166 results about "Description model" patented technology

The invention discloses a video content description method by means of space and time attention models. The global time structure in the video is captured by means of the time attention model, the space structure on each frame of picture is captured by the space attention model, and the method aims to realize that the video description model masters the main event in the video and enhances the identification capability of the local information. The method includes preprocessing the video format; establishing the time and space attention models; and training and testing the video description model. By means of the time attention model, the main time structure in the video can be maintained, and by means of the space attention model, some key areas in each frame of picture are focused in each frame of picture, so that the generated video description can capture some key buy easy neglected detailed information while mastering the main event in the video content.

The invention relates to an image description generation method based on a depth LSTM network, comprising the following steps: (1) extracting the CNN characteristics of an image in an image description dataset, and acquiring an embedded vector corresponding to the image and describing the words in a reference sentence; (2) building a double-layer LSTM network, and carrying out series modeling based on the double-layer LSTM network and a CNN network to generate a multimodal LSTM model; (3) training the multimodal LSTM model by means of joint training; (4) gradually increasing the number of layers of the LSTM network in the multimodal LSTM model, carrying out training each time one layer is added to the LSTM network, and finally, getting a gradual multi-objective optimization and multilayer probability fused image description model; and (5) fusing the probability scores output by the branches of the multilayer LSTM network in the gradual multi-objective optimization and multilayer probability fused image description model, and outputting the word corresponding to the maximum probability through common decision. Compared with the prior art, the method has such advantages as multiple layers, improved expression ability, effective updating, and high accuracy.

The embodiment of the specification discloses a processing method, a device and a system for predicting transaction risk. The method comprises the following steps: acquiring real-time transaction data, extracting features from the real-time transaction data, and obtaining a first feature set; the first feature set is classified by using a constructed classification model to obtain a classificationresult, the classification model comprising a model obtained by training based on the marked transaction data in the historical transaction data; detecting outlier of the first feature set by usingthe constructed description model, and obtaining the detection result, the description model comprising a model generated by clustering processing based on historical transaction data; determining a risk prediction result of the real-time transaction according to the classification result and the detection result. By utilizing the various embodiment of the present specification, it is possible toincrease the probability of predicting the risk of a transaction or a user, and at the same time to identify emerging means of committing a crime, thereby preventing from being bypassed by an unlawfulelement.

The invention discloses a deep learning model-based image Chinese description method and belongs to the field of computer vision and natural language processing. The method comprises the steps of preparing an ImageNet image data set and an AI Challenger image Chinese description data set; pre-training the ImageNet image data set by utilizing a DCNN to obtain a pre-trained DCNN model; performing image feature extraction and image feature mapping on the AI Challenger image Chinese description data set, and transmitting image features to a GRU threshold recursive network recurrent neural network;performing word coding matrix construction on an AI Challenger image mark set in the AI Challenger image Chinese description data set; extracting word embedding features by utilizing an NNLM, and finishing text feature mapping; taking the GRU threshold recursive network recurrent neural network as a language generation model, and finishing image description model building; and generating a Chinese description statement. According to the method, the blank of image Chinese description is filled up; a function of automatically generating the image Chinese description is realized; the accuracy ofdescription contents is well improved; and a foundation is laid for development of Chinese NLP and computer vision.

The invention discloses a commodity recommendation method based on model matching, which mainly comprises the following steps of: establishing a commodity description model, wherein the commodity description model comprises two parts, i.e. a character introduction and description part and an attribute information part; establishing a user preference model, ensuring that the user preference model is based on the commodity description model and fusing a plurality of commodity description models through the user preference model; calculating the similarity between the commodity description model and the user preference model; and ranking according to the similarity to recommend to a user. According to the commodity recommendation method, the potential interest of the user can be fully explored, and the accuracy of commodity recommendation can be realized.

The invention relates to an analysis and prediction system for cooperative correlative tracking and global situation of network social events. The system comprises an information fusion unit, a tracking unit, and an analysis and prediction unit. The information fusion unit is used for fusing multimodal features of network social event data to obtain fused information of the multimodal data of social events, thus establishing a semantic description model for the multimodal data of cross-social events. The tracking unit is connected with the information fusion unit and is used for acquiring sematic correlative tracking data of the social events on each aspect according to cross-modal property, cross-platform property and trans-time-and-space property of the network social events in face of network contents including rich multimedia information on the basis of the semantic description model for the multimodal data of the cross-social events. The analysis and prediction unit connected with the tracking unit is used for acquiring social-event-based global situation analysis and prediction data on the basis of the semantic correlative tracking data.

An energy consumption simulation and evaluation system for embedded device in energy consumption evaluation technology for electronic devices, which solves the problem that the energy consumption cannot be simulated under tasks operation condition with the existing systems. The present invention includes a graphical configuration management module for inputting graphical configuration parameters, a device energy consumption model building block for obtaining the AADL description model for software and hardware components through AADL abstract description of software and hardware at system-level, a model transformation module for AADL to GSPN model transformation; and a GSPN simulation module which utilizes QPME tool to simulate GSPN model; selectively loads an external energy-consuming event from the energy-consuming events sequence; and perform energy consumption simulation for the GSPN model in response to the external energy-consuming event by QPME tool to obtain a simulation result of the residual energy and lifetime of the system.

The invention relates to the field of building information models, in particular to an integration, collaborative design and delivery method and system based on BIM and design information. The integration method comprises the following steps: constructing a three-dimensional basic scene; constructing a hierarchical description model, wherein the hierarchical description model comprises a metadatalayer; extracting BIM model information and design information from the multi-specialty BIM software, and storing the design information and the BIM model information to a cloud service; respectivelyassociating the design information and the BIM model information with a metadata unit; step 5, conducting format mapping on the original BIM model, and writing metadata units, model component types and design positioning information corresponding to the BIM model into all components of the GIS format of the BIM model; and step 6, integrating the GIS format of the BIM model into a three-dimensionalbasic scene to generate an integrated three-dimensional scene. The invention also provides a collaborative design method and a delivery method based on the integration method, and a corresponding system.

The invention discloses a design method for an intelligent transformer station model based on CAD (Computer-Aided Design) graph and model integration. According to the method, the design of a secondary loop and the generation of a model file of the intelligent transformer station are synchronously completed; a data object corresponds to the back of each entity, and the model can be operated by a graph, therefore, the consistency of the design of the secondary loop and the model file is ensured, and the problem that the traditional intelligent transformer station is not deep enough is solved; and synchronously the design of the transformer station model can be finished by designers not familiar with 61850 design details, the role and the function of a design organization are strengthened in the establishing of the intelligent transformer station, and the visibility in operation and maintenance of the transformer station is increased; furthermore, the method is combined with the habits of traditional designers and is used for performing graphical loop design based on an AutoCAD platform so that the operation of the designers is convenient; an IEC61970CIM model is used in a primary system to describe the model, therefore, the data supporting other applications can be exported, and the openness of the method is increased.

The invention relates to a failure knowledge storage and push method for the FMEA (failure mode and effects analysis) process. The failure knowledge storage and push method comprises four parts in eight steps: 1, creating a failure product storage module; 2, constructing a failure information storage module; 3, constructing a failure knowledge storage module based on the ontology theory; 4, constructing a word segment module for a failure mode description statement; 5, constructing a failure mode theme matching module; 6, enabling a semantic description model to describe a failure mode theme obtained in the step 2 of the part 2 by using a WDF (word description model); 7, calculating the semantic similarity of information in a failure database and the to-be-inquired fault mode theme; and 8, sequentially reasonably determining the fault mode, the fault reason and the fault influence of a designated product from the lowest indenture level to the initial indenture level of the product according to the step-by-step solution characteristics of the blackboard process. The failure knowledge storage and push method for the FMEA process achieves the effective storage and management and the efficient utilization of the history failure data, and completes the analysis work of the FMEA from the lowest indenture level to the initial indenture level.

The invention relates to a video description method and system based on an information loss function, and the method comprises the steps: obtaining a training video, and obtaining the semantic information of each frame of a set training video; Inputting the semantic information of the training video into an LSTM-combined hierarchical attention mechanism model to obtain character description of thetraining video; According to the importance of each word in the character description to the expression video content, performing loss weighting on the words to obtain an information loss function, and taking the information loss function as an objective function to perform back-propagation gradient optimization on the hierarchical attention mechanism model to obtain a video description model; Obtaining a to-be-described video, respectively inputting the to-be-described video into the target detection network, the convolutional neural network and the action recognition network to obtain a setof target features, overall features and motion features of each frame of the to-be-described video as semantic information of the to-be-described video, and inputting the semantic information into the video description model to obtain character description of the to-be-described video.

The invention belongs to related technical field of optimum structural design, discloses a method for topological optimization of multiphase material flexible mechanisms under stress constraints, and aims at optimizing structures of the multiphase material flexible mechanisms. The method comprises the following steps of: (1) constructing a multiphase material level set topological description model to describe distribution of a multiphase material structure; (2) constructing a rigidity interpolation model and a separable stress interpolation model to respectively calculate elastic rigidity and stress of the multiphase material structure; and (3) constructing a weighing method and stress punishment-based multiphase material flexible mechanism parameterized level set topological optimization model, optimizing output displacement and flexibility of a flexible mechanism, and controlling local stress of the multiphase material structure. The method is applied to topological optimization design of multiphase material flexible mechanisms under stress constraints, the optimized multiphase material flexible mechanisms have the advantages of being high in flexibility and high in rigidity, flexible knot parts of the multiphase material flexible mechanisms do not have single point hinge phenomenon, the structural strength requirements are satisfied, and the stress concentration problems are eased.

The invention relates to a data requirement standardization method and standardization system. The data requirement standardization method comprises the steps that a data requirement standardization description model is created, and the data requirement standardization description model generates the description document of the business data requirements; a data source layer, a requirement model layer and an application layer are created in the data requirement standardization description model, and the data source layer converts the data assets and the data of other data sources into the understandable business data source of the business personnel; the requirement model layer performs information extraction of indicator description under each component of the data source layer and describes the algorithm and operation rules of the indicators under each component of the data source layer so as to generate the description information set of the understandable business data indicators of the business personnel; and the application layer records the combination, the presentation mode, the frequency and the path through which the indicators of the requirement model layer are convertedinto the description information of the understandable business data graphs, tables and information of the business personnel. The barriers of the business and the technology can be broken through.

The invention relates to an image semantic description generation method, and belongs to the technical field of image semantic description. The method comprises the following steps: S1, extracting global features and local features of an image; S2, combining the global features and the local features to obtain combined information; and S3, focusing image features and focusing texts according to the combination information through an adaptive attention mechanism to generate a semantic description model, and generating words and phrases of image semantic description by utilizing the semantic description model. According to the method, the problems that the focus key point of the image is determined, higher-level semantic information is mined, and detail information describing words or sentences is perfected in the existing image semantic description technology are solved.

The invention discloses a method for quantitatively identifying the performance degradation degree of a rolling bearing. The method comprises the following steps of: extracting time domain statistical characteristics of a vibration signal of the rolling bearing, selecting a stability characteristic and a sensitivity characteristic as input vectors, establishing a fuzzy support vector description model, and calculating a fuzzy monitoring coefficient; and introducing operation time on the basis of the fuzzy monitoring coefficient to construct a monotonic damage degree index and quantitatively identify the performance degradation degree of the rolling bearing. By the method, the performance degradation degree of the rolling bearing in the whole life cycle can be quantitatively identified, initial damage and final failure are found as soon as possible, and major accidents are prevented.

The invention discloses an intelligent auxiliary monitoring system and method for a substation. The system comprises an adaptive communication module, an intelligent electronic device (IED) model automatic configuration module, a version automatic matching module and an XML based interface configuration module. The adaptive communication module is used for achieving adaptive communication between a comprehensive monitoring host and various sub-monitoring systems. The IED model automatic configuration module is used for generating IED performance description files rapidly after the system configuration is changed. The version automatic matching module is used for achieving automatic matching of an integrated power supply system and the comprehensive monitoring host software version. The XML based interface configuration module is used for enabling XML files to perform description modeling on the whole character lattice LCD human-computer interface to generate an interface according to the model files. By the aid of the system and the method, high multi-system fusion is performed on technologies of IED model automatic configuration, version automatic matching, adaptive communication, interface configuration and the like, and a uniform novel intelligent auxiliary monitoring system platform which is intelligent and efficient and adaptive to substation development requirements is built.

The invention discloses a data interaction method and a data interaction system for a scheduling master station and a transformer substation based on MMS (Multimedia Messaging Service). The data interaction method comprises the following steps: expanding a communication model based on MMS, obtaining a system description model of the transformer substation, confecting an agency service interface to the transformer substation; setting up the corresponding relation between a resource mark number of the scheduling master station and the resource access path of the transformer substation according to the agency service interface and the system description model of the scheduling master station, performing data interaction with the transformer substation. According to the data interaction method and the data interaction system, the communication is performed in an object-oriented way, thereby solving the problems in the traditional technology that necessary protocol conversion and message point table maintain are needed, and all information types in the intelligent transformer substations cannot be covered. Furthermore, the work load of manual maintaining and the error are reduced.

The invention relates to a video content description method based on frame selection, comprising the following steps: a feed-forward neural network is used for constructing a screening model, and thescreening model screens the video frame according to the visual richness and semantic consistency of the video frame; constructing a description model for describing the content of the video to be described; training the screening model and the description model with training data; Selecting a description frame in the video to be described through the filtering model; the visual features of the description frame are extracted and the description model is input to obtain the description sentence of the video to be described.

A laser-point cloud and image fusion data classification method based on multi-characteristic comprises the following steps: 1, data preprocessing: preprocessing aviation image and unmanned plane laser-point cloud data; 2, sample extraction: fully utilizing geometry characteristics of the point cloud data and spectrum characteristics of the aviation images so as to carry out sample extraction of various types; 3, fusion data classification based on multi-characteristic: using a vector description model to classify the sample data; 4, precision evaluation: evaluating the precision of the classified data. The method is complete in surface object extraction and high in classification precision. The method starts from the angle of the fusion image spectrum information, carries out data fusionaccording to application purposes and surface object classification demands, sets corresponding classification rules for classification of main surface objects, and builds a corresponding relation between classification types and classification characteristics, thus extracting complete surface object areas, and reducing misclassification phenomenon.

The invention discloses an individual tree automatic extraction method based on multispectral LiDAR data. The method comprises the following steps that the nearest neighbor search method is adopted on each laser point in the benchmark point cloud data with the point cloud of any band of the LiDAR data acting as the benchmark data, and band information of the nearest laser point is acquired in the data of other bands so that the single fusion point cloud data including multispectral information are generated; multi-perspective projection is performed along the Z-direction of the single fusion point cloud data including multispectral information, and the point cloud data are segmented into ground point cloud data and non-ground point cloud data; clustering and normalized segmentation are performed on the non-ground point cloud data, and segmentation is performed according to the geometric and spectral characteristics of the point cloud data so that semantically independent local point cloud blocks are obtained through separation; and a tree target overall characteristic description model based on the three-dimensional local abstract class characteristics is established to perform individual tree automatic extraction processing based on deep learning. The individual tree automatic extraction method based on the multispectral LiDAR data has the advantages of enhancing the tree identification accuracy.

A computer aided design (CAD) system. A template generation engine generates templates from interconnect configuration files. A field solver generates high frequency passive element relationships from the templates. A circuit builder generates circuit description files from device technology models and from high frequency passive element relationships. Parameterized circuit description models may be generated for large range of sensitivity analyses. A simulator simulates circuit responses for transmission line models from the circuit description files. Interconnect configuration files may be generated by a geometry and material definition module that receives process description data from a designer.

A resource agent facing to a lighting heterogeneous equipment is provided. The resource agent is design into a physical communication protocol adapter, a service communication protocol adapter, a task execution scheduler and a uniform access interface from top to bottom by adopting hierarchical architecture design on the basis of unified description model of virtual device resources to shield thephysical communication protocol heterogeneity and service communication protocol heterogeneity( of devices. Shielding heterogeneous nature of devices layer by layer and providing uniform interface for accessing, controlling and monitoring devices make distributed, heterogeneous and autonomous static physical devices packed into a unified and dynamic virtual device grid service resource for being invoked by the upper layer for application. The devices provided by the invention can solve the problems of diverse access methods of the devices, monitoring of global devices and difficult cross-region management resulted from different purchase periods of devices, different information degree of each unit, autonomous management of each region and other factors.