoring mapping table and an anchoring confidence value set; generating an enhanced constraint graph based on the semantic anchoring mapping table and the anchoring confidence value set; outputting a consistency detection result set and a minimum conflict subgraph set based on the heterogeneous knowledge graph and the enhanced constraint graph; and based on the consistency detection result set and the minimum conflict subgraph set, generating a difference check report and a repair suggestion set. According to the method, heterogeneous knowledge graph construction and semantic anchoring are adopted, and automatic consistency checking of the soft material is realized.

The invention provides a dynamic space-time air quality prediction method based on a physical constraint graph attention network, and the method comprises the steps: carrying out the collection and preprocessing of multi-source data, and obtaining the pollutant concentration and meteorological element historical sequence of a monitoring station; constructing a dynamic space-time diagram network fusing distance attenuation and a wind direction driving transmission path based on the longitude and latitude of the station and the real-time meteorological field; multi-cycle time sequence features are extracted in a self-adaptive mode through an enhanced time sequence network ETNet, and attention weight distribution is guided through physical priori; time sequence features are embedded into the dynamic graph, a physically constrained graph diffusion attention network PC-GDAN is input, and space-time diffusion modeling is achieved through a differential operator embedded into an atmospheric diffusion equation; time sequence and space features are fused, mass conservation and diffusion smoothness constraints are introduced in decoding prediction, and a concentration prediction result conforming to a physical rule is generated. The method effectively fuses physical mechanisms and data, has strong space-time modeling capability and physical interpretability, and significantly improves the precision and robustness of air quality prediction.

The invention belongs to the field of electric power inspection, and provides an electric power scene full-chain ubiquitous dynamic sensing body intelligent inspection operation method and system, and the method comprises the steps: carrying out the semantic analysis of an initial electric power task, combining a target electric power scene model, constructing an executable constraint set, and matching each body intelligent agent in an inspection domain; fusing the task information and a pre-constructed electric power knowledge graph, performing graph path search on the fused electric power knowledge graph by using a pre-trained large language model to obtain a disassembled atomic action sequence, and issuing the disassembled atomic action sequence to a matched and determined corresponding body agent; and evaluating the risk of candidate actions in the atomic action sequence by using the multi-modal task-environment constraint graph to realize preliminary optimization, and dynamically adjusting according to real-time environment information in the target power scene, the state of the agent with the body and the current task completion condition until the atomic action sequence is executed. According to the invention, the intelligence, the accuracy and the safety of the inspection operation can be obviously improved.

The invention discloses an industrial equipment remote operation and maintenance management method based on digital twinning, and aims to solve the problems of deduplication, association and suppression of events and alarms and root cause fusion positioning under complex working conditions. According to the method, a multi-relation layered signed causal graph is constructed, a dynamic causal mask is generated according to a twinning state, causal structure learning is carried out under constraint, and graph anti-fact intervention and simulation of physical budget constraint and historical dual-channel consistency verification are combined; and the root cause priority and the evidence chain are generated by adopting topology perception contribution degree sorting, and the suppression strategy parameters are formed, so that the technical effects of reducing false alarms and repeated alarms, improving the root cause identification accuracy and processing efficiency and stably outputting the linkage work order are realized.

The invention discloses a multi-terminal detection score offset correction method and system, particularly relates to the field of financial risk control, is used for solving the problem of steady-state offset generated in the merging process of multi-terminal detection scores, and comprises the following steps: receiving cross-terminal original behaviors and collecting metadata under unified time reference and unified field specifications; the method comprises the following steps: extracting a terminal fingerprint table according to a sampling beat, a sensing path, a rendering mode and a quantification strategy, generating a time alignment table, dividing a stable fragment, a change fragment and a transition fragment by taking the terminal fingerprint table and the time alignment table as conditions, and constructing a segmented order-preserving mapping relation table; a cross-terminal consistency constraint graph is generated based on the segmented order-preserving mapping relation table to form a correction rule set, rules are verified in an offline playback window, a micro-correction list is output, and the correction rule set and the micro-correction list are read in the online merging stage to execute correction and output a unified scale score. And meanwhile, generating a feedback abstract, submitting, filing and publishing a version.

The invention relates to a physical constraint fused graph structure UNet neural network for long-time time domain difference optical simulation method and system, and the method comprises the steps: 1, randomly constructing a device parameter data set through an FDTD method, obtaining FDTD uniform grid data, and carrying out the data preprocessing; 2, constructing a graph structure Unet model; 3, inputting the device parameter data set into the graph structure Unet model for end-to-end training; 4, predicting electromagnetic field scattering after a long-time light source encountering period is carried out on given device parameters and initial electromagnetic field distribution; and 5, comparing the electromagnetic field distribution of different device shapes, which is predicted for a long time, with the electromagnetic field value calculated by FDTD, and further judging the prediction accuracy of the graph structure Unet network model. According to the method, one-time long-time sequence prediction of electromagnetic scattering is completed by using the graph structure Unet model.

The invention discloses a high-precision robust robot detection method and system based on constraint pose graph optimization, and the method comprises the steps: collecting point cloud data of a detection object in multiple poses, and introducing a standard sphere and a standard ball rod in the collection process for constructing a graph constraint; a robust constraint pose map optimization algorithm framework SYMM-RCPGO is constructed; substituting the point cloud data into SYMM-RCPGO to construct an attitude graph and a constraint attitude graph to obtain a target function; iteratively solving the SYMM-RCPGO in the SE3N by combining the objective function according to the graph constraint, and using a dynamic constraint boundary algorithm to relax the tightness of the constraint to obtain the three-dimensional point cloud data of the detection object with a complete appearance; and registering the measured three-dimensional point cloud of the detection object with the designed CAD model, and accurately calculating the pose matrix of the detection object to complete the positioning of the detection object. According to the method, the SYMM distance function with a higher D-optimal measurement upper limit, the robust Welsch function and the constraint graph are utilized, so that the method is superior to a traditional PGO in the aspects of accuracy and robustness.

The embodiment of the invention provides a shipyard steel framed bent scheme automatic optimization method and system. The method is applied to the technical field of buildings and comprises the following steps: selecting an initial size of a shipyard steel bent column beam section and bent design information based on engineering experience library data, and constructing high-precision input characteristics based on factory design information; based on the high-precision input features, constructing a physical constraint graph neural network, and predicting an initial design scheme meeting the physical constraint; and carrying out multi-objective reinforcement learning optimization on the initial design scheme meeting the physical constraint, and carrying out dynamic adjustment in a virtual simulation environment to obtain optimized steel bent design parameters. In this way, according to the method, the shipyard data are monitored in real time, the graph neural network is constructed for feature extraction and parameter optimization, the optimal steel frame parameters are obtained in dynamic adjustment, and the practical value is high.

The invention relates to the technical field of city cadastral data analysis, and discloses a city update district boundary division method and system based on cadastral data, and the method comprises the steps: collecting the division associated cadastral data of a target city update district, and employing an improved DTW algorithm to obtain a city update district boundary; the method comprises the following steps: screening current situation boundary effective points and newly-added candidate points with history continuity from current situation measured data, extracting spatial features and ownership features for fusion, constructing a feature matrix by using a fusion feature vector, constructing a spatial ownership double-constraint graph neural network, predicting a final boundary point set of a target city update area, and obtaining a target city update area. And generating a final land parcel boundary. Therefore, by introducing the improved DTW algorithm and the spatial ownership double-constraint graph neural network, while accurate association of historical and current boundary data is established, joint optimization of spatial rationality and ownership rationality is realized by utilizing double-constraint feature fusion, and the division precision and division efficiency of the urban update district boundary boundary are improved.

The invention relates to the technical field of electronic design automation, and discloses a standard cell layout legalization method and system based on a graph theory constraint graph, and the method comprises the steps: building a bidirectional directed constraint graph based on an existing layout and a design rule, and building a directed edge between adjacent graph edges according to the design rule; traversing the constraint graph, detecting whether a directed edge which does not meet a constraint condition exists, and if so, identifying the directed edge as a constraint violation edge; for two end points of each constraint violation edge, searching a key path based on a depth-first search algorithm, and determining a maximum movable distance in combination with a greedy strategy; moving the endpoint to a legal position according to the critical path and the maximum movable distance, and updating the position of a related node in the constraint graph; and mapping the position changes of the nodes in the constraint graph back to the corresponding graph edges in the layout, and outputting the legalized layout, so that efficient and automatic repair of design rule violation is realized through the method, and the layout compliance rate and the area optimization effect are remarkably improved.

The invention relates to the technical field of power task execution, and provides a vision-language-action fused power task security execution method and system, and the method comprises the steps: obtaining a natural language instruction, a visual image and an intelligent agent state of a current task, constructing a current state, initializing a candidate action set of an intelligent agent in the current state, and executing the current state; in combination with the multi-modal task-environment constraint graph, evaluating risk scores of candidate path points contained in the candidate actions, and carrying out preliminary screening on the candidate actions; for each candidate action subjected to preliminary screening, through a risk assessment model, obtaining an action risk probability, and screening out candidate actions of which the action risk probabilities are greater than a threshold value; and encoding the natural language instruction, the visual image and the intelligent agent state, obtaining observation information characteristics through a cross attention mechanism, selecting an optimal action from the reserved candidate actions based on the observation information characteristics, controlling the intelligent agent to execute an electric power task, and continuously updating the state and the action. And dangerous operation is avoided from the source.

The invention discloses a motor temperature prediction transfer learning method based on a physical constraint graph neural network, and the method comprises the following steps: S1, constructing an undirected graph with a label based on a physical structure of a source domain motor; s2, on the basis of the undirected graph of the label, constructing a physical constraint graph neural network; s3, training the physical constraint graph neural network by using operation data and temperature data of a source domain motor to obtain a pre-training model; s4, on the basis of the physical structure of the target motor, constructing an undirected graph of the target motor, and multiplexing the heat capacity parameter and the heat conductivity function in the pre-training model so as to obtain an initialized physical constraint graph neural network for the target motor; and S5, obtaining a temperature prediction model of the target motor. According to the method, the physical heat transfer rule is embedded into the transferable graph neural network structure, so that a high-precision temperature prediction model with physical interpretability can be quickly constructed for a new motor only by using small-batch data.

The invention relates to the technical field of the intelligent power grid industry, in particular to a power distribution and utilization ring network fault intelligent troubleshooting method and system. According to the method, a looped network situation primitive is generated through multi-source heterogeneous data fusion. The primitives are input into synchronous compression transformation to extract transient fault initial features, and the transient fault initial features are mapped into fault propagation path features through a graph convolutional network considering topological structure constraints; then, inputting the features into a space-time correlation prediction model, integrating defect type judgment and trend pre-judgment features, and generating a fault diagnosis evidence set; and starting an investigation mechanism based on the evidence set to determine an interval to be investigated, and performing closed-loop verification by constructing a fusion judgment matrix based on multi-modal physical sensing information to realize physical space locking of a fault defect point. According to the method, deep learning diagnosis and multi-mode positioning are integrated, and the detection precision and the positioning efficiency of looped network defects and the identification capability of transient hidden faults are effectively improved.

The invention discloses a dynamic constraint driven chemical three-dimensional pipeline construction drawing generation system and method, and the system comprises a multi-modal data fusion module which obtains multi-modal fusion data; the dynamic constraint map module is used for carrying out structured representation on the three-dimensional pipeline system by utilizing the multi-modal fusion data to obtain a pipeline dynamic constraint map; the three-dimensional space reasoning module is deployed with a graph neural network, carries out deep fusion of a space attention mechanism based on a pipeline dynamic constraint graph, and generates a three-dimensional pipeline construction drawing; and the intelligent drawing output and verification terminal module is used for automatically marking the three-dimensional pipeline construction drawing and synchronously verifying the marking information and the mechanical property. According to the method, the dynamic constraint graph is constructed by fusing the dynamic process parameters and the construction constraints, and spatial reasoning is carried out by utilizing an intelligent algorithm, so that the spanning of chemical pipeline design from static modeling to dynamic response is realized, and the problems of low efficiency, high conflict rate, dynamic constraint deficiency, cooperative lag and the like in a traditional method are effectively solved.

The invention discloses a wind turbine blade internal damage diagnosis method based on a sound field graph neural network, and belongs to the technical field of wind turbine blade state detection, and the method comprises the steps: deploying a microphone array in a cabin, and collecting an acoustic signal when a blade rotates; constructing a space sound field graph structure by taking the microphone as a node and the sound wave propagation path as an edge; extracting nonlinear acoustic features by using a physical constraint graph neural network PC-GNN; generating a damage embedding vector based on self-supervised training contrast learning; and outputting a damage probability thermodynamic diagram and positioning information. According to the method, a directional microphone array is deployed in a cabin, and a sound wave propagation space diagram structure is constructed; designing a physical constraint graph neural network PC-GNN, and embedding an acoustic wave equation as a regularization item; the problem of scarcity of damaged samples is solved by adopting self-supervised contrast learning; and finally outputting a positioning thermodynamic diagram of the internal damage of the blade.

The invention discloses a standardized work management system for a standard system revision project, and the system comprises a heterogeneous data fusion module which captures the multi-modal process data of each stage in real time, and converts the multi-modal process data into a process state vector sequence; a time sequence constraint graph rule base is built in the compliance graph calculation module, a process state graph with time sequence attributes is constructed, and the multi-dimensional compliance deviation degree is calculated through graph mode matching; the strategy generation and conflict resolution module is used for generating a candidate set with a weight adjustment instruction through a learnable strategy priority ranking model based on the deviation degree and resolving logic conflicts; and the trusted execution and feedback closed loop module atomizes an execution instruction and carries out uplink evidence storage, and collects project response data to update rule base parameters and model weights online. According to the method, automatic and accurate compliance evaluation is realized through graph calculation, self-evolution of the system is realized through intelligent strategy generation and credible execution feedback, and the management efficiency, compliance and intelligent level are comprehensively improved.

The invention relates to the technical field of multi-modal information processing, and discloses a multi-modal information generation and enhancement method based on an AI large model, comprising the following steps: step 1, acquiring multi-modal original data and task prompts, generating a unified token sequence and constructing a constraint graph; 2, setting a modal quota, a coverage threshold value and a consistency threshold value according to the importance degree of the constraint node; step 3, generating a visibility mask and establishing a quota account book, and controlling the visibility of dependent resources; 4, when the generation section meets a threshold value, determining a commitment section and generating an abstract fingerprint; step 5, weighting candidate output according to the quota account book, and adjusting mask regeneration when no feasible candidate exists; step 6, calculating a quality score, and performing backfill enhancement on the low-quality sub-segment; and 7, carrying out statistics on a modal conflict rate and a semantic drift rate, and adaptively adjusting a quota and a constraint weight. According to the method, semantic consistency maintenance, resource adaptive allocation and generation quality enhancement in the multi-modal information generation process are realized.

The invention belongs to the technical field of numerical weather forecast post-processing, and discloses a physical consistency intelligent optimization method of numerical weather forecast data and a related device. The physical consistency intelligent optimization method comprises the steps that numerical weather forecast data to be optimized and a dynamic graph adjacency matrix of the numerical weather forecast data to be optimized serve as input, a trained physical constraint graph neural network is used for conducting correction prediction, and correction is obtained; and superposing the numerical weather forecast data to be optimized with the correction to generate optimized numerical weather forecast data. The physical consistency intelligent optimization scheme disclosed by the invention can be used for carrying out increment correction and physical consistency constraint on the gridding meteorological field generated by the existing NWP, and can realize correction of local errors and coordination of physical relationships among meteorological elements while keeping the overall trend of numerical weather forecast, so that the prediction accuracy of the numerical weather forecast is improved. And the reliability and the stability of short-time strong weather and medium-term trend forecasting are effectively improved.

The invention belongs to the technical field of earth science, and particularly discloses a geological constraint-based graph reinforcement learning mineral product prediction method, which comprises the following steps of: constructing states, rewards and actions in graph reinforcement learning, adding geological constraints in a process of constructing graph reinforcement learning environment reward feedback, rewards are fed back to the intelligent agent by the environment under the constraint of geological knowledge, and the intelligent agent is promoted to make a multi-angle decision on the mineralization potential; constructing a Markov chain decision process of graph reinforcement learning, and realizing deep coupling of an intelligent agent and an environment; the method comprises the following steps: designing a loss function, establishing a target for learning of an intelligent agent, training a model by continuously reducing a gap between a target network and an intelligent agent network, gradually identifying and distinguishing known mineralization, potential mineralization and non-mineralization in an interaction process of the intelligent agent and an environment, and delineating a metallogenic prospective area. The method can effectively improve the recognition precision of model mineral prediction.

The invention belongs to the technical field of power distribution network fault removal, and discloses a power distribution system fault rapid removal method, which comprises the following steps: in the operation process of a power distribution system, collecting multi-dimensional operation data and a corresponding control log in real time, and applying a timestamp mark to the multi-dimensional operation data based on a synchronous triggering index mechanism; mapping the multidimensional operation data with the timestamp mark into a compressed state expression, constructing a topological electrical parameter constraint graph, and generating a topological perception state matrix; in combination with instruction response delay in a control log, action lag factors are extracted, state representation is embedded, and a composite state vector sequence is formed; on the basis of the composite state vector sequence, behavior events with path influence in the power distribution system are extracted, behavior influence accumulation is calculated by adopting an index time decay mechanism in combination with occurrence time and action intensity, and a dynamic convolution behavior index is generated; the timeliness and reliability of fault response of the power distribution system are improved, and safe and stable operation of the power system is guaranteed.

The invention belongs to the technical field of data processing, and provides a BIM data intelligent matching and conflict detection method, and the method comprises the steps: generating a fusion feature of a component through fusing a geometric topology feature and a semantic constraint feature; through alignment and coupling of the installation state sequence and the compliance state sequence, a space-time state sequence of the component is formed, so that the construction constraint graph can fit the actual dynamic change of construction; the method comprises the following steps: injecting a space-time state sequence into entity nodes by taking components as the entity nodes and a design specification and a construction process as constraint nodes, constructing a construction constraint graph, and predicting a conflict propagation path between the components through a time sequence graph neural network; the prediction result of the conflict propagation path is subjected to actual construction verification, the verified prediction result is subjected to influence tracing and grading, and the comparison data of the actual construction result and the prediction result is combined to adjust the parameters of the time sequence diagram neural network and the edge weight of the construction constraint diagram, so that the prediction accuracy of the conflict propagation path is further improved.

The application discloses a power metering drift correction method and system based on transformer area line loss and a medium, relates to the technical field of power meters, and comprises the following steps: constructing a four-level topology of a transformer area-branch-meter box-meter and attaching a communication quality label; calculating a measured line loss rate in layers, combining a physical constraint graph neural network to obtain a theoretical line loss; comparing a deviation and combining communication quality to determine suspicious nodes; performing time sequence decomposition on candidate meters to identify drift types; and adaptively verifying and grading correcting according to the drift types. The application solves the technical problems that it is difficult to determine the power metering drift node in the prior art, the drift type cannot be effectively distinguished, the operation and maintenance efficiency is low, and drift identification is prone to errors, and achieves the technical effects of improving line loss analysis accuracy by constructing a refined topology and a graph neural network, accurately identifying the drift type by combining a communication quality label and time sequence decomposition, and improving the reliability of power metering drift processing and the operation and maintenance efficiency.

The invention provides a rock mass fracture network medium reconstruction method and system based on image tracking, and belongs to the technical field of fracture network reconstruction, and the method comprises the steps: obtaining a to-be-reconstructed fracture network, and carrying out the preprocessing of the to-be-reconstructed fracture network, and obtaining a binary image; skeletonizing the binary image to obtain a single-pixel wide fissure main ridge; identifying end points and intersections of the fracture skeleton based on a neighborhood connection relation, and performing constrained graph tracking by taking the end points and the intersections as graph nodes, taking a connection skeleton sequence between the end points and the intersections as graph edges and taking the end points as starting points to obtain an ordered vertex coordinate sequence of each fracture broken line; and performing multi-point coordinate calibration on the ordered vertex coordinate sequence, mapping the ordered vertex coordinate sequence to a target two-dimensional coordinate system, generating a polyline command text file based on the mapped coordinates, and reconstructing the fracture network medium by using the polyline command text file. The method is oriented to data flow of heterogeneous software, a special interface is not needed, batch processing is supported, and the efficiency and accuracy of cross-platform modeling are improved.

The invention discloses a heuristic-based iterative legalization method and system for an analog circuit layout, and the method comprises the steps: dividing layout groups according to a special structure and an array module in a layout based on the initial layout of the layout containing overlapped devices or modules; constructing a constraint graph for describing the relative position relationship between the layout groups, and solving symmetric constraint conflicts in the layout so as to ensure that all symmetric constraints are met; performing simulation processing on a non-overlapping constraint, a minimum area constraint and a minimum spacing constraint through a code simulation solver to generate an initial legal layout; and iteratively adjusting the constraint graph based on a heuristic method, identifying a suspicious group and adjusting the constraint edge of the suspicious group, and re-verifying and repairing the symmetric constraint after each adjustment until the layout area converges. According to the method, the symmetric constraint is synchronously verified and repaired in each iterative optimization, so that the layout area is minimized on the premise of ensuring the layout legality, and the solving efficiency is greatly improved at the same time.

The invention relates to a pointer analysis-based incremental data flow path extraction method. The method comprises the following steps of: firstly, acquiring an initial pointer variable related to an incremental code as a task variable set; then, performing reverse query on the task variables based on the pointing relation constraint graph to obtain a basic variable set and a memory object pointed by the basic variable set; and after the working set is initialized, executing pointing relation propagation, propagating the memory object to a pointing set of the task variable, and classifying a new association pointer found in propagation into an indirect variable set. And executing explicit dependency query on the indirect variable set, adding a new dependency variable into the task set, repeating the process until the constraint graph is stable, and finally obtaining pointing information of all variables. And an inter-process data dependence sub-graph is constructed, and an incremental data flow path is extracted. According to the method, through a locality principle and a reverse query strategy, an analysis range is limited to incremental code related variables, and the overhead and memory problems of global analysis are avoided.

The invention relates to an industrial quality prediction method based on priori knowledge constraint graph convolution, and the method comprises the steps: collecting multivariable time series data containing quality variables and process variables, deeply mining the Granger causality between the variables based on the multivariable time series data, preliminarily obtaining a directed information transfer matrix, and carrying out the deep mining of the Granger causality between the variables; combining a prior sub-process knowledge mask matrix to dynamically adjust and refine an information transfer relationship between variables so as to generate a dynamic adjacency matrix; and designing a multi-head space-time diagram convolution long-short-term memory network based on the dynamic adjacency matrix to learn long-short-term space-time characteristics. According to the method, correlation between a quality variable and a process variable is effectively mined by adopting a Granger causal relationship based on constraint priori knowledge, and a long-short term dependency relationship is captured by utilizing a multi-head space-time diagram convolution long-short term memory network, so that the accuracy of quality prediction of an industrial system is improved.

The invention relates to the technical field of artificial intelligence, in particular to a customer service reply method and system of a multi-modal fusion architecture and a storage medium, and the method comprises the following steps: obtaining initial input data of a user, preprocessing the initial input data, and generating a standard data source; performing decision compliance analysis on the standard data source to generate a decision judgment result; and generating a corresponding decision scheme according to the decision judgment result. According to the invention, through a space-semantic double-constraint graph network, the technical defects of fragmentation and low standardization degree of multi-modal data identification by traditional intelligent customer service are effectively solved, and the problem of insufficient adaptation of a general identification technology to a government affair format is overcome; in a semantic constraint dimension, spatial features and a government and enterprise professional semantic dictionary are deeply fused based on a node association mechanism of a graph network, conversion from multi-modal data to structured policy elements is automatically completed, and the recognition accuracy is improved by more than 40% compared with that of a traditional NLP technology.