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

62 results about "Extended model" patented technology

Building carbon emission analysis method based on STIRPAT algorithm and building information model technology

The invention discloses a building carbon emission analysis method based on an STIRPAT algorithm and a building information model technology, and the method comprises the steps: collecting the whole life cycle information of a building through the BIM technology in a data collection and integration stage, covering the data of all links of design, construction and operation, widely collecting the external environment data of a project location, such as social economy, energy technology, etc. Multi-source data such as geographic space and weather are introduced and fused, and real-time monitoring is performed by means of an intelligent sensing technology. Secondly, an extended STIRPAT model is constructed, core independent variables such as population, economy and technology and control variables such as building scale are comprehensively considered, and the influence of factors is accurately quantified through dynamic parameter adjustment and scene simulation prediction; according to the method, carbon emission attributes are given to components and equipment based on a BIM model, full-life-cycle carbon emission is simulated and calculated, integrated optimization is carried out with solar photovoltaic, intelligent control, waste management and other technologies, finally, results are comprehensively analyzed, carbon emission distribution and trend are displayed by utilizing a BIM visualization function, and a building low-carbon decision is provided.
Owner:EQUOTA ENERGY TECH SHANGHAI CO LTD

Modeling method of engineering cement-based composite material under impact load based on near-field dynamics

The invention discloses a near-field dynamics-based modeling method for an engineering cement-based composite material under an impact load, and the method comprises the steps: constructing an engineering cement-based composite material full-discrete model, and discretizing a cement matrix and fibers in the engineering cement-based composite material full-discrete model into a plurality of space material points; constructing an impact crack propagation model based on near-field dynamics through the plurality of space material points; correcting the impact crack propagation model in combination with a failure criterion of a model strain rate effect to obtain a corrected impact crack propagation model; and analyzing the engineering cement-based composite material under the impact load through the corrected impact crack propagation model. The method can effectively predict and analyze the mechanical behavior and damage problem of the ECC under the impact load.
Owner:ZHENGZHOU UNIV

Master data management method and system based on enterprise data governance

The invention relates to the technical field of data processing, in particular to a main data management method and system based on enterprise data governance, and the method comprises the steps: defining a basic model structure based on an enterprise general main data object, and storing the basic model structure in a metadata form; obtaining a business system change event and a manual expansion demand and converting the event and the demand into a structured description; zero-coding extension configuration of attributes, classification and association relationships is completed through a visual interface; executing interface adaptation, data mapping adaptation and rule conflict detection; and synchronizing the effective extension model to a service layer, a cleaning layer and a monitoring layer. The system comprises a demand awareness module, a dynamic modeling module, an extension configuration module, an adaptation engine module and a life cycle management module. According to the method, full-process management of the master data model from demand capture to business landing is realized, and the model expansion efficiency and the business adaptation capability are improved.
Owner:BEIJING HKRSOFT TECH CO LTD

Natural confrontation sample generation method and system based on expansion model

The invention discloses a natural adversarial sample generation method and system based on an expansion model. The natural adversarial sample generation method comprises the steps of obtaining noise images of a to-be-tested sample at a plurality of preset moments; inputting each noise image into a diffusion model for single-step denoising to obtain a plurality of denoised images at different moments; inputting each de-noised image into a classifier for classification to obtain a log-likelihood probability of each de-noised image corresponding to a target category; obtaining norms of each de-noised image and the to-be-tested sample, and calculating gradients of the log-likelihood probability and the norms to the to-be-tested sample to obtain a gradient value; and updating the to-be-tested sample according to the gradient value to obtain a natural confrontation sample. According to the method, the adversarial sample which is more natural and has certain readability can be generated, additional neural network components and modules are not needed in the process, the calculation cost is lower, plug and play can be realized, and the method can be widely applied to the technical field of artificial intelligence security.
Owner:CHINA TELECOM CORP LTD

Model extension method, electronic equipment and storage medium

The invention discloses a model extension method, electronic equipment and a storage medium, and belongs to the technical field of computers. The method comprises the following steps: based on the type of a target service, configuring an index field for a TR181 extension model file of the target service; based on the description parameters of the target service, at least one description field is configured for a TR181 extension model file of the target service, and one description field is used for expressing features of the description parameters; and setting the length of each description field based on the globalized fitness and global specification of the target service so as to realize the extension of the TR181 extension model file.
Owner:ZTE CORP

A master data management method and system based on enterprise data governance

The present application relates to the technical field of data processing, and more particularly to a kind of master data management method and system based on enterprise data governance, including based on enterprise general master data object definition basic model structure and store in the form of metadata;Business system change event and artificial extension demand are obtained and converted into structured description;Attribute, classification and correlation are completed by visual interface Zero coding extension configuration;Interface adaptation, data mapping adaptation and rule conflict detection are executed;Synchronization of effective extension model to service layer, cleaning layer and monitoring layer.The system includes requirement perception, dynamic modeling, extension configuration, adaptation engine and life cycle management module.The present application realizes the whole process management of master data model from requirement capture to business landing, improves model extension efficiency and business adaptation capability.
Owner:BEIJING HKRSOFT TECH CO LTD

Dynamic evaluation method for safety risk of maintenance construction of village-town crossing road

The invention discloses a village-town crossing road maintenance construction safety risk dynamic evaluation method, and belongs to the field of artificial intelligence modeling and road engineering safety, and the method comprises the steps: building a risk index system comprising multi-level indexes; based on the risk indicator system, a dynamic Bayesian network model is constructed, and the model comprises a root node and a conditional dependency relationship between the nodes; determining the fuzzy probability of each root node in the dynamic Bayesian network model by adopting a fuzzy set theory in combination with an expert scoring method; the conditional probability of a non-root node in the dynamic Bayesian network model is corrected on the basis of a Leak Noise-or Gate expansion model; forward reasoning is carried out through a dynamic Bayesian network model, and a dynamic probability curve of the town-crossing highway maintenance construction safety risk is generated; backward reasoning is carried out through a dynamic Bayesian network model, and the posterior probability of the key risk factors is calculated. According to the method, the safety in the village-town-penetrating maintenance construction process can be guaranteed.
Owner:FUZHOU UNIV

Audio classification method, model training method, system, equipment and medium

The invention relates to an audio classification method, a model training method, a system, equipment and a medium. The training method comprises the following steps: acquiring an extended audio set and an audio tag set corresponding to the extended audio set; inputting the extended audio set into the trained original model, and extracting generated reference original category labels and original features; inputting the extended audio set into an extended model, and extracting a generated original category label and / or a newly added category label and an extended feature generated by the same network layer as the original model; and updating parameters of an expansion model based on the difference degree between the original category label and the reference original category label, the difference degree between the expansion feature and the original feature, and the difference degree between the original category label, the newly added category label and the corresponding audio label in the audio label set, and taking the trained expansion model as an audio classification model. According to the method, the problem of knowledge forgetting possibly occurring when the extension model learns a new category is solved, and meanwhile, the learning ability of the extension model to a new category is enhanced.
Owner:BESTECHNIC SHANGHAI CO LTD

New energy power system minimum inertia demand rapid assessment method based on mechanism-data fusion

The invention relates to the technical field of power system dynamic safety analysis, and discloses a mechanism-data fusion new energy power system minimum inertia demand rapid evaluation method, which comprises the following steps: in an offline stage, establishing an expansion system frequency response model containing new energy; generating a plurality of sample scenes in a given disturbance power range and a new energy permeability range; a model minimum inertia demand and a simulation minimum inertia demand are obtained through expansion of a system frequency response model and stepping type time domain simulation calculation, a polynomial regression method and a polynomial regression equation are utilized, and the coefficient and order of the polynomial regression equation are determined by adopting an optimization algorithm; during online evaluation, real-time disturbance power and new energy permeability of a power grid are obtained; and obtaining a model minimum inertia demand and a corresponding minimum inertia demand deviation compensation value, and outputting a final system minimum inertia demand evaluation result. According to the method, the final evaluation result has the calculation efficiency of the extended SFR model and the evaluation precision of stepping simulation.
Owner:STATE GRID SICHUAN ECONOMIC RES INST

Massive multi-transmit multi-receive system channel prediction method, system and medium

This invention provides a method, system, and medium for channel prediction in a massive MIMO TDD system using a base-extended extrapolation model. The method includes: modeling the uplink channel using a base-extended model, representing the uplink channel as a small number of unknown BEM coefficients, and reconstructing the uplink channel by estimating these unknown BEM coefficients; representing the estimated uplink channel as a linear combination of a discrete ellipsoidal BEM and its corresponding coefficients to reduce the complexity of downlink channel prediction; predicting the coefficients of a downlink time-varying DPS-BEM, and using the predicted DPS-BEM coefficients for downlink channel recovery. This invention alleviates the channel aging problem in massive MIMO TDD systems.
Owner:HARBIN INST OF TECH SHENZHEN GRADUATE SCHOOL

Carbon emission accounting and tracking method and system based on process construction method and BIM

The invention provides a process construction method and BIM-based carbon emission accounting and tracking method and system, and the method comprises the steps: constructing an extended BIM model fusing geology, process and carbon emission attributes, and building a construction carbon emission model through employing a mixed life cycle method; and mechanical energy consumption is simulated and predicted through discrete events, and model components are associated, so that dynamic tracking and visualization of the carbon footprint during the construction period are realized. According to the invention, accurate accounting, dynamic tracking and visual display of carbon emission in the construction period can be realized, a scientific basis is provided for carbon emission management in the construction process, and optimization of construction organization, reduction of carbon emission and improvement of environmental benefits and sustainability of construction projects are facilitated.
Owner:POWER CHINA KUNMING ENG CORP LTD

Protein cysteine residue redox modification prediction method based on MusteDeep framework

PendingCN119920315ASequence analysisInstrumentsProtein oxidationData pre-processing
The invention relates to the technical field of protein redox, and particularly discloses a protein cysteine residue redox modification prediction method based on a MusteDeep framework, and the method comprises the following steps: data preprocessing, convolutional neural network construction, attention mechanism introduction, Bootstrap method introduction, model result output and the like. The extended model is used for predicting the redox reaction of the cysteine residue in the protein, the accuracy of the extended model in predicting the redox reaction of the cysteine residue is remarkably improved, the redox reaction can be learned and predicted from original protein sequence data, predefined feature engineering is not needed, and the accuracy of the extended model in predicting the redox reaction of the cysteine residue is greatly improved. Manual intervention and subjective judgment are greatly reduced, and the objectivity and efficiency of prediction are improved.
Owner:NUCLEAR BIOTECHNOLOGY RES INST OF XINJIANG ACAD OF AGRI SCI (XINJIANG UYGUR AUTONOMOUS REGION BIOTECHNOLOGY RES CENT)

A method for optimizing support resource allocation based on aircraft usage maintenance activities

The application provides a kind of guarantee resource configuration optimization method based on aircraft use maintenance activity, including the following steps: aircraft use maintenance activity classification, basic model construction, construction extension analogy derivation model, guarantee resource configuration project set acquisition, guarantee resource demand set acquisition, subset, function integration, guarantee resource configuration list.The application uses a basic model and three extension models to derive the preliminary configuration of guarantee resources for aircraft field use maintenance activities, and then uses set operation and function integration to finally form the standard configuration of guarantee resources, which solves the problem of aircraft grounding due to the shortage of certain guarantee resources during aircraft field use maintenance, affecting the completion of the task, and solves the problem of excessive guarantee resources, resulting in increased guarantee costs, affecting the efficiency and cost ratio target and aircraft seller market, and also reduces the specifications of guarantee resources through function integration, improving the aircraft transfer capacity.
Owner:中航贵州飞机有限责任公司

Human-in-the-loop Text2SQL iterative data annotation method fused with large model

The invention provides a human-in-the-loop Text2SQL iterative data labeling method fused with a large model. The method comprises the following steps: firstly, carrying out training and fine tuning on a generative model by utilizing a pre-constructed manual labeling training sample; performing same-semantic diversification expansion on the training sample by using the expansion model; inputting the plurality of extended training samples into a generative model to obtain a first SQL query statement; for each first SQL query statement, reward and punishment assignment is carried out on model parameters of the expansion model according to an expert analysis result and a reinforcement learning mechanism to determine a target training sample; performing manual annotation on the second SQL query statement, and performing incremental training on the generation model by utilizing a manual annotation result to obtain a second SQL query statement; and when the verification result of the second SQL query statement shows that the query expectation is not met, manually marking the target training sample circularly until the verification result meets the query expectation. The efficiency of data annotation is improved through man-machine cooperation and reinforcement learning.
Owner:JOINT WARFARE COLLEGE NAT DEFENSE UNIV OF THE CHINESE PEOPLES LIBERATION ARMY

Method and system for carbon emission intensity prediction simulation and land utilization management and control

The invention discloses a carbon emission intensity prediction simulation and land utilization management and control method and system, and belongs to the technical field of carbon emission calculation prediction.The method comprises the steps that firstly, multi-source space and statistical data are integrated to establish a basic database, and the carbon emission intensity of all kinds of space is calculated according to land space classification; then, constructing an extended STIRPAT model to screen a core driving factor, and constructing a feature project containing residual, lag and trend features based on the output of the extended STIRPAT model; training an XGBoost machine learning model by adopting a Bayesian optimization algorithm so as to capture a complex nonlinear relationship between the carbon emission intensity and the influence factors; and finally, inputting future scene feature data, generating a territorial space carbon emission intensity distribution map in combination with a PLUS model, and identifying a high-emission hot spot region. According to the method, the problem that the technical progress influence is difficult to quantify and the spatial explicit prediction is difficult to carry out in the prior art is solved, the prediction precision is remarkably improved, and a scientific basis and decision support are provided for regional territorial space planning and low-carbon management and control.
Owner:ARCHITECTURAL DESIGN RES INST OF GUANGDONG PROVINCE

A distributed power distribution system unified scale load modeling simulation method and system

The application relates to the field of power distribution system simulation, in particular to a distributed power distribution system unified scale load modeling simulation method and system, which comprises the following steps: denoising and fitting a wideband signal to obtain a multiscale signal; obtaining a Hankel matrix time-varying extended model based on the multiscale signal and historical modal parameters and solving to obtain real-time modal parameters; constructing a coupling matrix based on the real-time modal parameters, fitting a fast action layer state equation and a slow action layer state equation by using the coupling matrix to obtain a unified coupling equation; transforming and decoupling the unified coupling equation to obtain a mixed differential-algebraic equation group; and solving the mixed differential-algebraic equation group by using a preset step length switching rule based on a dominant dynamic type to obtain full-time domain simulation results of the distributed power distribution system. The application can effectively solve the problems of time scale fragmentation and insufficient parameter identification accuracy in traditional load modeling.
Owner:CHINA ELECTRIC POWER RESEARCH INSTITUTE CO LTD +2

Enterprise power load prediction method based on multi-influence factor STIRPAT model

The invention discloses an enterprise power load prediction method based on a multi-influence-factor STIRPAT model, and the method comprises the steps: carrying out the correlation analysis through collecting the related historical data of the power consumption of an enterprise, and screening key influence factors having significant correlation with the power consumption of the enterprise; establishing a relation model between each key influence factor and the enterprise power load by adopting an extended STIRPAT model; and performing time sequence prediction on the data of each influence factor in the future by using the ARMA model to obtain variable values of each key influence factor in the future time period, inputting the variable values into the STIRPAT model, and calculating the power load in the corresponding time period. By adopting the method, the demand prediction and optimization problems of enterprises in medium and long term power load planning and management can be solved.
Owner:STATE GRID JIANGSU ECONOMIC RES INST

Deep learning models processing time series data

An Artificial Intelligence (AI) based data processing system transforms a plurality of time series data sets for processing by one or more deep learning (DL) models for generating forecasts. The DL models are initially trained on training data generated from the historical data. During operation, a plurality of transformed time series data sets are generated from the plurality of time series data sets associated with different entities in an entity hierarchy via data flattening and data stacking. A primary model of the one or more DL models is trained on first-party data for generating the forecasts. An extended model of the one or more DL models is trained on third-party data from external data sources. Whenever new data is available in the first-party data or the third-party data, the primary model and the extended model are correspondingly updated.
Owner:ACCENTURE GLOBAL SOLUTIONS LTD

Automatic Construction Method and System for Programming Task Context Based on Eclipse Plug-in

The present invention discloses a method and system for automatically constructing a programming task context based on an Eclipse plug-in, which automatically adapts and selects a context pattern library by identifying programming task features; tracks code elements accessed by a developer within a sliding time window, forms a context model view of the code elements according to the declaration relationships between the code elements and uses the Eclipse SWT technology to display the view; extends the context model by using the AST technology, and then assigns abstract role types; extracts subgraphs of the extended model, performs pattern matching on the subgraphs by using the context pattern library, and records all successfully matched subgraphs; generates a list of recommended code elements by using the Eclipse SWT technology and displays the list. The developer can locate and analyze the recommended code elements by clicking on them. The present invention explicitly captures the programming task context model by a visual method, helps developers quickly understand the programming task context model, and effectively improves the efficiency of software development.
Owner:ZHEJIANG UNIV

A method for airborne software security modeling and automated testing

The embodiment of the application provides a kind of airborne software security modeling and automated testing method, comprising: according to the field of airborne software, the security property description required for security testing is extracted using security property extraction module, while the SysML model is designed, the XML file of SysML model and security property description are uploaded to model automatic expansion module;Model automatic expansion module is expanded according to the uploaded security property description XML file of SysML model, and the node type and path information of SysML model are supplemented, to generate the extended model;Test case automatic generation module generates the comprehensive test path and test data of the extended model according to the extended model output by model automatic expansion module. Since the scheme adds security testing through security property extraction module, model automatic expansion module and test case automatic generation module, the efficiency of test case generation is improved.
Owner:BEIHANG UNIV

Self-evolution multi-agent event semantic understanding system and method based on dynamic extensible model

The invention discloses a self-evolution multi-agent event semantic understanding system and method based on a dynamic extensible model, and the system comprises a multi-agent cooperation framework construction module, a demand understanding and task planning module, a capability verification module, a model dynamic extension module, a task execution module, and a completion module which are connected in sequence. The multi-agent collaboration framework construction module comprises a central control type multi-agent, a task execution type multi-agent and a model expansion type multi-agent, the method focuses on a news event semantic understanding scene, a dynamic extensible system architecture is constructed through a multi-agent collaboration mechanism, a capability verification link is introduced to optimize a task execution process, and the task execution efficiency is improved. And on-demand iteration of the model capability is realized by using an active learning framework, and adaptive evolution of the agent capability is realized, so that the problems of model stiffness, poor adaptability and the like of a traditional event semantic understanding system in a complex news event scene are solved, and the dynamic adaptability and accuracy of news event semantic understanding are remarkably improved.
Owner:UNIV OF SCI & TECH OF CHINA

Modeling method and device of intelligent manufacturing system, electronic equipment and medium

The invention relates to the technical field of industrial system simulation, and provides a modeling method and device of an intelligent manufacturing system, electronic equipment and a medium. The method comprises the steps of obtaining a model library, wherein the model library comprises a preset model; when a model expansion instruction is received, taking a preset model corresponding to the model expansion instruction as an initial model, expanding the initial model according to the model expansion instruction to obtain a new expansion model, and adding the new expansion model as a new preset model into the model library; when a model instantiation instruction is received, determining a target model corresponding to the model instantiation instruction from all preset models in the model library, and generating a model instance of the target model; an intelligent manufacturing system is determined based on each model instance. The intelligent manufacturing system is determined according to the model instance generated by instantiating the preset model in the model library, the efficiency is greatly improved, in addition, the preset model in the model library has an efficient and rapid expansion approach, and the modeling efficiency of the intelligent manufacturing system based on the preset model is improved.
Owner:HANGZHOU HOLLYSYS AUTOMATION

A method and apparatus for generating code from a sysml model

The application discloses a kind of SysML model to code generation method and device, belong to model-driven development technical field, including: first, by embedding code generation configuration parameter and the semantic structure of target programming language through extending SysML model;Then, the model element instance repository containing model element instance and its associated relationship is built and validity check is carried out;Subsequently, the SysML model element instance that passes through check is converted into the abstract syntax tree of target programming language using conversion engine;Finally, the correct syntax tree of verification is filled to pre-defined code template, and executable code is generated.By the application, the structure and behavior semantic mapping of system model to code is realized, the code automatic generation of dynamic behavior such as state machine, activity diagram is supported, and the code generation efficiency and system development quality are significantly improved.
Owner:ZHEJIANG UNIV

A network unknown threat detection method based on feature extended CNN

The present invention discloses a method for detecting unknown network threats based on feature-expanded CNN. A feature-expanded CNN model is constructed based on the fact that many unknown network threats and known threats originate from the same family and are characterized by similar sample features. First, a convolution operation is performed on the original data at each layer of the CNN network to obtain a native feature map; then a linear random operation is performed on the native feature map to obtain an extended feature map; finally, the two are combined to obtain extended reconstructed data of the original data, whose dimension is lower than that of the original data, thereby realizing dimensionality reduction and extended reconstruction of the data. A security data classification model is then constructed based on a shallow machine learning algorithm to realize the detection of unknown threats in network security big data. The method for detecting unknown network threats based on feature-expanded CNN proposed by the present invention generates extended reconstructed features, which not only realize dimensionality reduction, but also expand the data representation of unknown threats, realize high-precision detection of unknown threats, and reduce computational complexity.
Owner:HANGZHOU DIANZI UNIV

Deviation path-based method for identifying and classifying dangerous sources of medium and large unmanned surface vehicles

The invention provides a method for identifying and classifying danger sources of medium and large unmanned ships based on a deviation path, which comprises the following steps of: firstly, comprehensively describing a task process from four dimensions of events, processes, information and states through an MVMOS multi-view modeling method so as to realize panoramic description of personnel, a ship body, an environment and an interaction relationship thereof, secondly, realizing deviation analysis and task process adaptation, and finally, identifying and classifying the danger sources of the medium and large unmanned ships. By expanding time deviation and state deviation guide words, the HAZOP method better fits the characteristics of high task time sequence and discrete state of the unmanned ship; injecting the deviation attribute into a SysML behavior graph, constructing an expansion model based on a deviation path, visually simulating and deducing the propagation and influence of the deviation in a task process, and achieving the dynamic positioning of a hazard source and the analysis of a causal chain; and finally, a four-layer hazard source hierarchical structure is established, typical hazard sources are classified and merged, risk sources, influence paths and associated elements are clarified, systematicness and operability of hazard source management are improved, and clear and reliable input is provided for subsequent risk assessment and safety measure formulation.
Owner:NORTHWESTERN POLYTECHNICAL UNIV

Unmanned aerial vehicle control method and device, unmanned aerial vehicle and storage medium

The invention relates to the technical field of unmanned aerial vehicle autonomous control, and discloses an unmanned aerial vehicle control method and device, an unmanned aerial vehicle and a storage medium, and the method comprises the steps: obtaining a kinetic equation of the unmanned aerial vehicle under a rigid coordinate system, and determining a relational expression of motion parameters, such as position, linear velocity, angular velocity and attitude, with respect to control quantity; constructing an extended state vector, and obtaining an extended state space model through linear conversion; based on a motion parameter measurement value under a world system, the extended model and an observation equation containing observation noise, state estimation is carried out on motion parameters of the unmanned aerial vehicle, and a state estimation value composed of a state item and a disturbance item containing disturbance force / torque is obtained; the target control quantity is solved according to the state item and the disturbance item in the estimated value, the unmanned aerial vehicle is controlled according to the target control quantity, disturbance force / torque is modeled as a dynamic state variable, online real-time estimation is conducted, the disturbance force / torque is fed back to the control quantity, dynamic compensation is conducted, and the strong interference resistance and the control precision of the unmanned aerial vehicle are remarkably improved.
Owner:CHINA HUADIAN ENG CO LTD +1

Audio classification method, model training method, system, device and medium

The present invention relates to an audio classification method, a model training method, a system, a device and a medium. The training method comprises: obtaining an extended audio set and a corresponding audio tag set; inputting the extended audio set into a trained original model, extracting the generated reference original category labels and original features; inputting the extended audio set into an extended model, extracting the generated original category labels and / or newly added category labels and extended features generated by the same network layer as the original model; based on the difference between the original category labels and the reference original category labels, the difference between the extended features and the original features, and the difference between the original category labels, the newly added category labels and the corresponding audio tags in the audio tag set, updating the parameters of the extended model, and using the trained extended model as an audio classification model. The present invention improves the problem of knowledge forgetting that may occur when the extended model learns new categories, while enhancing its learning ability for newly added categories.
Owner:BESTECHNIC SHANGHAI CO LTD

Method and system for generating an extended FEM model

The invention relates to a method for generating an extended FEM model (570) with a 3D geometry of a formed technical component (10), in particular a sheet metal component for a vehicle body, for an FEM simulation for calculating, testing and / or validating design parameters of the technical component (10), wherein the extended FEM model (570) comprises a forming profile (380) of at least one forming parameter (370) occurring during a forming of the technical component (10), comprising the following steps: - Input of at least one FEM model (270); - Performing a modeling of a spatial forming profile (380) of at least one forming parameter (370) by a trained AI model (350), - Generating an extended FEM model (570) with the 3D geometry of the component (10) and the integrated spatial forming profile (380); - Performing an FEM simulation of the component (10) for the calculation, testing and / or validation of design parameters of the component (10), in particular a crash simulation, using the extended FEM model (570).
Owner:DR ING H C F PORSCHE AG

Bathroom Cabinet Kit (Extended Version - Seeking the Boundary Series)

1. Name of the design product: Bathroom cabinet set (extended model - Explore the Boundary series). 2. Use of the design product: This design is a bathroom cabinet set for washing and storing items. 3. Design key points of the design product: lie in the shape and pattern of the product. 4. Picture or photo that best shows the design key points: combined state reference drawing. 5. Other situations that need to be explained Kit description: Kit 1 is a bathroom cabinet; Kit 2 is a storage cabinet.
Owner:FOSHAN DONGPENG SANITARY WARE +3

A large-scale recommendation system training method based on distributed heterogeneous computing

The application discloses a kind of large-scale recommendation system training method based on distributed heterogeneous computing in the field of communication technology, comprising the following steps: Cache initialization;Cache replacement algorithm construction;Index mapping algorithm construction;In GPU, the calculation of Sparsefeature is carried out using the Embedding or Embedding_bag function provided by PyTorch, and the calculation of the subsequent Dense part of the model, the technical solution can make full use of GPU computing resources, and the calculation speed is increased;Replace in and out by row optimization is replaced in and out by block, and the throughput is improved;Finally, the scale of the model is further expanded by making full use of computing resources, and the model is distributedly expanded, and the method has creativity and market promotion value.
Owner:BEIJING LUCHEN TECH CO LTD