Patents
Literature
Patsnap Eureka AI that helps you search prior art, draft patents, and assess FTO risks, powered by patent and scientific literature data.

366results about How to "Efficient mining" patented technology

Optimal configuration method and device based on virtual power plant resources and computer equipment

The invention relates to an optimal configuration method, device and equipment based on virtual power plant resources. The method comprises the following steps: acquiring resource parameters of each virtual power plant resource, historical load data associated with each virtual power plant resource, external load data and user behavior data, and constructing a resource characteristic model for each virtual power plant resource according to the resource parameters, the historical load data, the external load data and the user behavior data, according to historical load data, external load data and user behavior data, load prediction is carried out according to the historical load data, the external load data and the user behavior data to obtain a load prediction result, adjustable capability indexes are determined according to the load prediction result and the resource characteristic model, and weighted fusion is carried out according to the adjustable capability indexes and the index weights of the adjustable capability indexes to obtain a resource evaluation result. And performing optimal configuration on each virtual power plant resource according to a preset configuration demand and the resource evaluation result corresponding to each virtual power plant resource. The method can improve the resource utilization rate and reduce the energy consumption.
Owner:CHINA SOUTHERN POWER GRID BIG DATA SERVICE CO LTD

Deep learning-based stroke prognosis prediction method and system

The invention discloses a deep learning-based stroke prognosis prediction method and system. The method comprises the steps of obtaining a head magnetic resonance medical image data set of a stroke patient before treatment and a corresponding mRS score of stroke prognosis; segmenting a core infarction focus image from the obtained magnetic resonance medical image by adopting a DWI sequence; establishing a six-score cSVD load scoring standard, scoring the core infarction focus image, and constructing a model data set; 3, constructing a prediction model for predicting the stroke prognosis mRS score, dividing the model data set in the step 3 into a training set, a verification set and a test set, training the prediction model by adopting the training set, verifying the prediction model on the verification set, testing by adopting the test set, screening out the prediction model of which the precision meets the requirement, and determining the stroke prognosis mRS score. And predicting the mRS score of the stroke prognosis of the to-be-evaluated patient by adopting the screened prediction model. According to the method, the prediction capability of the image features obtained by MRI (Magnetic Resonance Imaging) flat scanning of the LAA-AIS patient on prognosis is fully excavated, and a new reference direction is provided for prognosis evaluation of the LAA-AIS patient.
Owner:YANCHENG HOSPITAL OF TRADITIONAL CHINESE MEDICINE

User experience design method based on natural language processing technology

The invention discloses a user experience design method based on a natural language processing technology, relates to the technical field of natural language processing technologies and user experience design, and adopts the natural language processing technology to intelligently collect and analyze user feedback information. The method is constructed on the basis of advanced natural language understanding and sentiment analysis technologies, and an intelligent feedback management system realizes efficient mining and deep insight of mass text data and covers deep information extraction, sentiment recognition and quantification and automatic report generation; according to the application of the natural language processing technology in the field of user feedback collection and analysis, not only is the crossing from passive receiving and manual processing to active detection and intelligent analysis realized, but also the user experience is improved through seamless voice interaction and accurate path guidance. The user experience and the response speed of the enterprise to the user demand are greatly improved, and the intelligent upgrading of the feedback system is really realized.
Owner:NANJING FIBERHOME STARRYSKY CO LTD

Transform-based automobile part defect detection method and system

The invention discloses an automobile part defect detection method and system based on Transform, and the method comprises the steps: taking Vision Transform as an encoder, carrying out the spatial feature and long-distance dependence feature extraction of a two-dimensional automobile part image, and obtaining four different levels of automobile part coding feature maps; sequentially inputting the fourth-level automobile part coding feature map into a Sobel filter and a saliency edge guiding module, and performing feature enhancement and extraction on boundary information of a defect region in the image to obtain a high-level saliency coding map; performing texture feature extraction and enhancement on the automobile part coding graphs of different levels by using a texture feature enhancement method to generate texture feature graphs of different levels; and performing foreground and background mark segmentation on the high-level saliency coding image and the texture feature images of different levels by using a context enhanced decoding module to obtain an automobile part defect positioning result image. According to the invention, the precision and generalization of automobile part image defect area detection can be improved.
Owner:JILIN DONGGUANG RUIBAO LAMP

Hotspot mining system and method based on Internet financial information, and medium

The invention discloses a hotspot mining system and method based on Internet finance and economics information and a medium, and relates to the field of Internet finance and economics hotspot mining systems, and the system comprises an information collection module which collects finance and economics information in a multi-source manner, determines a collection period according to source characteristics, and carries out the screening and storage; the text preprocessing module is used for cleaning and analyzing texts, standardizing terms and carrying out classification labeling; the hotspot evaluation index calculation module is used for calculating indexes such as classification weighting, TF-IDF and the like; the hotspot mining and analyzing module is used for setting threshold mining hotspots, analyzing emotion and the like; and the visual display module is used for visually displaying the hot spots and supporting interaction and pushing. According to the invention, financial information is integrated, and hotspots are mined accurately. The information acquisition efficiency is improved, and decision-making investment and research are assisted. Information spreading is promoted, and financial field development is promoted.
Owner:HEXUN INFORMATION TECHNOLOGY CO LTD

Semi-supervised speech imagination intention decoding method based on electroencephalogram collaborative clustering

The invention discloses a semi-supervised speech imagination intention decoding method based on electroencephalogram collaborative clustering. The method comprises the following steps: step 1, collecting electroencephalogram data of speech imagination; 2, performing feature extraction on the preprocessed electroencephalogram data; 3, performing feature clustering and sample clustering on the original data matrix to obtain a clustering indication matrix of features, a clustering indication matrix of samples and a coefficient matrix; step 4, completing optimization solution of the clustering indication matrix of the features, the clustering indication matrix of the samples and the coefficient matrix under a semi-supervised framework; and 5, displaying a decoding result obtained by the semi-supervised learning model based on the electroencephalogram sample and feature collaborative clustering through a visualization means. The method can fully explore and utilize a bidirectional complex interaction mode of features and samples and label information of speech imagination in speech imagination intention decoding research and system based on electroencephalogram signals.
Owner:HANGZHOU DIANZI UNIV

River channel desilting device for water conservancy project

The invention discloses a river channel desilting device for hydraulic engineering, and belongs to the field of river channel desilting devices.The desilting device comprises a floating plate, a collecting box arranged on the floating plate and a desilting structure used for removing river channel silt; the desilting structure comprises two supporting plates fixedly arranged at the end, close to the collecting box, of the floating plate and a conveying belt movably installed between the two supporting plates. The bucket is fixedly arranged on the conveying belt and used for cleaning sludge in the river channel; and the hydraulic oil cylinder is hinged to the floating plate, connected with the conveying belt and used for adjusting the inclination angle of the conveying belt. Operation is directly carried out on the floating plate, water does not need to be drained in advance, the operation preparation time is greatly shortened, meanwhile, continuous and efficient sludge excavation and conveying are achieved through the design that the conveying belt is matched with the excavator bucket, and the dredging speed is remarkably increased. The problem that the trolley is prone to clamping stagnation in sludge is solved, an operator only needs to control equipment on the floating plate, and the labor intensity and the operation risk are reduced.
Owner:宋悦豪

Light guide plate surface defect detection method based on GCC-YOLOv8 network model

The invention provides a light guide plate surface defect detection method based on a GCC-YOLOv8 network model. The method comprises the following steps: acquiring an image of a light guide plate to be detected; the image is input into a GCC-YOLOv8 network model for defect detection; outputting surface defect information of the light guide plate based on a detection result; wherein the GCC-YOLOv8 network model is constructed based on a YOLOv8 network model, and comprises a backbone network unit, a neck network unit and a detection head network unit with three parallel detection branches, which are arranged in sequence; two GKA-C2f1 modules are arranged at the tail end of the backbone network unit; the neck network unit comprises an FPN network unit and a PAN network unit; the FPN network unit comprises two up-sampling modules, two C2f2 modules, two Concat modules, an SPPF module and a CPCA module; the PAN network unit comprises two CBS modules, two Concat modules, two C2f2 modules, three CPCA modules, a CFC module and an SFC module. The light guide plate surface defect detection method has good light guide plate surface defect detection performance and robustness, and also has good detection speed.
Owner:SI CHUAN JIN CHENG RUI ZHI HU LIAN WANG KE JI YOU XIAN GONG SI

Temperature monitoring method and system based on dual-light fusion image

The invention provides a temperature monitoring method based on a dual-light fusion image. The method comprises the following steps: S11, synchronously acquiring a visible light image and an infrared thermal image inside and outside a switch cabinet in an intelligent station building; s12, performing local adaptive convolution on the visible light image and the infrared thermal image, extracting local shallow layer features, and performing heterogeneous association to obtain heterogeneous association degree; s13, based on an anisotropic diffusion mechanism, decomposing the infrared thermal image interfered by sunlight in the switch cabinet to obtain a detail layer image and a base layer image, and performing enhanced fusion on texture features of the detail layer image and the base layer image through the distribution gradient and heterogeneous association degree of textures in the visible light image to obtain an enhanced infrared thermal image; and S14, inputting the enhanced infrared thermal image into a pre-trained temperature detection model for identification, obtaining a temperature abnormal region, and carrying out early warning on a temperature characteristic value. According to the invention, the heterogeneous correlation enhancement of the dual-light fusion image is realized, so that the reliability of temperature monitoring is improved.
Owner:YUNNAN ELECTRIC POWER TESTING & RES INST (GRP) CO LTD +1

Deep heat storage large-scale fracturing construction scheme design method based on weak structure

The invention discloses a deep heat storage large-scale fracturing construction scheme design method based on a weak structure, and relates to the technical field of high-temperature geothermal resource exploration and development, and the method comprises the following steps: depicting and characterizing the weak structure of a deep heat storage large-scale fracturing stratum to obtain a comprehensive characterization large model; based on the comprehensive characterization large model, delineation and fracturing performance evaluation are conducted on a weak structural zone of the deep heat storage scale fractured stratum, and a delineated area is obtained; on the basis of the delineated area, the farthest disturbance range and distance of fracturing transformation of the weak structure of the deep heat storage scale fractured stratum are predicted; and determining a well spacing, designing a well location, deploying a well pattern and optimizing a fracturing layer section based on the fracturing property and the predicted farthest disturbance range and distance of thermal storage fracturing transformation, and completing the design of a deep thermal storage large-scale fracturing construction scheme. According to the method, through construction and analysis of the comprehensive characterization large model, the stratum weak structural zone suitable for fracturing can be accurately delineated, the farthest disturbance range and distance of fracturing transformation are predicted, and efficient mining is achieved.
Owner:CHINA UNIV OF MINING & TECH

Preparation method of coal gangue composite grouting material

The invention discloses a preparation method of a coal gangue composite grouting material, which comprises the following steps: step 1, preparing coal gangue aggregate: drying crushed coal gangue, and screening to obtain the coal gangue aggregate consisting of fine aggregate and coarse aggregate; step 2, preparing a coal gangue gel material: uniformly mixing the crushed coal gangue with graphite, performing microwave activation, adding a chemical activator, mixing, and grinding to obtain the coal gangue gel material; and mixing the coal gangue gel material and the coal gangue aggregate, adding water, and uniformly stirring. According to the method, the technological process is simple and easy to implement, efficient utilization of mine solid waste is achieved, the mining cost is greatly reduced, the grouting filling material with the excellent mechanical property is successfully obtained, and progress and development of the coal gangue based grouting technology are powerfully promoted.
Owner:XIAN RES INST OF CHINA COAL TECH & ENG GRP CORP

Ship trajectory prediction and anomaly detection method, device and equipment based on deep cross LSTM network, and storage medium

The invention provides a ship trajectory prediction and anomaly detection method, device and equipment based on a deep cross LSTM network, and a storage medium, and the method comprises the steps: S1, preprocessing AIS data, and extracting ship navigation features; step S2, constructing a deep cross LSTM network model based on the attention mechanism and a coder-decoder, which comprises the following steps: in a coder part, converting input multivariate time series data into high-dimensional vector representation, and in combination with the multi-head attention mechanism and an LSTM network, capturing a dependency relationship and periodic change among the time series data, and constructing a deep cross LSTM network model based on the multi-head attention mechanism and the LSTM network; the learning of the network model on the data relationship is enhanced and the calculation efficiency is improved; and S3, training the deep cross LSTM network model, and predicting ship navigation features by using the trained deep cross LSTM network model. The network model is applied to two different prediction tasks, namely a joint prediction task and a historical data prediction task.
Owner:AEROSPACE INFORMATION RES INST CAS

Enhanced urban geothermal exploitation system and method based on nanofluid

The invention relates to an enhanced urban terrestrial heat exploitation system and method based on nanofluid. The system comprises a heat collection pipeline, a heat exchange assembly and a nanofluid adjusting assembly. The nanofluid adjusting assembly is connected into the heat collecting pipeline. The nanofluid adjusting assembly comprises a storage pool, a control unit, a concentration detection sensor, a temperature sensor and a concentration adjusting unit, the concentration detection sensor, the temperature sensor and the concentration adjusting unit are connected with the control unit, the storage pool communicates with the heat collecting pipeline, the concentration detection sensor is used for detecting the concentration of nanofluid in the storage pool, and the temperature sensor is used for detecting the temperature of the nanofluid. And the concentration adjusting unit is used for adjusting the concentration of the nanofluid in the storage tank. Compared with the prior art, the method has the advantages that the fluid concentration in the whole circulation pipeline is adjusted in real time according to the appropriate heat transfer concentration of the hybrid nanofluid at all temperatures, then the heat exchange performance of the nanofluid is effectively guaranteed, and the heat exchange efficiency of the geothermal circulation system is improved.
Owner:TONGJI UNIV +1

Intelligent cattle herd health analysis and prevention and control system

The invention relates to the technical field of intelligent monitoring, in particular to an intelligent cattle health analysis, prevention and control system, which comprises a behavior data acquisition module, a data processing module, a data processing module, a data processing module, a data processing module and a data processing module, wherein the behavior data acquisition module is based on sensors at a plurality of positions in a pasture and is used for collecting position movement data of cattle, recording time information, linking the data with time labels and storing the data in a time sequence format; and obtaining behavior time sequence data. According to the method, the real-time monitoring capability of health management of the cattle herds is improved through a comprehensive data acquisition and analysis technology, a behavior network can be constructed and real-time anomaly detection can be carried out by adopting a data processing and analysis technology, and potential health problems are effectively mined from complex data through continuous analysis of behavior data, so that the health management efficiency of the cattle herds is improved. Early recognition is carried out on subtle changes of behavior modes, the accuracy of disease prediction and the perspectiveness of decision making are remarkably improved, errors of manual judgment are reduced through automatic data analysis, economic losses are effectively reduced, and the overall efficiency of pasture management is enhanced.
Owner:第八师石河子市畜牧水产发展服务中心

Coronary heart disease auxiliary diagnosis system, system operation method and diagnosis device

The invention discloses a coronary heart disease auxiliary diagnosis system, a system operation method and a diagnosis device, and belongs to the field of medical auxiliary diagnosis devices. A coronary heart disease classification module of the system comprises a fine tuning encoder, a foreground prototype extraction module, a multi-view feature fusion module and a classifier. The fine-tuning encoder comprises a VIT frozen encoding block, a global feature fine-tuning module and a local feature fine-tuning module. According to the method, local and global prior information in the coronary heart disease vascular structure image is effectively mined, and the problem of difficulty in early screening of the coronary heart disease is solved; compared with the prior art, the coronary heart disease early-stage screening auxiliary system can be constructed quickly and accurately based on a small amount of coronary artery angiography data, and the coronary heart disease early-stage screening auxiliary system has the advantages that the coronary heart disease early-stage screening auxiliary system can be constructed quickly and accurately based on a small amount of coronary artery angiography data; therefore, rapid, accurate and high-throughput coronary heart disease diagnosis is realized.
Owner:JINLIN MEDICAL COLLEGE

Remote sensing image change detection method and system based on progressive distraction mining

The invention relates to a remote sensing image change detection method and system based on progressive distraction mining, and belongs to the technical field of dual-time-phase remote sensing image processing. Multi-scale features of remote sensing double-time-phase images are respectively extracted through a convolutional neural network, difference of double-time-phase features of the same scale is respectively carried out, an absolute value is solved, a multi-scale difference feature map is obtained, and a change area prediction map is obtained by using the obtained small-scale difference features. The method comprises the following steps: acquiring a change area prediction map, guiding a large-scale difference feature map to distraction mining a change area according to the change area prediction map, then acquiring the change area prediction map by using large-scale difference features after guiding refinement, and guiding refinement of larger-scale difference features, thereby realizing progressive guidance, and finally inputting an obtained multi-scale fusion difference feature map into a predictor for prediction. And calculating a loss value updating model. According to the method, secondary mining is carried out on the uncertain fuzzy change area, and the problems of false change and missing detection change are effectively avoided.
Owner:INSPUR OPTOELECTRONICS SATELLITE TECHNOLOGY (SHANDONG) CO LTD

Time sequence prediction system based on causal relationship

The invention belongs to the field of data mining in deep learning, and particularly relates to a time series prediction system based on a causal relationship. The time series prediction method comprises a causal inference and enhancement algorithm suitable for time series data; a causal-frequency domain attention module based on a graph attention mechanism and Fourier transform; the invention discloses a decoder based on a multi-layer perceptron. According to the method, the modeling capability of the existing multivariable time sequence prediction method on the dependency relationship between the sequence variables and the feature extraction capability on the long-time sequence data can be improved, and the error of the prediction result is effectively reduced.
Owner:FUDAN UNIVERSITY

Mutual inductor prediction method based on MHA-CNN-SLSTM and error compensation

The invention discloses a mutual inductor prediction method based on MHA-CNN-SLSTM and error compensation. The mutual inductor prediction method comprises the following steps: decomposing mutual inductor error data into different high and low frequency components by adopting VMD; key parameters of VMD are optimized based on an improved seagull optimization algorithm ISOA, and optimal decomposition of error data is achieved; performing cross processing on error influence features based on a multi-head attention mechanism (MHA), mining relevance among the features, and establishing a deep relation between weak correlation features and errors through a relation between strong correlation features and the errors; the CNN is arranged at the rear end of the multi-head attention mechanism MHA, and the output feature dimension of the multi-head attention mechanism MHA is reduced; the correlation between a training set and a test set is considered, an LSTM neural network is improved, the network weight and bias are dynamically adjusted, and an MHA-CNN-SLSTM combined prediction model is constructed; and inputting the error between the predicted value and the actual value as a training set into the constructed combined prediction model, and generating compensation data to compensate the predicted value. The method can accurately predict the error state change of the mutual inductor in a short period in the future, and has the characteristics of high prediction precision, strong generalization and the like.
Owner:CHINA THREE GORGES UNIV +2

Intelligent office process processing method and system based on large language model

The invention relates to an intelligent office process processing method and system based on a large language model, and belongs to the field of artificial intelligence. Comprising the steps of obtaining enterprise documents; sequentially carrying out document analysis, document cutting and key information extraction on the enterprise document; constructing an enterprise knowledge base based on an ES library according to the extracted content; receiving a user question about the office process; a semantic text block and a keyword text block about the user question are recalled in the enterprise knowledge base, and the semantic text block and the keyword text block are spliced by using a large language model to obtain a user question answer of the user question; taking the user question and the answer of the user question as question-answer pairs, and performing office process work order automatic filling of the question-answer pairs by utilizing a large language model according to a preset work order filling rule; and outputting the filled office process work order. According to the invention, comprehensive intellectualization and automation of the office process are realized, the office efficiency is improved, the error rate is reduced, and the user experience is improved.
Owner:CHINA TELECOM DIGITAL INTELLIGENCE TECH CO LTD

Test task scheduling method and system of control system based on graph model and deep reinforcement learning

The invention belongs to the technical field of automatic testing, and particularly relates to a test task scheduling method and system of a control system based on a graph model and deep reinforcement learning. The method comprises the following steps: analyzing a test task demand, and then constructing a Single-rate DAG task scheduling model; according to a Single-rate DAG task scheduling model, generating a directed graph as the input of a GNN; establishing a heat map about the edge based on the directed graph and the GNN; generating a task scheduling sequence according to the heat map, and planning a scheduling scheme according to the task scheduling sequence; calculating a loss function, and updating the graph neural network according to the loss function; repeatedly establishing the heat map to calculate the loss function until a high-quality solution conforming to an expected target can be planned or a difference value between a mean value and a minimum value of the planned test scheme makespan is smaller than a certain threshold value, and ending iteration; and step 7, outputting an optimized solution. The method is used for solving the problem that a planning result of a high-correlation test task scheduling process is often high in randomness and is difficult to meet actual requirements.
Owner:HARBIN INST OF TECH

Drilling engineering drilling rate prediction network training method and application method

The invention provides a drilling engineering drilling speed prediction network training method and an application method, and belongs to the technical field of drilling engineering optimizing.The training method comprises the steps that data screening is conducted on obtained drilling construction data to obtain effective construction data, variational mode decomposition algorithm noise reduction is conducted on the effective construction data to obtain noise reduction data, and the noise reduction data is used for obtaining the drilling engineering drilling speed prediction network; mechanism model prediction is carried out on the noise reduction data to obtain a mechanism prediction drilling speed, and the noise reduction data and the mechanism prediction drilling speed are combined to obtain drilling speed prediction training data; and inputting the drilling speed prediction training data into the initial drilling engineering drilling speed prediction network, and performing iterative optimization on network parameters of the initial drilling engineering drilling speed prediction network based on a particle swarm optimization algorithm to obtain a well drilling engineering drilling speed prediction network which is completely trained. According to the method, data noise reduction is carried out through the variational mode decomposition algorithm, a drilling mechanism model is combined, data features can be effectively mined, the model accuracy is improved, the calculated amount in the training process is reduced through the particle swarm optimization algorithm, and the training efficiency is improved.
Owner:YANGTZE UNIVERSITY

Environmental sound classification method and system based on metric learning

PendingCN119296550AImprove classification accuracyefficient miningSpeech analysisEnvironmental sound classificationMedicine
The invention discloses an environment sound classification method and system based on metric learning, and the method comprises the following steps: collecting an environment sound audio, and converting the environment sound audio into a Mel spectrogram; performing data enhancement on the Mel spectrogram; an SPA module and an FPL module are constructed, the SPA module or the FPL module is used to process the environment sound audio after data enhancement, a pre-trained convolutional neural network model is used to carry out feature extraction on the processed environment sound audio, and sound features are obtained; inputting the extracted sound features into a representation space; sound features are optimized in combination with comparison, metric learning and cross entropy loss in the representation space, and the optimized sound features are classified; according to the method, the intra-class variation is considered by using metric learning, so that the features of the same kind of samples are closer and the features of different kinds of samples are more separated in the process of carrying out the classification task, and the classification accuracy is effectively improved.
Owner:NANCHANG HANGKONG UNIVERSITY

Enterprise knowledge graph reasoning and complementing method and reasoning and complementing system

The invention discloses an enterprise knowledge graph inference completion method and inference completion system, and the method comprises the steps: obtaining a to-be-completed knowledge graph of a target enterprise, inputting the to-be-completed knowledge graph into a trained knowledge graph embedding model, and predicting an incomplete triple to achieve the inference completion of the knowledge graph; wherein the knowledge graph embedding model comprises an entity node coding module, a multilayer graph convolutional neural network module, a Softmax pooling layer and a gating neural network based on a self-attention mechanism. According to the method, the knowledge graph embedding model can provide more accurate semantic information for feature interaction, multi-level feature interaction and aggregation of the triple on sub-tensor kernel granularity and global granularity can be realized, and hidden semantics behind entities and relationships of the knowledge graph can be more efficiently mined. Therefore, the knowledge graph completion accuracy, the semantic modeling capability, the interpretability and the like are obviously improved.
Owner:CHINA TOBACCO ZHEJIANG IND CO LTD

Microorganism control analysis method and system

The invention discloses a microorganism control analysis method and system, and relates to the field of microorganism control analysis, and the method comprises the following steps: identifying a key microbial community; constructing a microbial diffusion prediction model based on historical microbial concentration information, environmental parameters and the identified key microbial community data, and predicting a microbial diffusion trend and a potential pollution area in a future time period by using the trained microbial diffusion prediction model; synchronizing a prediction result and real-time monitoring data into a digital twinborn model, and updating a simulation state of microbial diffusion; performing a simulation experiment by using the digital twinborn model, and verifying the effectiveness of preset microbial culture environment control measures; the system comprises an identification module, a prediction module, a diffusion simulation module and a measure verification module. The control measure effect is evaluated under the condition that the actual environment is not intervened, and the trial and error cost is reduced.
Owner:NANTONG TUMOR HOSPITAL

Three-dimensional directivity target segmentation method under weak supervision setting

The invention discloses a three-dimensional directivity target segmentation method under weak supervision setting, and relates to the technical field of three-dimensional directivity target segmentation. The method comprises the following steps: an instance extraction process: obtaining a point cloud scene and text description, extracting 3D instances in the scene, and representing the 3D instances as instance masks and category score prediction; in the multi-expert mining process, information is extracted from semantic clues of three dimensions of complete context, attribute and category through a plurality of expert models, and semantic distribution of targets is calculated respectively; and a multi-expert aggregation process: according to characteristics of a current sample, automatically endowing input distribution from a more accurate expert model with a larger weight, and eliminating interference information while highlighting a target instance. According to the three-dimensional directivity target segmentation method under weak supervision setting provided by the invention, only the three-dimensional point cloud data and the corresponding directivity description are provided, and mask annotation is not provided, so that time and energy do not need to be consumed for mask annotation.
Owner:XIAMEN UNIV

Intelligent software security vulnerability mining method and system based on large language model

The invention belongs to the technical field of software security, and aims to provide a software security vulnerability intelligent mining method and system based on a large language model. The method comprises the steps of obtaining a preprocessed vulnerability code data set; obtaining CWE vulnerability data, and constructing a CWE vulnerability library based on the CWE vulnerability data; constructing a program dependency graph set based on the preprocessed vulnerability code data set; creating a sample data set according to the program dependency graph set, obtaining a pre-training large language model, and training the pre-training large language model by using the sample data set and a lightweight fine tuning method to obtain a software security vulnerability intelligent mining model; and receiving a to-be-detected code, and performing vulnerability mining processing on the to-be-detected code based on the software security vulnerability intelligent mining model and the CWE vulnerability library to obtain a vulnerability mining result. According to the method, the potential security vulnerabilities in the software source code can be automatically, efficiently and accurately mined, and a complete vulnerability mining result can be obtained.
Owner:SICHUAN UNIV +1

Multifunctional barren slope repairing machine

The invention discloses a multifunctional barren slope repairing machine, and belongs to the field of seeding, the multifunctional barren slope repairing machine comprises an irrigation assembly, a seeding assembly and a drilling assembly, the irrigation assembly comprises two first side plates and second side plates, and crawler belts are mounted on the opposite surfaces of the two second side plates; according to the barren slope repairing machine, multiple functions such as earth cutting, sowing, fertilization, irrigation and backfilling can be integrated in one device through integrated design, the operation efficiency and the automation level are remarkably improved, and it is ensured that seeds are sown under the proper condition through the unique hole digging mechanism and the screening mechanism; meanwhile, a water spraying mechanism can provide water for the seeds in time to promote growth, through the design of a movement mechanism and a soil covering mechanism of the shell, the barren slope repairing machine has good trafficability under different terrains and can automatically conduct soil backfilling operation, and the operation convenience and efficiency are further improved; on the whole, the barren slope repairing machine effectively reduces manual operation and improves the accuracy and efficiency of crop sowing.
Owner:YANGTZE NORMAL UNIVERSITY

Suggested decision-making method and device based on front-end application system user portrait model

The invention belongs to the technical field of data processing, and particularly relates to a suggestion decision-making method and device based on a front-end application system user portrait model, and the method comprises the steps: collecting the user behavior data of each user according to a pre-constructed user access behavior monitoring index system; constructing a user portrait model based on the user behavior data; acquiring a user portrait report by using the user portrait model; based on the user portrait report, utilizing a pre-established analysis and prediction model to obtain a predicted suggested decision scheme; the analysis and prediction model is constructed by using a historical user portrait report and a historical suggestion decision scheme. According to the technical scheme provided by the invention, the user demand can be efficiently and quickly mined, and the investigation work of the user is greatly reduced; and core users can be quickly distinguished in a labor-saving manner, which are key marketing groups and which are key construction directions are determined, and the problem of long demand confirmation time caused by multiple internal levels of an enterprise is relieved.
Owner:BEIJING GUODIANTONG NETWORK TECH CO LTD +1

Intelligent detection system for switch cabinet

The invention discloses an intelligent detection system for a switch cabinet, and the system comprises a data collection and transmission module which collects parameter data in the switch cabinet in real time; the data preprocessing module is used for preprocessing the collected original data, including data cleaning, denoising and abnormal value processing; the real-time data storage and management module is used for storing the preprocessed data in a MongoDB database; the data analysis and mining module is used for grouping or classifying the data and identifying an abnormal or fault mode; and the real-time response and early warning module is used for generating early warning information in real time based on the analysis result, and sending an early warning notification to related personnel when the monitoring data is abnormal. Through intelligent data processing and analysis, efficient processing and mining of large-scale data are realized, more reliable and intelligent switch cabinet monitoring and early warning services are provided for users, and improvement of stability and safety of a power system is facilitated.
Owner:STATE GRID HEBEI ELECTRIC POWER CO LTD

Transformer fault detection method, device and equipment based on KFWAdaBoost and medium

The invention provides a transformer fault detection method and device based on KFWAdaBoost, equipment and a medium, and relates to the technical field of transformers, the method combines a feature generation technology and an ensemble learning strategy, uses a KPP algorithm to generate clustering features, and analyzes the relationship between samples and cluster centers to encode the distribution mode of the samples. The model generates a weighted soft label for each sample in the classification process, and utilizes the soft labels to express fuzziness and similarity between categories, so that the information expression ability of a data set is effectively improved. And sample weights are adjusted through multi-round iteration, learning of samples difficult to classify is enhanced, and a final classification result is output in combination with a voting mechanism. Besides, the generated new data set is not limited to a KFWAdaBoost model, but can be widely applied to training of various classifiers, so that the utilization efficiency of data and the accuracy of diagnosis are improved.
Owner:KEBO ELECTRICAL EQUIP CO LTD