44 results about "Three dimensionality" patented technology

The invention discloses an automatic generation method for health science popularization articles, and belongs to the crossing field of computer technology and health science popularization. The method comprises the following steps that S1, health authority information is collected through a multi-source crawler, and an original material pool is obtained through Sension-BERT vectorization duplicate removal and BioBERT medical entity labeling; s2, classifying nine types of health themes by using a RoBERTa fine tuning model; s3, multiple AI model supplementary materials are made into a structured package; s4, predetermining optimal selection questions in combination with AI three-dimensional scoring and editing; s5, calling the multi-dimensional knowledge base to generate a professional knowledge packet; s6, generating an outline and performing three-dimensional five-score system auditing of knowledge point fullness and the like; s7, the LLM expands and writes the text and marks knowledge sources; s8, performing double-stage auditing; s9, anthropomorphic draft moistening; s10, intelligently illustrating the picture; s11, when the matching degree is smaller than a preset threshold value, automatically updating the knowledge base; and S12, integrating and outputting. According to the method, the health science popularization article is generated, the manual participation time is shortened to be within 30 minutes, the medical error rate is reduced to be below 5%, the daily output of a 10-person team is improved by 3-5 times, the annual human cost is reduced by 60% or above, and the large-scale and high-quality science popularization requirements are met.

The invention discloses a water quality monitoring probe fault prediction and life evaluation method and system based on time sequence deep learning, and belongs to the technical field of water quality monitoring. The objective of the invention is to solve the problems of passive maintenance, no performance attenuation quantification, inaccurate service life evaluation and high early warning false alarm of the existing probe. The system obtains and preprocesses multi-dimensional time sequence data including response time, measurement precision and the like, inputs the data to a time sequence feature deep learning module, and extracts depth features by using an LSTM-CNN hybrid network in combination with an attention mechanism; the probe performance attenuation quantification module establishes a baseline according to the initial data, calculates a three-dimensional attenuation index and weights the three-dimensional attenuation index to obtain a comprehensive attenuation index; the Weibull life evaluation module performs modeling according to the data and calculates the residual life, the reliability and the confidence interval; the fault risk scoring module fuses multiple parameters and generates 0-100 score risk scores and four-level early warning and maintenance suggestions; through core algorithm innovation, predictive maintenance is realized, cost is reduced, monitoring continuity is improved, and the method is suitable for full-life-cycle management of the probe.

The invention discloses an anti-unmanned aerial vehicle interference effect evaluation method and system, and belongs to the technical field of anti-unmanned aerial vehicle interference effect evaluation, and the method comprises the following steps: collecting two types of anti-unmanned aerial vehicle scheme cases, carrying out the standardization and classification processing of cases which are good in evaluation but failing in off-site simulation, and building a failure case library; cases which are both good in evaluation and off-site simulation are classified, then the economic benefit range of each type of evaluation method is determined, and a successful case library is established. According to the method, a failure case library containing structured data and a success case library with a clear economic benefit range are established, a to-be-evaluated scheme is evaluated, then three-dimensional data are extracted from the failure case library to construct a multi-dimensional association model, and an evaluation blind area is positioned; and finally, three evaluation methods are generated in combination with a success case library, economic benefits are calculated and sorted, the difference between evaluation and off-site simulation can be effectively reduced, the evaluation accuracy is improved, and off-site simulation failure conditions are reduced.

The invention relates to the technical field of human-computer interaction, in particular to a B2C customer service response system based on sentiment analysis, which comprises the following steps: acquiring an input statement of a user in a B2C customer service dialogue and constructing a semantic sequence, then deconstructing semantic conflicts to judge the context polarity, and correcting a progressive link path to determine the final sentiment deviation of the user. And matching and adjusting the mood of the customer service response corpus according to the emotion deviation, screening the corpus of which the emotion and the mood are matched, and combining to generate a customer service response output result. According to the method, context linkage between entries is realized by constructing a semantic flow, a directional screening mechanism is introduced at a conflict position to unify emotion judgment, final sentence correction is combined in an emotion link to avoid deviation, and dynamic replacement is performed by comparing user expression amplitude in a mood level, so that generated corpora are kept consistent in strength and style; corpora are organized through continuous units, so that connection is natural, and synchronous optimization of dialogue content in three dimensions of emotional polarity, mood coordination and semantic coherence is achieved.

This invention discloses an automatic statistical method for three-dimensional product structure analysis across multiple production lines based on stored procedures. It encapsulates statistical logic through database stored procedures and utilizes GROUP BY and CASE WHEN interval judgments to achieve automatic aggregation by steel type, thickness range, and width range. This method supports unified processing of data from 14 or more production lines and allows for flexible simplification of statistical dimensions. It solves the problems of complexity and inefficiency in traditional multi-production-line product structure analysis, and features high efficiency, stability, and ease of maintenance. This invention has been practically implemented in state-owned steel plants for many years.

The invention relates to the technical field of computers, and discloses a large model reasoning service dynamic connection number limiting method based on service attributes, which comprises the following steps of: 1, carrying out initialization setting on service attributes, resource parameters and a weight model; 2, collecting multi-dimensional data, wherein the collected multi-dimensional data comprises load data, service data and resource data; step 3, carrying out data preprocessing and data fusion on the collected multi-dimensional data to form a three-dimensional fusion data set; step 4, calculating a three-dimensional dynamic weight based on the three-dimensional fusion data set; 5, generating a connection number threshold value according to the three-dimensional dynamic weight, and carrying out connection request distribution and over-limit processing; and step 6, forming closed-loop feedback through real-time monitoring, historical backtracking and algorithm tuning. By utilizing the scheme of the invention, the refined management and control of the connection number in the large model reasoning service are realized, and the resource cost is optimized while the high-priority service quality is ensured.

Surface determination systems, threat detection systems and medical treatment systems are described. According to one aspect, a surface determination system includes processing circuitry configured to access a three-dimensional complex-valued image volume of a target, define image locations of the image volume using first and second dimensions of the image volume, for each of the first locations, identify voxels along a third dimension of the image volume that correspond to the respective first location, for each of the first locations, select one voxel that corresponds to the first location as having an increased amplitude compared with other voxels that correspond to the first location, and for each selected voxel, use phase information to identify a second location in the third dimension for the selected voxel that corresponds to a location of a surface of the target and that is different than a third location of the respective voxel.

This application discloses a method, apparatus, device, and medium for item recommendation across multiple user groups, scenarios, and tasks, relating to the field of artificial intelligence technology. The method involves acquiring user-side information, item-side information, and corresponding scene environment information in different scenarios; calculating weight dimensions, shared feature vectors, and environment dimensions to obtain a first dimension, a dense vector, and a second dimension; performing attention mechanism feature cross-calculation to obtain a third dimension; calculating item recommendation scores using the first dimension, dense vector, second dimension, and third dimension to obtain an item recommendation value; determining the target item recommendation value; identifying the target item information corresponding to the item recommendation value; and recommending the target item information to the user. This application can construct a three-dimensional perspective model encompassing multiple scenarios, user groups, and tasks, thereby improving the ability to represent differentiated user groups, meeting diverse user needs, and achieving comprehensive item recommendation across multiple user groups, scenarios, and tasks.

The invention discloses a three-dimensional space material grabbing method and device and a vehicle assembling system, and relates to the technical field of process control. Comprising the steps of determining two-dimensional coordinates of a material according to an image of the material; the sensing arm extends to the material from the first position until the sensing arm makes contact with the material so as to obtain a second position when the sensing arm makes contact with the material; according to the first position and the second position, coordinates of the material in a third dimension perpendicular to a two-dimensional plane corresponding to the two-dimensional coordinates are determined; and the materials are grabbed according to the two-dimensional coordinates of the materials and the coordinates of the materials in the third dimension. According to the technical scheme, compared with an existing 3D visual recognition technical scheme, the same grabbing function can be achieved, and the equipment cost is greatly reduced. Through measurement and calculation, the 2D vision cost of photographing to obtain images is reduced by one order of magnitude compared with the 3D vision cost, so that the method has a wide application prospect.

The invention provides a vehicle end picture dimension reduction method, device and system based on principal component analysis, and a medium. The method comprises the following steps: acquiring an image in an RGB format acquired by an intelligent vehicle camera; the image in the RGB format is converted into a first three-dimensional array, the first dimension of the first three-dimensional array is the height of the image, the second dimension of the first three-dimensional array is the width of the image, and the third dimension of the first three-dimensional array is an R color channel, a G color channel and a B color channel; separating pixel data of R, G and B color channels in the first three-dimensional array to obtain R, G and B pixel matrixes; carrying out dimensionality reduction on the R, G and B pixel matrixes by adopting a principal component analysis method to obtain R, G and B pixel matrixes after dimensionality reduction; and combining the R, G and B pixel matrixes after dimension reduction into a second three-dimensional array, and generating a dimension reduction image based on the second three-dimensional array. According to the invention, dimension reduction based on the principal component analysis algorithm can be independently carried out on each color channel in the RGB image, cross-channel interference is avoided, and independent optimization of color information is ensured.

The invention relates to a method for testing a fracture track of a metal material under a multi-field coupling effect. The method comprises the steps of sample preparation, high-temperature speckle preparation, sample loading, strain field acquisition, fracture morphology observation, data processing and fracture model establishment. By constructing a new fracture test method under the action of multi-field coupling, the problems of lack of physical field coupling factors, non-uniform test systems, limitation of fracture models and the like at present are effectively solved. A coupling loading test system capable of synchronously applying complex stress, high strain rate and transient high temperature is established, and real simulation of the multi-physical field coupling effect under the actual disaster working condition is achieved; a metal material fracture test system with universality and comparability is formed; key data support is provided for building a fracture model comprehensively considering the stress three-dimensional degree, the strain rate effect and temperature softening, the accuracy and reliability of fracture track prediction of the steel structure under the extreme disaster are remarkably improved, and an important basis is provided for engineering disaster resistance design and safety evaluation.

The invention provides a personnel positioning method and device, electronic equipment, a storage medium and a program product, relates to the technical field of data processing, combines the machine room characteristics of a data center, constructs a positioning tracking model, promotes a three-dimensional model to a three-dimensional space, realizes three-dimensional acceleration triple consistent distribution, and is beneficial to solving the blind spot problem in UWB positioning. Through the motion data and the acceleration at the current moment, the motion data at the next moment is calculated, and the accurate estimation of the motion state of the person at the next moment is realized. According to the invention, through positioning the tracking model, calculating the acceleration, calculating the motion data of the personnel at the next moment according to the motion data and the acceleration at the current moment, and correcting the motion data at the next moment, accurate positioning of the personnel in the machine room is realized.

The invention discloses a transmission conductor galloping displacement abnormity monitoring method based on Beidou RTK, relates to the technical field of conductor displacement monitoring, solves the technical problems of low reliability of galloping parameter calculation and abnormity judgment and lack of continuous over-limit trend prediction capability, and provides a transmission conductor galloping displacement abnormity monitoring method based on Beidou RTK by realizing accurate coordinate system conversion and eliminating projection deformation influence. Two form reconstruction methods are provided, flexible selection can be performed according to requirements, meanwhile, environmental parameters such as real-time icing and wind speed are fused into a three-dimensional kinetic equation, the goodness of fit between a model and an actual galloping state is improved, an LSTM-ARIMA mixed time sequence model is constructed for a continuous overrun scene, and the abnormal trend is predicted in combination with meteorological data in the future 72 hours, so that the prediction accuracy is improved. And meanwhile, a three-dimensional composite risk index is introduced, three types of trends of slow remission, stable maintenance and rapid deterioration can be pre-judged in advance, and operation and maintenance personnel can deploy resources in advance according to the trends, so that the conversion from passive alarm to active prevention and control is realized, and the fault response time is shortened.

The invention discloses an urban and rural fusion development evaluation method based on a combined weighting and coupling coordination model, and relates to the technical field of data processing. The method comprises the following steps: S1, constructing an urban and rural fusion index evaluation system; s2, three-level combination weighting is carried out; s3, constructing a space-time two-dimensional improved coupling coordination model; step S4, performing evaluation based on ESDA and an obstacle degree model; and S5, multi-dimensional verification is carried out. According to the method, a high-quality data basis is provided for dynamic empowerment through accurate indexes screened in a three-dimensional mode, a scientific basis is provided for space-time coupling coordination calculation through reasonable weights of the dynamic empowerment, reliability is guaranteed through multi-scene verification of accurate results of space-time optimization, and finally, an intelligent evaluation report and a policy suggestion library are output. The method directly serves urban and rural fusion development decision, improves the overall evaluation efficiency, and improves the conversion efficiency from the evaluation result to the policy decision.

To achieve an automatic inspection of a biological tissue capable of highly accurate determination.SOLUTION: In the inspection image Tim, a feature amount is associated with a combination of the first coordinates, the second coordinates, and the specific physical quantity. The first coordinate and the second coordinate specify a position on the focal plane when the biological tissue is imaged. The specific physical quantity includes a physical quantity related to light from a substance of a biological tissue present at a position specified by the first coordinates and the second coordinates. The feature quantity includes the intensity of light having a specific physical quantity. The inference model Md extracts the information on the abnormality from the inspection image Tim by performing a three dimensional convolution operation on the feature amount with the first coordinate, the second coordinate, and the specific physical quantity as a first dimension, a second dimension, and a third dimension, respectively, on the inspection image.SELECTED DRAWING: Figure 6

This invention relates to a visualization method and application system based on multidimensional state space, within the field of visualization processing. It acquires and standardizes multidimensional signals and historical data from detection points. In the first visualization region, the radius of concentric circles in a Cartesian coordinate system is dynamically determined based on the historical data distribution density. The first and second dimensions are mapped to coordinates to obtain data points. Directional features are generated using quadrant positions, and deviation degree and level labels are generated using Euclidean distance. In the second visualization region, the third dimension is mapped using a bar chart array. The output is a comprehensive interface including dual regions and tables, presenting the evolutionary state using directional features. Finally, by using real-time statistical frequency analysis, when a structural drift of the expected benchmark occurs, the radius or level is corrected. This invention improves the accuracy and adaptability of monitoring.

The application relates to the technical field of blast furnace ironmaking control, in particular to a metallurgical electric appliance data-oriented query analysis method and system. Silicon content in molten iron, slag byproduct alkalinity multiple and coal gas conveying capacity of key nodes are collected in real time through sensors. When any parameter exceeds a preset threshold value, it is determined that there is high-emission tail gas abnormality. Based on real-time data and historical records, corresponding device node sets are determined respectively. Sample data sets are constructed by using monitoring records in the high-emission abnormal state. A similarity calculation model is trained to convergence, and a deep learning model is generated. The three device node sets are input into the model, and potential leakage nodes are output. Visual warning information of target nodes is generated. The application integrates three-dimensional data, improves the high-emission determination accuracy, realizes accurate leakage point tracing by combining the device node space topological relationship and the graph neural network, identifies CO2 high-emission risk by the alkalinity change duration, and establishes a hierarchical warning system.

The application belongs to the technical field of electric power storage, and discloses a power warehouse operation simulation and bottleneck diagnosis system based on digital twinning. A multi-source perception architecture is constructed through an acquisition and adaptation module, covering full-factor data such as equipment operation and personnel operation. Combined with a five-dimension fusion model of a twinning modeling module and a laser scanning modeling technology, the physical space and operation logic of the power warehouse are accurately restored. A simulation and deduction module adopts a fusion simulation algorithm and a double-scene template library to realize linkage simulation of the whole process and control instructions of warehousing, storage, picking, and delivery, can simulate the operation state under different scenes, and avoid efficiency problems such as equipment waiting and path redundancy in advance. A bottleneck diagnosis module constructs a three-dimension evaluation system of operation state, control parameters, and equipment performance, integrates real-time working condition data and simulation analysis results, scientifically distributes index weights through an analytic hierarchy process, and realizes quantitative division of bottleneck grades.

The embodiment of the invention provides an image semantic alignment re-recognition method and device, equipment and a storage medium, and relates to the technical field of image processing. The method comprises the following steps: obtaining a global semantic matching degree according to an identification query feature and an identification candidate feature, obtaining a target semantic matching degree according to an identification mask query feature and an identification mask candidate feature, and obtaining a matching pair according to an alignment mask query feature and an alignment mask candidate feature; and obtaining an alignment semantic matching degree of the identification mask query feature and the identification mask candidate feature based on the matching pair, and obtaining a re-identification result of the query image and the candidate image according to the global semantic matching degree, the target semantic matching degree and the alignment semantic matching degree. The unsupervised alignment capability of the hidden space diffusion model is utilized to reduce the labeling cost, the accuracy of feature association is enhanced through global, target and alignment three-dimensional semantic matching, and finally dual optimization of the labeling cost and the recognition accuracy in a semantic re-recognition task is realized.

The present invention provides an integrated hyperspectral water quality analysis method, which belongs to the field of hyperspectral water quality analysis. First, data preprocessing is conducted by water quality data collection and water quality image collection in early stage; second, three dimensionality reduction methods are adopted to conduct dimensionality reduction processing, and fused dimensionality reduction is conducted by parameter trade-off selection; third, machine learning algorithms are adopted to train and test hyperspectral water quality inversion models on spectral data after dimensionality reduction; finally, the hyperspectral water quality inversion models are selected and optimized. The present invention adopts an innovative fusion strategy in the aspect of data dimensionality reduction processing, which can achieve a better data dimensionality reduction effect, effectively remove noise and redundant information, and provide a more accurate and reliable data basis.

The invention discloses a pixel-level identification and objectification semi-automatic classification integration method for frozen and thawed slope deposits in a cold and cold mountainous area, and relates to the technical field of remote sensing and geographic information systems. Under the background of complex terrains and strong seasons of the cold and cold mountainous area, freeze-thaw slope deposits are stably and accurately extracted on a long time sequence scale by using multi-source remote sensing data, a pixel-level result is converted into a plaque object available for engineering, and the plaque object is accurately extracted under the condition of lack of year-by-year complete environment covariants. And realizing three-dimensional interpretable semi-automatic classification of formation mechanism-spatial distribution-substance composition, and finally outputting a reproducible and updatable space-time evolution database and a graph product.

The invention relates to a cable fault detection method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: acquiring historical fault data of a cable, and preprocessing the historical fault data to obtain processed historical fault data; determining a time sequence feature, an artificial time domain feature and a two-dimensional image feature of the cable based on the processed historical fault data, and performing dimensionality reduction to obtain three dimensionality reduction depth features; fusing the dimension reduction depth features to obtain multi-view fusion features; constructing a training set for fault detection; inputting the training set into a preset detection model for training, and obtaining a trained preset detection model under the condition that the detection precision of the preset detection model reaches a preset degree; and obtaining target detection data of the cable, and inputting the target detection data into the trained preset detection model to obtain a fault detection result of the cable. By adopting the method, the accuracy of fault detection can be improved.

The invention discloses a complex layout file keyword extraction method and apparatus, and an electronic device. The method comprises the steps of obtaining a complex layout source file; after a complex layout source file is obtained, a redundancy check text with space coordinates is obtained through multi-tool parallel analysis, the source file is converted into a picture with the original layout reserved and serves as a reference coordinate system, and a fusion text set is generated through space coordinate conflict resolution; after images and texts are accurately aligned, inputting the images and texts into a multi-modal large model, completing layout grading, image and text cross validation and content reasoning by means of system cues, and outputting key fields with layout anchor points; and carrying out deduplication normalization, standardized formatting and three-dimensional cross check, carrying out directional re-reasoning on abnormal fields, and finally generating a standardized structured text. According to the method, through multi-tool redundancy analysis, image-text deep fusion and closed-loop directional error correction, the keyword extraction accuracy is greatly improved, the method is adaptive to a multi-format mixed complex layout scene, and the universality and the error correction efficiency are remarkably improved.

The invention is suitable for the technical field of collection training, and provides a self-adaptive collection learning path planning method and system, and the method comprises the steps: constructing a three-dimensional procurement capability model containing compliance operation, a negotiation strategy and risk management and control, decomposing the procurement capability model into capability units, and setting a quantitative index; collecting operation, decision and assessment data in the simulation scene in real time, and generating a dynamic capability assessment matrix; constructing a three-dimensional space model based on the matrix, calculating the association strength and the lifting priority of the power unit, and outputting a personalized learning path; recombining the knowledge module according to the evaluation value change of the capability unit; dynamically adjusting model parameters in combination with policies and industries and re-planning a path; a cross-dimension learning path is generated when the plurality of capability units reach the standard. According to the method, dynamic evaluation, personalized path adjustment and cross-dimension capability integration can be realized, the problems of one-sided evaluation, fixed path and insufficient adaptability in the prior art are solved, and the accuracy and efficiency of auditory collection training are improved.

The invention specifically relates to a metal oxide arrester leakage current on-line monitoring system and method, and relates to the technical field of power equipment monitoring, and the system comprises a multi-physical field sensing array module; a synchronization and data aggregation gateway module; the central analysis server is connected with the fusion module; and a multi-dimensional state evaluation and decision module. According to the invention, through the collaborative design of the electrical, acoustic and thermal three-dimensional sensing units, the problems of early defect misjudgment and interference signal misjudgment caused by the fact that traditional MOA monitoring only depends on power frequency parameters are solved; nanosecond-level partial discharge pulses are captured in the electrical dimension; the acoustic dimension realizes three-dimensional positioning of defects; the thermal dimension is combined with an environment compensation algorithm, so that tiny temperature rise can be identified; and in cooperation with triple cross validation of a space-time correlation algorithm, internal defects and external interference are effectively distinguished.

This invention relates to the field of railway assembly technology, and discloses a method and system for autonomous assembly and positioning of railway fasteners based on multimodal visual fusion. The method includes: aligning the image and depth information of the railway fasteners; constructing three-dimensional spatial coordinate information and extracting the fastener position information using image pixel coordinates as a two-dimensional reference and depth distance as a third dimension; performing inverse kinematics on the robot arm based on the three-dimensional coordinates and posture data of the position information and smoothing constraints to obtain path planning parameters and encode them into control commands; continuously monitoring the assembly status and identifying the current position deviation parameters; establishing a mapping rule between deviation and pose adjustment based on kinematic analysis and test data, mapping the current deviation to the target adjustment amount; updating the path planning parameters according to the target pose adjustment amount, and encoding the updated parameters into correction control commands, executing the correction control commands to complete the assembly of the fasteners; this invention can improve the efficiency of railway fastener assembly and positioning.

The application discloses a kind of carbon metering algorithm verification methods for multi-source heterogeneous data fusion, including collecting the multi-source heterogeneous data of preset computing power center, based on physical and statistical mixed construction high-fidelity simulation test environment, generate test case data stream, the test case data stream is pretreated;Dynamic operating condition generator generates dynamic operating condition signal based on physical model, based on space-time scheduling strategy dynamic injection preset anomaly on dynamic operating condition signal to the high-fidelity simulation test environment;The dynamic operating condition signal includes load fluctuation, fuel switching transient, start-stop process;Theoretical carbon emission calculation model based on material balance and chemical metrology obtains theoretical reference value, three-dimensional consistency analysis is carried out to the output value of measured carbon metering algorithm and the theoretical reference value, and comparison data is obtained;The three-dimensional is internal logic, abnormal response and trend rationality;According to the comparison data, construct multi-source heterogeneous fusion carbon metering verification model, input the data to be verified into the multi-source heterogeneous fusion carbon metering verification model, and output verification result.

The invention discloses an aviation dynamic credit evaluation method and device and computer equipment. The method comprises the following steps: selecting three types of core data sources, namely airline historical data, real-time flight data and macropolicy data, and comprehensively covering historical operation-real-time operation-external environment dimensions; a TCN-RiskModel is constructed based on a time sequence convolutional network TCN, and a federal transfer learning architecture and a hyper-parameter optimization framework are adopted to train and optimize the model; setting an operation stability-airline potential-policy risk three-dimensional basic weight, introducing a dynamic adjustment mechanism, and obtaining a dynamic credit score through weighted summation of a three-dimensional score * a corresponding dynamic weight, omega1 + omega2 + omega3 = 1, omega1 is an operation stability weight, omega2 is an airline potential weight, and omega3 is a policy risk weight. In this way, a credible quantitative basis can be provided for financing matching, and the performance of the method is obviously superior to that of a traditional model.

The invention relates to the technical field of knowledge base evaluation, and discloses an RAG knowledge base evaluation and self-optimization method based on three-dimensional dynamic calibration. According to the method, a retrieval recall rate-answer correctness-knowledge base dependency degree three-dimensional evaluation system is constructed, and a dynamic weight distribution algorithm and an improved vector similarity calculation model are combined, so that quantitative evaluation and automatic optimization of a knowledge base effect are realized. The system comprises an index system configuration module which supports scenarized weight dynamic allocation; the evaluation data set automatic generation module is used for realizing large-scale benchmark problem construction; the dynamic evaluation execution module is used for calculating a three-dimensional index and outputting a comprehensive score; the self-optimization execution module is used for executing operations such as index optimization, parameter adjustment and knowledge supplementation for the abnormal scene; and the result visualization module is used for providing an evaluation report and optimizing trajectory tracking. The technical problems that traditional RAG knowledge base evaluation is subjective, cannot be reproduced and lacks a closed-loop optimization mechanism are solved, the evaluation efficiency is improved by 90% or above compared with a traditional method, the optimization effect is improved by 30%, and the method is suitable for various RAG knowledge base scenes such as enterprise research and development and government affair services.

The invention discloses a sports video key scene time node positioning method, which relates to the field of video processing, and comprises the following steps: obtaining sports data, and setting a sports management space; obtaining sports video data; dividing the historical sports video data in the sports management space according to the sports data, and setting a key sports data set; constructing a three-layer branch processing model based on the key motion data set, and performing importance evaluation processing from three dimensions; performing resource allocation processing according to the importance evaluation result to obtain a resource allocation result; performing three-dimensional division matching on a corresponding image area in the sports video data according to the resource allocation result to obtain a hierarchical division result of three dimensions; carrying out collaborative integration according to a hierarchical division result and a three-layer branch processing model, and obtaining key scene time node positioning information of the corresponding sports video data; the sports video processing efficiency is improved to a certain extent.