381 results about "Field based" patented technology

The invention relates to the technical field of natural language processing, and discloses a power generation industry data intelligent treatment method, device and equipment based on a large model, and the method comprises the steps: constructing a multi-source heterogeneous task data set covering structured and unstructured information, completing the fine tuning training of a plurality of industry sub-fields based on a language model, forming a large language model set with specific scene adaptability; constructing a multi-view semantic representation structure for actual input data, integrating modeling task intention, application scene and model adaptability, predicting an optimal target model and a Top-K candidate model, and generating a unified semantic embedding vector; and realizing accurate matching of the structured knowledge fragments through the graph neural network. The problems that an existing model is insufficient in semantic understanding, inflexible in model selection and inaccurate in knowledge calling in the data management process are solved, the requirements of diversified tasks for accuracy and specialty are met, and then management of data assets is facilitated.

The invention relates to a bidding document total-factor blind-state desensitization method fusing multi-modal artificial intelligence, and the method comprises the following steps: S1, obtaining an original bidding document package, carrying out the data preprocessing, and obtaining a standardized document set; s2, according to the standardized document set, performing multi-modal content identification and alignment to obtain a multi-modal content library; s3, constructing a sensitive information classification system based on domain expert knowledge and a historical case library; s4, constructing a dynamic rule base according to the sensitive information classification system; s5, according to the multi-modal content library and the dynamic rule library, in combination with deep learning, multi-modal sensitive information intelligent detection is carried out, and a sensitive area set is confirmed; and S6, based on the dynamic rule base, performing intelligent desensitization on the sensitive area set to obtain a desensitized document. According to the method, the risk of sensitive information leakage is effectively reduced while the bidding file expression effect is improved.

The invention relates to a business process standard dynamic optimization method and system based on a large language model, and belongs to the technical field of computer application, and the method comprises the following steps: constructing a data collection strategy template; performing anomaly detection on the key field based on a preset three-level anomaly detection mechanism to obtain standard structured data; constructing a process execution accurate quantitative evaluation model, and constructing a process topological graph by using a process mining technology; calculating an abnormal probability of each node in the process topological graph by using a multi-task graph neural network, and obtaining a high-attention area in the process topological graph based on the abnormal probability; analyzing the multi-source heterogeneous data associated with the high-attention area through a large language model, positioning root causes, calculating root cause influence weights, and obtaining a root cause list; establishing an intelligent bidirectional feedback mechanism to generate optimization suggestions of the standard process; according to the method, the problems of serious data redundancy, insufficient standard dynamics and lack of feedback mechanisms in the traditional flow standard optimization technology in the power industry are effectively solved.

The invention discloses an enterprise information dynamic modeling method based on multi-dimensional data driving, which relates to the technical field of enterprise information modeling, and comprises the following steps: calculating semantic similarity and semantic offset amplitude among fields according to time sequence changes of semantic vectors in a plurality of modeling cycles, identifying whether driving dimension fields are replaced or not, and if yes, judging whether the driving dimension fields are replaced or not; generating a field replacement matching pair for recording a field replacement relationship; and on the basis of the field replacement matching pair and in combination with a historical modeling path structure, establishing a dependency mapping relationship between the driving dimension field and the path node, executing path structure backtracking calculation, determining a modeling path dependency state under the condition that the driving dimension field is replaced, and generating a path dependency identification matrix. According to the method, the problem of modeling path mismatch caused by drive field replacement is solved, automatic identification and dynamic updating of the modeling structure are realized, and the accuracy and adaptability of the model are improved.

The invention provides a vertical field data construction method based on a large model, which belongs to the technical field of data processing and artificial intelligence, and comprises the following steps: converting a vertical field source document into an intermediate format text, and segmenting the intermediate format text into a plurality of text blocks; inputting the text blocks into a pre-trained generative language model, guiding the pre-trained generative language model according to pre-designed cue words to generate a plurality of candidate questions according to the content of the text blocks, and performing preliminary screening and fine screening on each candidate question to obtain a question set; pre-defining a mode of a knowledge graph according to field characteristics of the vertical field, processing all text blocks based on an information extraction model, and constructing a field knowledge graph; and performing local context retrieval on each final question in the question set based on the text block of the question source, performing global knowledge retrieval based on the domain knowledge graph, and generating a final answer and a final thinking chain. The method is suitable for different vertical fields, the data quality can be effectively improved, and the problem generation accuracy is guaranteed.

The invention relates to the technical field of network security, and discloses a data desensitization and access control method based on data consanguinity and sensitivity identification, which comprises the following steps of: constructing a data consanguinity graph for describing derivation, circulation and dependency relationships among fields; identifying a sensitivity level of each field based on a field consanguinity context provided by the graph; in response to a data access request of a user, dynamically generating an access control strategy for the target field in combination with a service scene, a user role and the sensitivity level of the target field, the strategy specifying access permission, access denial or access after desensitization; and executing the strategy through a data interceptor deployed in a data access key layer, releasing and shielding the data or outputting the data after real-time desensitization by calling a specified algorithm. According to the invention, accurate identification and dynamic and consistent fine-grained access control of the sensitivity of the network data are realized, and the security protection capability of the network data is improved.

Certain aspects of the disclosure provide for information retrieval that exploits context derived from document structure. Source documents can be preprocessed to identify fields and determine context attributes related to each field based on the structural layout of a source document. Resource documents can also be preprocessed to segment a resource document into passages and determine context related to the passages based on structural layout. Queries pertaining to a field can be enhanced by adding context metadata associated with the field. A query embedding can be generated and compared with previously generated passage embeddings to locate candidate matches based on similarity. A machine learning model can be provided with the top-ranked passages and tasked with re-ranking the passages based on relevancy to the original query. The highest re-ranked passage or set of passages can be output in response to the query.

The invention discloses a shock wave overpressure field reconstruction method based on physical information, and relates to the technical field of artillery launching, and the method comprises the following steps: building a multi-scene overpressure empirical formula library comprising a Henrych formula, an M.A.Sadovskyi formula and national defense design specifications, and embedding physical prior knowledge into a neural network for training through a weighted loss function; an attention module with a channel-space cooperative enhancement capability is formed by fusing the one-dimensional efficient channel attention, the extrusion excitation network and the dimension reduction CBAM; according to shock wave propagation characteristics, a forward shock wave field subtask and a muzzle rear wave front subtask are divided, and regional prediction is achieved through dynamic characteristic routing; and optimizing network parameters by adopting a low data volume training strategy, and outputting overpressure peak field distribution. According to the method, through construction of a physical data fusion constraint mechanism and an EC-SEAM mixed attention mechanism, network physical interpretation and near field mutation feature extraction capability are improved, and dependence on labeled data is remarkably reduced.

The embodiment of the invention provides a data query method and system. The data query method comprises the steps that a data query request associated with a target field is received; identifying query intention information corresponding to the data query request, and constructing a query prompt word corresponding to the data query request according to the query intention information; the query prompt word is input into a target distillation model to be processed, a data query sequence corresponding to the data query request is obtained, and the data query sequence comprises a data query statement and a data query reasoning chain associated with the data query statement; and executing the data query statement to obtain data query information, and reading target data in a target data table associated with the target field based on the data query information to serve as a response of the data query request.

The invention belongs to the technical field of marine environment monitoring and numerical simulation, and relates to a three-dimensional marine element field reconstruction method based on multivariable empirical orthogonal function decomposition. Comprising the steps of S1, multi-source data collaborative preprocessing; S2, MEOF joint modal extraction; S3, vertical projection operator optimization construction; S4, three-dimensional element field dynamic reconstruction; according to the method, through organic combination of multivariable modal decomposition and physical constraint projection, limitation of precision, efficiency and adaptability in three-dimensional field reconstruction in the prior art is broken through, key technical support is provided for real-time monitoring of the ocean dynamic environment, and the method can be widely applied to the fields of ocean disaster early warning, underwater engineering safety guarantee and the like.

The invention discloses an intelligent control system and method for wet raking and field returning operation of rice straw based on the Internet of Things, particularly relates to the field of agricultural machinery automation, and is used for solving the problems of tilling depth fluctuation, straw floating and winding and uneven burying thickness caused by muddy water in traditional wet raking operation. The method comprises the following steps: acquiring torque and position information of a rotary blade shaft in real time, establishing a torque sequence and a position index, executing edge detection on the sequence to divide stable sections, generating section labels, matching an action template from a rule base, issuing a tilling depth adjustment and speed correction instruction, and recovering an execution receipt to evaluate torque response. The burying uniformity and the tilling depth stability are calculated and written into a quality account book, and adjacent section optimization strategy switching is retrieved; torque is used as core sensing, sectionalized intelligent regulation and control of the operation process are achieved, continuous and consistent burying and ground surface smoothness are guaranteed, and operation efficiency and quality are improved.

The invention discloses an equipment system logistics support large model training method and system and a storage medium. The method comprises the steps of obtaining multi-source heterogeneous data in the field of equipment system logistics support; constructing a domain knowledge base based on the multi-source heterogeneous data, and generating a training sample of a to-be-trained large model based on knowledge extraction; constructing a multi-stage instruction template; based on the domain knowledge base and the instruction template, executing multi-stage instruction fine tuning processing on the to-be-trained large model to obtain a domain large model; the multi-stage instruction fine-tuning processing comprises a task cognition fine-tuning stage, a domain knowledge fine-tuning stage, a reasoning modeling fine-tuning stage and a generation optimization fine-tuning stage; an operation planning optimization code is generated through the field large model; executing the operation planning optimization code and recording an execution result, and constructing an error correction data set based on error information in the execution result; performing feedback fine tuning processing on the field large model based on the error correction data set; and evaluating the model performance of the field large model, and iteratively updating the field knowledge base and the instruction template according to an evaluation result.

The invention discloses a self-adaptive background grid simplification and size field generation method, which comprises the following steps of: reading a CAD (Computer Aided Design) geometric model, generating an initial fine triangular background grid, and generating an initial size field based on geometric characteristics and preset parameters; according to the background grid and the initial size field, carrying out feature coding on edges and nodes, and inputting into a graph convolutional network to score grid edges, thereby obtaining an edge collapse priority score; screening out a candidate edge set according to the edge collapse priority score, and executing edge collapse operation on edges in the set to obtain a background grid after topology simplification; based on the background grid after topology simplification, executing local re-gridding and vertex projection interpolation operation, and generating a simplified size field corresponding to the local re-gridding and vertex projection interpolation operation; and iteratively executing the steps until a termination condition is met, and outputting the finally simplified background grid and the size field thereof. According to the method, the background grid element number and the size field query time are remarkably reduced, and the method is suitable for automatic grid generation of a complex engineering model.

A floating-point data precision conversion method includes determining a bit width of a second mantissa field based on a coded value of a first exponent field. The floating-point data precision method further includes determining a reserved coded value and a discarded coded value in a first mantissa field. The floating-point data precision method further includes, if the coded value of the first exponent field is greater than or equal to a first preset threshold, performing a rounding operation on the reserved coded value based on a coded value that starts from a most significant bit and whose bit width is a preset bit width in the discarded coded value, to obtain a coded value of the second mantissa field.

The invention provides a method and device for identifying and diagnosing the wiring state of a combined junction box, and the method comprises the steps: collecting a multi-view two-dimensional image and multispectral three-dimensional point cloud data, and carrying out the calibration and fusion of a sensor, thereby forming a multispectral fusion point cloud; image features are decomposed into domain invariant semantic features and domain related style features by using a cross-domain feature decoupling mechanism, and a semantic segmentation result and a two-dimensional topological semantic graph are generated in combination with a double attention mechanism; constructing a three-dimensional geometric spectrum field based on implicit neural representation, and obtaining accurate three-dimensional space coordinates of a wiring assembly and a joint through differentiable rendering and consistency constraint optimization; and constructing a space-time consistency enhanced wiring information graph by taking the space coordinates as anchor points, carrying out multi-scale anomaly detection by adopting a self-supervised contrast learning model, and generating a defect significance graph and a diagnosis result. According to the method, the problems of poor identification stability, low cross-modal registration precision, difficulty in unknown defect detection and the like in a complex environment are effectively solved, and the automation degree and reliability of wiring state identification are remarkably improved.

The invention discloses a fault analysis system and method for ship equipment, and the system comprises a multi-source sensing synchronization module which is configured to carry out the phase synchronization of the operation data of multiple equipment through a hardware phase locking mechanism; the multi-physical field modeling module is coupled to the multi-source sensing synchronization module, and is configured to construct a device virtual state field based on the multi-device operation data after phase synchronization, and generate a dynamic parameter representing a multi-device state time sequence imbalance degree; the adaptive diagnostic analysis module comprises a deep learning fault analysis model, and recessive feature extraction parameters and fault evolution time sequence modeling parameters of the deep learning fault analysis model are adaptively adjusted in real time according to dynamic parameters so as to output system-level fault characterization parameters; and the predictive decision-making module is configured to generate decision-making information containing fault development trend prediction and a preventive maintenance window based on the system-level fault characterization parameters. And high-precision correlation diagnosis and predictive maintenance decision-making of early faults of multiple devices of the ship can be realized.

The invention discloses an industrial drawing structured extraction method and device, equipment and a medium, and relates to the technical field of drawing processing, and the method comprises the following steps: obtaining an industrial drawing file; the industrial drawing file is analyzed, candidate fields and corresponding evidence source information are obtained, and the evidence source information comprises an evidence text and a coordinate position in a drawing; calculating the evidence coverage rate of each candidate field according to the evidence source information; judging whether the evidence coverage rate is greater than or equal to a coverage rate threshold value or not; if the evidence coverage rate is greater than or equal to a coverage rate threshold value, retrieving in a standard knowledge base according to the candidate field to obtain a rule fragment; compiling the rule fragments into a constraint set; and if the candidate fields meet all the constraints in the constraint set, generating structured data according to the candidate fields meeting all the constraints in the constraint set. According to the method, the industrial engineering drawing can be accurately converted into the structured data.

The invention belongs to the field of land parcel information extraction technology, and provides an agricultural machinery operation suitability evaluation method comprising the following steps: S1, based on a remote sensing image and a digital elevation model, extracting field parcel geometric features, road features, field obstacle features and terrain parameters, and calculating field parcel indexes; and S2, a multi-dimensional environment correction factor is calculated, the obstacle influence function and the agricultural machinery operation efficiency loss are integrated, a multi-parameter weighted agricultural machinery operation suitability evaluation model is constructed, and an agricultural machinery operation suitability index is calculated. The invention also correspondingly provides an agricultural machinery operation suitability evaluation system. The method not only reflects the direct influence of field obstacles on agricultural machinery operation, but also objectively quantifies the efficiency loss in actual operation through path planning simulation, and provides a scientific basis for precise reconstruction of farmland and intelligent scheduling of agricultural machinery.

The invention discloses a shielding effectiveness monitoring method and system. The method comprises the following steps: modeling a space electromagnetic environment based on multi-sensor fusion and ray tracing; performing joint analysis on the mixed signals acquired by the sensor network in real time, and extracting fingerprint features representing different signal sources; dynamically defining an effective action boundary of the current shielding field based on the extracted fingerprint features and the type identification result, and distinguishing effective signals and boundary-crossing interference in the shielding area; meanwhile, a deep residual network is adopted to identify weak signals under strong interference, and power compensation is estimated; and finally, outputting a multi-dimensional evaluation result of shielding effectiveness monitoring. According to the invention, by using signal fingerprint analysis and intelligent separation, the anti-confusion and anti-interference capabilities are enhanced, and the misjudgment rate is reduced. And under strong background interference, weak shielding test signals can still be effectively identified, the root cause of a monitoring failure phenomenon can be accurately diagnosed, and the accuracy of a shielding effectiveness monitoring result is improved.

The invention discloses a data warehouse table structure generation method and device, a medium and a product, and belongs to the technical field of data warehouses. According to the method, the Excel template configured with multiple pieces of source database table structure information is obtained, table field definition is automatically analyzed, the source data type is intelligently mapped and converted into the target data type according to the source database type, meanwhile, partition key fields are automatically recognized based on the preset strategy, and finally a complete and accurate data definition language (DDL) statement is generated. According to the method, remarkable technical breakthrough and value are achieved in practical application, the table building efficiency can be improved by 90% or above, batch processing of hundreds of tables is supported, and manual errors are reduced by 95% or above; mainstream database type conversion is comprehensively supported, and various data models such as Dores are adapted; the method has the capabilities of automatic type detection and conversion, intelligent key field selection and partitioning and bucket dividing strategy optimization.

The invention relates to the technical field of data processing, in particular to a power generation field technical supervision and evaluation report automatic generation method based on a large model, and the method comprises the steps: carrying out the preprocessing of multi-source data in the power generation field, obtaining a standardized database, and constructing a vector knowledge base; obtaining prompt information input by a user, and retrieving the vector knowledge base through a mixed retrieval mechanism and a knowledge quality verification mechanism to obtain an original retrieval result; performing structured expansion on the original retrieval result to obtain a retrieval enhanced knowledge set; and according to the prompt information, retrieving the enhanced knowledge set and the prompt word engineering system, and driving each agent to generate a technical supervision examination and evaluation report through a preset multi-agent collaboration mechanism. According to the embodiment of the invention, full-process automation of the technical supervision and evaluation report from data processing, knowledge base construction and retrieval enhancement generation to multi-agent cooperation generation of the report is realized, and the generation efficiency and the content quality of production of the technical supervision and evaluation report are remarkably improved.

The invention provides a medical linear accelerator virtual source model optimization method which is characterized by comprising the following steps: constructing a virtual source model and inputting initial configuration parameters; obtaining simulated dose distribution data of a typical irradiation regular field based on the model; vertically irradiating on the standard water model through a medical linear accelerator to obtain corresponding measurement dose distribution data; performing priority evaluation and sorting on part of parameters of the virtual source model, and performing iterative adjustment according to the priority until the simulation and measurement dose distribution difference is within a preset range; and inputting the parameters after iteration is completed into the virtual source model to obtain an optimized model. According to the method, full-process automatic optimization of the virtual source model can be achieved, manual intervention is remarkably reduced, the labor cost is saved, the modeling precision and efficiency are greatly improved, and the method is suitable for an efficient and high-precision source modeling process under a complex structure.

The invention discloses an intelligent question and answer method and system for a large model ship field based on retrieval enhancement generation, and the system comprises a fine tuning data integration module, a model fine tuning module, a model reasoning module and a module integrating various technologies, and aims at achieving the efficient management, intelligent retrieval and precise question and answer of complex documents in the field. According to the system, through organic cooperation of three core modules of document preprocessing, high-precision retrieval and large-model fine-tuning reasoning, the system adopts multiple index structures, and rapid positioning and semantic matching of documents are achieved. Through a large model fine tuning technology, the system adapts a general pre-training model to a ship domain task and endows the ship domain task with a domain specifying ability, and the system can quickly respond to multiple rounds of question and answer requirements of a user and dynamically adjust generated answers to meet the requirements of different contexts. According to the method, the threshold of knowledge acquisition and utilization is reduced for the ship industry, resource optimization configuration and technical progress are promoted, and a wider space is created for future development of the industry.

The invention discloses a park source network load storage cooperative scheduling method and system based on a knowledge graph, and the method comprises the steps: constructing the knowledge graph of a field based on a scheduling rule, and coding the knowledge graph into low-dimensional vector representation; fusing the real-time system data and the priori knowledge vector to generate enhanced state representation; constructing a reinforcement learning decision-making environment; a distributed Actor-Critic framework is adopted, and all agents share environment information but independently optimize a strategy; and realizing collaborative decision-making under global constraints through iterative updating of a strategy gradient algorithm. The system comprises a knowledge graph unit, a vector fusion unit, an environment definition unit and an optimization unit. According to the invention, a cooperative scheduling scheme satisfying the industrial park source-network-load-storage integrated scheduling scene can be provided. The method can be widely applied to the technical field of power system dispatching.

A method and an apparatus for transmitting a physical layer protocol data unit (PPDU) are provided. The method includes: generating a synchronization header field based on a first sequence or a third sequence, where a side lobe of a periodic cross-correlation function of the first sequence and a second sequence is a constant value, the first sequence is a binary sequence consisting of 1 and 0, the second sequence is a binary phase shift keying sequence corresponding to the first sequence, a side lobe of a periodic autocorrelation function of the third sequence is a constant value, and the third sequence is a binary phase shift keying sequence consisting of 1 and −1; and sending the PPDU on a target channel, where the PPDU includes the synchronization header field.

The invention provides a domain reasoning model training method and device, a domain reasoning model using method and device, equipment and a medium, and relates to the artificial intelligence technology. The domain inference model training method comprises the steps that an original domain model is finely adjusted based on a preset inference data set, and an intermediate model with the inference capability is obtained; the preset reasoning data set is generated by reasoning general open source data through a reasoning model and comprises reasoning process data and reasoning conclusion data which conform to the output form of the target domain; based on the current inference data set, performing iterative training on the basis of the original domain model to obtain a domain inference model; wherein in each iteration process, the current reasoning data set is updated based on the field problem data set and a current intermediate model obtained based on the current reasoning data set in the current iteration process; the domain problem data set comprises task feature information of the target domain. By means of the method, the domain model reasoning effect can be effectively improved.

The invention relates to a crystal space group identification and uncertainty quantification method and system, and belongs to the technical field of material crystallography analysis. According to the method, an end-to-end deep learning system is constructed, implicit depth features and explicit physical features based on domain knowledge in an original diffraction pattern are extracted by using two channels, and then dynamic weighted fusion is performed on the features extracted by the two channels by using an adaptive gating fusion mechanism, so that a deep learning algorithm is obtained. And finally, carrying out classification prediction on the fused features and carrying out quantitative evaluation on uncertainty of a prediction result, thereby realizing comprehensive coverage from automatic classification to credibility analysis. Compared with a traditional automatic space group classification model for prediction, the method has the advantages that the classification accuracy, the analysis robustness and the interpretability of an evaluation result on noise-containing real data are remarkably improved, and an intelligent and reliable innovative solution is provided for automatic representation of modern high-throughput materials.

The invention discloses a method for site selection of a linear engineering waste slag site based on engineering geological conditions of a GIS (Geographic Information System), solves the problem that dynamic environmental factors are not effectively evaluated in a traditional site selection method, and realizes real-time tracking and dynamic optimization of site selection of the waste slag site. Through a multi-source data fusion and continuous feedback mechanism, the site selection deviation risk caused by environmental condition changes is reduced, the applicability of a site selection scheme under complex geological and ecological conditions is improved, through standardization processing, data of different sources are seamlessly connected in the spatial dimension and the attribute dimension, and the site selection accuracy is improved. And a reliable data basis is provided for space operations such as subsequent overlay analysis and buffer area calculation. In addition, efficient analysis of the multi-source data in the GIS platform is ensured in the format conversion process, and analysis interruption or result distortion caused by data format errors is avoided.

The invention provides a data processing method and device, equipment, a medium and a computer program product.The data processing method comprises the steps that original cues corresponding to data input by a user are processed through a domain model and an adapter matched with the domain model, and domain cues are obtained; obtaining a target cue word according to the original cue word and the field cue word; and taking the target cue word as an input of a large language model LLM obtained by pre-training to obtain reasoning data corresponding to the user input data. According to the scheme, the final reasoning data can be comprehensively obtained according to the original data and the cue word obtained by the corresponding field reasoning, so that the problem that the reasoning result obtained only based on the field reasoning is inaccurate is avoided, and the reasoning precision is improved; the problem that in the prior art, a data processing scheme for reasoning is poor in reasoning precision is solved.