20results about How to "Eliminate redundancy" patented technology

The invention provides a target identification method and system based on multi-source information fusion, and relates to the technical field of low-altitude target detection. The method comprises the following steps: acquiring target radar track data, interception equipment track data and position area information data; based on target radar track data, extracting a first feature in an RCS form dimension, and extracting a second feature in a motion dimension; based on the target radar track data and the track data of the monitoring equipment, determining a frequency spectrum monitoring correlation factor and regional position information features; performing feature fusion on the first feature, the second feature, the spectrum interception correlation factor and the regional position information feature to obtain a target feature, and identifying a target type; and identifying a target threat level based on the target type and the target radar track data. The method and the device are used in a target identification process based on multi-source information fusion, and the technical problem that the target threat degree cannot be accurately identified in a complex environment in the prior art is solved.

The application discloses a tunnel boring machine space dynamic compensation seismic wave advanced detection method and system, comprising: receiving the reflection echo data of the current space position; compensating the reflection echo data according to the attenuation compensation factor of the current space position, wherein the attenuation compensation factor of the current space position is obtained by a trained neural network, and the prediction process comprises: after curve fitting of the attenuation compensation factor of the reflection echo data of the detected space position, the attenuation compensation factor of the next space position is predicted by using the trained neural network; the compensated reflection echo data is subjected to geological body imaging by using the equal travel time imaging method, and the geological body detection result is obtained according to the amplitude difference on the imaging profile. The space dynamic compensation algorithm of the mechanical wave attenuation is used to adaptively correct the attenuation compensation factor at the next tunneling surface, so as to compensate the received reflection echo data and enhance the identification ability of the reflection echo.

The invention discloses a fluid pulsation accelerator architecture for reinforcement learning of a large language model, and belongs to the technical field of artificial intelligence hardware acceleration. The system aims at solving the problems that when a current universal processor executes an RLHF working load, an instruction driving normal form is not matched with data flow calculation, fixed parallel granularity cannot adapt to a dynamic load, and the efficiency is low due to the fact that a framework does not sense data statistical characteristics. The core of the system is a liquid systolic array calculation fabric capable of being dynamically reconstructed, execution is triggered through data flow, and kernel boundaries and instruction overhead are eliminated. According to the system, a global lookup table subsystem is fused, and nonlinear calculation is optimized through merging query and parallel lookup by using data distribution prior; and a dynamic scheduling unit is configured, and elastic parallelism is realized by adopting a resource allocation algorithm supporting work stealing. According to the method, the throughput and the energy efficiency of RLHF training and a large language model reasoning stage can be remarkably improved, and a new design normal form is provided for a next-generation AI special computing architecture.

This invention discloses a method and system for injecting knowledge into large-scale models for vertical industries, belonging to the field of large-scale model knowledge injection technology. The method involves: collecting professional knowledge data from vertical industries, performing denoising, structured transformation, and terminology standardization; dividing the data into knowledge modules and associating them with vectorized indexes and metadata tags to form a vertical industry knowledge base; parsing the semantic feature vector of the input question using a dynamic routing classifier and routing it to the target knowledge module in the vertical industry knowledge base; employing a hybrid retrieval strategy to recall relevant document sets from the target knowledge module and deduplicate them to generate enhanced retrieval results; based on this, concatenating the enhanced retrieval results with the input question using industry-specific prompt templates; and inputting the LoRA-tuned large-scale model to generate accurate answers conforming to industry standards. This invention solves the problem of insufficient knowledge adaptability of large-scale models in vertical industries, improving the accuracy and applicability of large-scale models in vertical industries.

The invention belongs to the technical field of artificial intelligence, discloses an expert knowledge and data fused intelligent aid decision-making model training method, and aims to solve the problem that expert knowledge and data are difficult to synergistically utilize by an intelligent model, the expert knowledge is converted into a feature and strategy bound decision-making sequence, so that the expert knowledge becomes a data sample; a way is provided for utilizing expert knowledge; a business target state is adopted to evaluate the strategy effect, the expert sample weight and the parameter learning rate are dynamically adjusted according to the model training process, the expert knowledge utilization effect is improved, and meanwhile, new expert samples are dynamically injected in a model evaluation mode; a sample fusion matching mode is adopted to eliminate redundancy of expert samples and data samples, the model training process is accelerated, a sample cluster with a small number of samples is identified through a density clustering mode, the number of similar samples is increased by taking the sample cluster as a starting point, and the robustness of an auxiliary decision-making model is enhanced through continuous improvement of sample richness.

The present application relates to a kind of multi-test bench railway vehicle test data integrated management system, belong to railway vehicle test data management technical field.The data acquisition module of the system is based on adaptive matching mechanism obtains multi-test bench data and completes local pre-processing;Data processing module constructs data correlation rule, generates dynamic cross-test bench data correlation atlas, realizes multi-source data fusion and value-driven storage scheduling by hierarchical fusion strategy;Core management module generates test task scheduling scheme and completes priority adjustment, constructs digital twin mapping and generates whole-process traceability link;Interactive collaboration module generates adaptive data display scheme, pushes abnormal early warning and completes field operation interaction.The present application solves the problems of existing multi-test bench data integration, scattered management and poor traceability, realizes the standardization and intelligent control of whole-process railway vehicle test, and improves the test efficiency and data management reliability.

This invention discloses a topology optimization design method for hot-formed steel automotive main frame welded tubes, comprising: establishing an initial model of the automotive main frame welded tube, the initial model including an installation area, a central load-bearing area, a collision energy absorption area, a weld candidate area, an optimizable area, and a non-design area; performing topology optimization on the optimizable area to obtain material density distribution results; dividing the area into high-density, medium-density, and low-density areas based on the material density distribution results, and determining the collision energy absorption area based on the collision deformation trend; converting different areas into stiffeners, locally reinforced structures, variable cross-section transition structures, cross-section reduction areas, and induced crushing structures, respectively; determining the longitudinal weld position based on stress distribution and collision deformation trend; compensating and correcting the structural parameters according to the hot forming process, and outputting a design scheme for the hot-formed steel automotive main frame welded tube. This invention can improve material utilization, weld reliability, and hot-formability manufacturability.

The application discloses a kind of urban resilience spatial combination mode identification method based on space big data mining, belong to wisdom city and urban space planning technical field.This method obtains the spatial element vector data of research area and carries out grid division, extract the spatial element and its characteristics in each grid unit, quantization topological, direction, distance and so on configuration relationship;Using "element-feature-relation" triple to carry out standardization coding, construct spatial transaction dataset;Based on the improved FP-Growth algorithm, potential characteristic factors are mined, and the output typical spatial combination mode characteristic factor set is output by hierarchical merging, spatial continuity correction and spatial instance verification.The system includes data preprocessing, relationship quantization, coding, pattern mining, post-processing and output module.The application can improve the objectivity, reproducibility and interpretability of identification, and provide quantitative support for typhoon disaster prevention, resilience assessment and spatial planning decision.

The application discloses a kind of synergic network perception method and device, it is related to data communication technical field, method includes: each function branch of network perception collects raw data, respectively pre-processes and generates intermediate data, after data conversion and security processing, it is shared to other function branch;Each function branch generates perception result according to receiving and own intermediate data, and the perception result that can be utilized by other function branch is shared;Meanwhile, operation suggestion is generated according to perception result and is executed, and operation execution result is fed back to itself, and is shared to other function branch that produces influence.The application breaks the barrier of each function branch, so that each function branch maintains relatively independent while being able to cooperate with each other.The technical problem that the current dependence single network function branch realizes leads to not comprehensive consideration factor, the operation after perception is not accurate enough.

This application provides an image classification method, training method, apparatus, and device based on a classification model, comprising: acquiring an image to be processed; inputting the image to be processed into a preset classification model; and acquiring a convolutional input vector from the input layer of the classification model; compressing and extracting features from the convolutional input vector based on a downsampling module in the classification model to obtain a downsampling output vector; extracting features from the downsampling output vector based on a normal module in the classification model to obtain an image feature vector; and classifying the image feature vector based on a fully connected layer in the classification model to obtain the image category of the image to be processed. This application reduces the amount of computational data required for feature extraction from the image to be processed, improves the computational speed of feature extraction, and ensures the comprehensiveness and accuracy of feature extraction.

The invention relates to the technical field of database architecture and management, and provides a database tool plug-in routing management method, which comprises the following steps of: dynamically distributing a globally unique namespace identifier for each functional plug-in; adding a namespace prefix for a front-end routing path of each plug-in according to the namespace identifier to realize centralized registration and life cycle management of routing; all application program interface requests are intercepted in a front-end communication link and a rear-end communication link, namespaces of corresponding plug-ins are automatically associated based on request paths, and standardized request and response processing flows including identity authentication and error processing are implemented; based on an event-driven architecture, an event bus supporting plug-in identifier tracking is established, and an event publishing and subscribing mechanism with context information is provided between plug-ins. According to the invention, routing conflicts can be eliminated, stable operation of the system is guaranteed, standardization of a communication process and centralization of management and control are realized, communication efficiency and consistency are improved, and plug-in development and system maintenance costs are greatly reduced.

The application discloses a glacier change prediction method and system, and belongs to the technical field of glacier monitoring; the application is based on a data-driven glacier mass balance model, considers the feedback mechanism of glacier morphological evolution on glacier mass balance and dynamic process, reduces the uncertainty of glacier change prediction results through iterative processing, and further improves the accuracy of future glacier change prediction. On this basis, the glacier mass balance model used by the application comprises a dimension reduction module, an attention mechanism module and a mapping module; the nonlinear mapping relationship between the glacier meteorological parameter sequence, the glacier morphological parameter and the glacier mass balance data is learned through the joint dimension reduction module and the attention mechanism module, the multiscale characteristics of the time series data can be fully mined, and the error of the glacier mass balance estimation is further reduced.

The invention relates to a method for constructing a knowledge graph in a sensing-communication and calculation-control integrated field. The method comprises the following steps: firstly, obtaining PDF and Word data sources from paper, standard and industry libraries, and analyzing and cleaning to generate a preprocessed data set; constructing a domain ontology layer taking sensing and calculation control integration as a root from top to bottom, and synchronously constructing a terminology library and a synonym dictionary; based on an ontology and term system, high-confidence entities are screened through entity boundary probabilities and term weights, and high-reliability entity pairs are obtained through position / type filtering and relation strength quantification; inputting the data and the preprocessed data into a large model, and extracting an initial triple by means of a cue word template; fusing the multi-source entities through the comprehensive similarity to obtain a standardized triple; and finally, importing a graph database, mapping into nodes / edges, and visually displaying, so that the problems of semantic unification and accurate association of cross-domain knowledge are solved.

This invention relates to the field of robot control technology, specifically a multimodal inspection robot swarm control method. The method includes: collecting heterogeneous environmental data such as laser point cloud sequences and visible light video streams through multiple sensing terminals; performing spatiotemporal alignment and feature-level fusion on these data to construct a unified spatiotemporal perception map containing information such as a semantically defined 3D environmental model; based on this map, using an improved distributed model predictive control algorithm with rolling optimization based on an online learning interactive cost function, generating and executing local control command sequences for each robot; and receiving incremental perception data returned by the robots and asynchronously updating the unified spatiotemporal perception map and the interactive cost function. This method improves the comprehensiveness of swarm environmental perception and control accuracy, and enhances the coordination and dynamic adaptability of robot swarm collaborative operations.

This application discloses a method and system for tracking cigarette product production information, relating to the field of tobacco technology. The method implements unique identification for each level of cigarette product through predefined coding rules; industrial IoT devices continuously collect the timestamps, status identifiers, and location coordinates corresponding to the codes to form a raw data stream; a stream processing engine based on Apache Flink performs real-time grouping, time-series alignment, and state merging on the raw data stream, outputting a structured event sequence; a graph database transforms the event sequence into a multi-dimensional relationship graph of coded nodes and process nodes, constructing a semantically complete full-process topology structure; a dynamic programming algorithm traverses the graph, extracting the complete flow path of any code according to the time sequence to generate a production trajectory chain; finally, a force-directed graph layout algorithm spatially rearranges the nodes and edges in the trajectory chain, achieving a natural balance in node distribution through physical simulation, and outputting an interactive visual graphic.

This invention provides a method and system for predicting multimodal drug interaction events based on compact representation learning. The method includes: acquiring multimodal features of each drug in a drug pair to be predicted, wherein the multimodal features include at least two of biological features, molecular structure features, and knowledge graph features; performing intramodal compact representation learning on each modal feature of each drug to eliminate redundant information within a single modality and generate compact sub-representations for each modality; using the compact sub-representations corresponding to the molecular structure features as anchors, performing cross-modal alignment of the compact sub-representations corresponding to the biological features and knowledge graph features through mutual information minimization constraints to eliminate intermodal redundant information; constructing a multimodal fusion representation for each drug based on the compact sub-representations of each modality; concatenating the multimodal fusion representations of the two drugs in the drug pair and inputting them into a classifier to output the prediction result of the interaction event type of the drug pair.

The utility model discloses a vacuum gauge joint and a vacuum gauge, relates to the technical field of vacuum measurement, and aims to solve the problems that in the prior art, when impurities of a vacuum gauge are filtered, space adaptability is poor due to the fact that a traditional filtering device is used, and convenient development of equipment is not facilitated. The vacuum gauge connector comprises a vacuum gauge connector body, a first hole and a second hole are formed in the two ends of the vacuum gauge connector body respectively, the vacuum gauge connector body is provided with an inner cavity, the first hole and the second hole are communicated through the inner cavity, the first hole, the second hole and the inner cavity form a gas circulation channel, and at least one filtering structure is arranged in the gas circulation channel. The filtering structure is provided with filtering holes, the first hole is used for communicating with a to-be-tested vacuum system to ensure the same vacuum degree with the vacuum system, and the second hole is used for installing a vacuum sensor.

The invention discloses a water turbine thrust pad temperature prediction method based on working condition self-adaption, and relates to the technical field of hydroelectric generating set operation and maintenance monitoring and intelligent diagnose.The method comprises the steps that a working condition boundary fitting model is built based on the corresponding relation between a water head and active power according to historical operation interval data of a set; slicing an original sequence by adopting a rolling window mode, and then automatically extracting derivative features from each window by utilizing a feature derivation tool; constructing training data sets, grouping the training data sets according to different working conditions, and constructing an XGBoost regression model; selecting a corresponding XGBoost prediction model according to different working conditions, and performing anti-standardization processing on the output predicted tile temperature to obtain an actual temperature predicted value; and calculating a residual error between the predicted value and the real thrust pad temperature, and carrying out statistical analysis on the residual error in a set time window. According to the method, the XGBoost regression model is independently trained, so that the prediction robustness, the model training efficiency and the prediction stability under a complex operation condition are improved.

The application belongs to the technical field of aircraft design and cost engineering analysis, and discloses a key dimension identification method for low-cost design of an aircraft based on dimension reduction and clustering, comprising the following steps: constructing a design variable matrix of an aircraft design scheme and a total cost sequence; performing standardization processing on the design variable matrix, and then performing principal component analysis to obtain a feature matrix after dimension reduction; constructing a sparse regression model with the total cost sequence as a response variable by using the feature matrix after dimension reduction; constructing an optimization problem on the basis of the sparse regression model, obtaining a regression coefficient vector by solving the optimization problem, and selecting key cost influence variables on the basis of the regression coefficient vector; extracting a load vector of the key cost influence variables in a principal component space, and clustering the key cost influence variables based on the load vector; and identifying key dimensions according to the categories after clustering. The application is suitable for cost key variable analysis under the conditions of multiple schemes and multiple parameters in the engineering design stage.