30 results about "Si model" patented technology

The application provides a bearing multi-scale lightweight RUL prediction method and device based on a dynamic sparse space-time graph, and belongs to the field of residual life prediction.The method comprises the following steps: adopting an adaptive multi-scale identifier to extract the potential periodicity of a time sequence, and creating a multi-scale representation robust to noise; using a ProbSparse attention mechanism to construct a dynamic and sparse connection ST graph for each scale, and updating the weight of an edge by using a decay matrix, and realizing multi-hop feature propagation by combining a Mixhop GCN, which overcomes the limitation of isolated modeling of traditional time networks and graph networks, and realizes integration of high-precision ST features; a lightweight method is proposed, which adopts two pruning types, one for full connection pruning and the other for hierarchical propagation pruning; the method solves the problems of ignoring key cross-time sensor correlation, the influence of noise on prediction and model complexity, and improves the accuracy and efficiency of RUL prediction.

A computer-implemented method for model building with explainability is provided. The method includes receiving, by a hardware processor, a first metric and a second metric of minimum model performance. The first metric relates to data modeling quality and the second metric relates to model to business rule correlations. The method further includes performing, by the hardware processor, auto Artificial Intelligence model generation responsive to training data and a combination of the first and the second metrics of minimum model performance to obtain a model that is trained and meets model prediction accuracy and model prediction explainability requirements represented by the combination of the first and the second metrics of minimum model performance.

This invention discloses a load-balanced automatic parallel modeling method, device, and storage medium. First, it analyzes the key factors affecting the execution performance of operators and models (operator in-degree, tensor shape, and tensor data type), and proposes a method for constructing a performance evaluation model based on operator characteristics to assess the computational, communication, and overall costs of operators, as well as the training performance costs of the model. Then, with the goal of load balancing across devices, a layer-by-layer partitioning method based on topological sorting is used to quickly divide the neural network model into multiple sub-models with balanced overall costs, achieving coarse-grained partitioning. Finally, based on the model's training performance evaluation model, and according to the communication characteristics between operators, a fine-grained model partitioning and scheduling method based on communication optimization is used to fine-grainedly adjust the coarse-grained sub-models, reducing the amount of cross-device communication tensor transmission to achieve optimal global model scheduling.

The invention provides an online high-precision map construction method and a model construction method thereof, and the model construction method comprises the steps: sequentially constructing a backbone network, a view conversion network, a hierarchical query embedded network and a model optimization network based on a MapTr model architecture, and obtaining an initial model; training the initial model to obtain a map construction model; and carrying out model integration on the plurality of map construction models to obtain an online high-precision map construction model. According to the method, efficient and accurate high-precision map construction can be realized through a stronger backbone network, an effective view conversion and data enhancement technology and a model integration strategy. The method has excellent performance in the aspect of multi-category semantic information extraction, and is suitable for real-time map construction tasks in an automatic driving system.

The application provides a pyramid knowledge distillation framework model compression limit analysis method, comprising the following steps: constructing N groups of online deep mutual learning models in a pyramid structure; performing online deep mutual learning on each group of online deep mutual learning models, and recording the parameter quantity and model performance of two models in each group of online deep mutual learning models; wherein, starting from the second group of online deep mutual learning models from bottom to top, while performing online deep mutual learning, the previous group of online deep mutual learning models is accepted for offline knowledge distillation; the potential representation of all models from the first group to the N-1th group is extracted and sent to an adapter to generate teacher importance weight soft labels; the Nth group of online deep mutual learning models is subjected to online deep mutual learning, and the parameter quantity and model performance of the Nth group of models are recorded; and the balance point of the model compression ratio and accuracy is analyzed according to the parameter quantity and model performance of two models in each group of online deep mutual learning models and the parameter quantity and model performance of the Nth group of models.

The application discloses a bridge risk identification method and system fusing integrated learning and an attention mechanism, and the method comprises the following steps: processing an i-th input feature in an input data set to obtain a reconstructed feature vector; constructing XGBoost and RF-MLP models to perform deep feature extraction, and acquiring fusion features by using an attention mechanism; inputting the fusion features into a back-end multi-task perception decoder, and using parallel decoding branches to decouple and identify the fusion features, wherein a first full connection layer and a first Softmax activation function are used as a risk type decoding branch to output a risk type probability distribution; a second full connection layer and a second Softmax activation function are used as a position decoding branch to output a position probability distribution; a third full connection layer and a Sigmoid activation function are used as a damage degree decoding branch in cooperation with linear mapping to output a damage degree quantitative index; and the application has the advantage of high identification precision.

The present application relates to the field of intelligent prediction, and particularly relates to a weak adhesion hydrogel adhesion performance prediction method and system, comprising the following steps: S1: obtaining hydrogel material parameters and interface state information and preprocessing; S2: identifying state labels through a classification model to form an extended data set with state labels; S3: constructing a segmented physical model system driven by the interface state to obtain state-adaptive physical prediction results; S4: embedding the state-adaptive physical prediction results into a physically-constrained neural network structure to construct a prediction model and obtain model prediction results; S5: using a multi-model collaborative fusion mechanism, comprehensively considering the state-driven physical model output and the model prediction results, and obtaining the final prediction value through a weighted fusion manner; S6: taking the final prediction results and model internal parameters as inputs, performing feature contribution degree analysis, identifying key influencing factors, and outputting parameter optimization suggestions. The present application realizes high-precision and high-generalization-capability prediction.

A sample data acquisition and model training method, medium, device and program product, the method comprising: acquiring a thinking chain template, the thinking chain template being used for defining a target information point, the target information point being used for describing information extracted from an input question for thinking chain reasoning; in response to the acquired current input question, generating question solving step information corresponding to the current input question based on the thinking chain template; the problem solving step information is used for describing a reasoning step adopted when thinking chain reasoning is carried out on the current input problem based on the information described by the target information point; generating sample data based on the current input question and the question solving step information; the sample data is used for training a first question generation model for reasoning based on a thinking chain.

The application provides a bridge construction scene data modeling system and method based on space-time objects; the system comprises a data classification module, a framework modeling module, a multi-dimensional description module and a model integration module; the application introduces dimensions such as cognitive intelligence, behavior action and correlation relationship, so that the digital object has the ability of perception, analysis and decision-making, and realizes evolution from a static model to a dynamic intelligent entity; through systematic description of the composition structure and the correlation relationship, a knowledge network between objects is constructed, and a leap from data integration to semantic correlation is realized; through fusion of real-time perception and cognitive intelligence, the system can actively predict risks and optimize construction schemes based on environmental changes and object states, realizes self-adaptation and self-optimization of the construction process, and improves the intelligent management level of the project.

The application belongs to the field of highway engineering information modeling, and particularly relates to a digital method and device for extracting engineering drawing elements and real-time quantity checking. The method comprises the following steps: S1, generating a three-dimensional BIM model according to a two-dimensional engineering drawing; S2, splitting the three-dimensional BIM model to obtain a BIM model component and a size parameter of the BIM model component; S3, calculating the volume of the component according to the size parameter of the three-dimensional BIM model component, and connecting the three-dimensional BIM model component with an engineering progress; S4, calculating the volume of the component corresponding to the completed engineering in real time according to the engineering progress to generate a bill of quantities; wherein the volume of the simply supported beam bridge comprises the volume of the beam piece and the volume of the wet joint, and the longitudinal slope rate of the road longitudinal slope and the transverse slope rate of the road transverse slope are taken as the calculation parameters of the volume of the simply supported beam bridge. The application is used for realizing more accurate engineering calculation.

The application relates to the technical field of probabilistic graphical models, in particular to a power consumption portrait construction method and system based on a dynamic Bayesian network. Multiple hidden variables are introduced to describe the preference degree of a user for various power consumption attributes, and a power consumption portrait based on a DBN is constructed to intuitively and reasonably explain the influencing factors of power consumption rules; a VQVAE model and a GPN model are introduced to reduce the power consumption preference hidden variables, the reduction does not change the original power consumption rule distribution, the power consumption portrait accurately describes the dynamic relationship between the variables, the construction cost of the power consumption portrait is reduced, the power consumption portrait template is constructed to uniformly learn the relatively fixed relationship within and between time slices, and the related local structure in the template is updated in combination with the power consumption data of adjacent time slices, so that the dynamic relationship existing within and between time slices is captured, and efficient construction of the power consumption portrait is realized. The application aims to solve the problem of how to optimize the construction of the power consumption portrait.

This application relates to the field of large model technology, and more particularly to an evaluation method, system, device, medium, product, and chip. In this method, an inference environment corresponding to each model is pre-built. When evaluating a model, the evaluation environment and the corresponding inference environment are run. Based on the inference environment, the inference service interface corresponding to the model is started, and an inference request is initiated by calling the inference service interface. Based on the evaluation environment, the inference results from the inference service interface are received, and the evaluation task for the model is executed based on the evaluation environment and the inference results. In this way, the inference process and the evaluation process of the model are decoupled, allowing the evaluation environment to adapt to the evaluation of various models.

Receive a question via a graphical user interface (GUI), obtain a passage of text potentially relevant to the question, and receive, via the GUI, a selection of a number of question-answering models to be ensembled. Produce a plurality of answers to the question by running a plurality of question-answering models, consistent with the selection of the number of question-answering models to be ensembled, on the passage of text. Produce an ensembled answer by ensembling the plurality of answers according to their respective confidence scores. Display, via the GUI, the ensembled answer in context of the passage of text, with the ensembled answer visually marked in the passage of text. Optionally, repeat these steps for a second passage of text.

This invention relates to an efficient machine learning model management method and system. The method includes: S1, receiving a model type and model name input by a user, and determining pre-set model configuration information corresponding to the model type and model name input by the user; the model configuration information includes: a pre-set model identifier ID corresponding to the model type and model name, pre-specified feature parameters, a pre-set training duration, a pre-set training algorithm, a pre-set evaluation duration, and a pre-set model update cycle; S2, obtaining the model corresponding to the model configuration information based on the model configuration information; S3, evaluating each model in the model group according to a preset evaluation mode, and obtaining the evaluation result; S4, storing the models that meet the preset conditions in the evaluation results.

This invention discloses a 3D model delivery system, method, and storage medium, comprising a model and data input layer, a model decomposition layer, a model annotation layer, a model reconstruction layer, and a model delivery layer. The model and data input layer provides an interface for data conversion and preprocessing of uploaded 3D object models. The model decomposition layer calls the 3D object models and decomposes them into minimum delivery units. The model annotation layer obtains the preprocessed 3D object models, sets annotation parameters for different 3D object models, and fills in the annotation parameters. The model reconstruction layer constructs the relationships between the minimum delivery units and reassembles them into an engineering building model group. The model delivery layer stores the 3D object models, annotation parameters, minimum delivery units, and engineering building model groups, interacts with other layers, and generates delivery documents, including a list of minimum delivery units and engineering files.

The application provides a process system modeling method and device, electronic equipment and readable storage medium, and belongs to the field of three-dimensional modeling. The method comprises the following steps: obtaining project requirement information of a process system to be modeled, determining preset standard unit modules, module parameters and connection parameters of each preset standard unit module according to the project requirement information; constructing each unit module according to the preset standard unit modules and the module parameters, constructing the connection pipe sections of each unit module according to the connection parameters, and obtaining a system structure model; determining the labeling information of each unit module according to the module parameters, constructing the labeled axonometric drawing of each unit module according to the labeling information, and performing three-dimensional labeling on the system structure model according to the labeled axonometric drawing to obtain a process system model. The unit module is constructed through the model parameters and the preset standard unit modules, the parameterized process system modeling can be realized, the modeling efficiency and the accuracy of the model are effectively improved, and the process system model is conveniently used through the three-dimensional labeling.

The embodiment of the present application provides a kind of processing method and device of three-dimensional model, it is related to three-dimensional model processing technical field.The method comprises: the model file of the three-dimensional model to be processed is parsed, obtains the texture image and model description file of the three-dimensional model to be processed;Image compression processing is carried out to the texture image, and compressed texture image is obtained;According to the compressed texture image and the model description file, the target model file corresponding to the three-dimensional model to be processed is generated.The embodiment of the present application is used to reduce the data amount of three-dimensional model.

The present invention relates to the field of Building Information Modeling (BIM) technology, specifically to an intelligent processing and automatic generation system and method for building 3D point cloud data and BIM models. By constructing multi-scale component feature representations using continuous synchronization theory, utilizing a topology-aware graph network for relational reasoning, and employing homotopy group theory to achieve topological preservation in component completion, the system significantly improves the accuracy and completeness of BIM model generation.

The application belongs to the field of industrial internet, and particularly relates to a kind of intelligent interaction three-dimensional modeling method, device, equipment and storage medium. First, in response to the interaction operation triggered by the client, the modeling text is determined, then the workpiece to be assembled and its three-dimensional modeling parameters are determined according to the modeling text, the corresponding three-dimensional modeling template is found out and spliced with the parameters to generate a three-dimensional modeling scheme, then the scheme is input into the preset modeling model to obtain a three-dimensional target model, and finally the model is output. The method automatically analyzes the modeling text and completes three-dimensional modeling with the help of the preset template and model, solves the problem that non-professional users cannot operate modeling software for modeling, realizes efficient and convenient three-dimensional modeling, and reduces the modeling threshold.

A kind of double-agent model optimization method of wind snow-removal robot pipeline structure parameter, specific steps are as follows: (1) according to the initial design of wind snow-removal robot pipeline, determine the design variable and value range of pipeline structure parameter;(2) establish high-dimensional data set;(3) establish low-dimensional data set;(4) select a proxy model to obtain high-dimensional proxy model A and low-dimensional proxy model B;(5) generate active learning candidate sample pool;(6) calculate the under-learning degree score of each candidate learning point;(7) establish active learning point set;(8) carry out CFD simulation to obtain updated data set, continue to train model A, when the prediction accuracy of model A, the prediction accuracy of model B and the inconsistency degree of model A model B prediction result all reach specified threshold, stop active learning.The present application provides an efficient, reliable and engineering applicable method for the optimization of complex pipeline parameters with multiple structural variables and multiple outlet flow responses.

The present invention discloses a method and system for performing building design calculations and analyses The method includes receiving user inputs in various formats, retrieving data such as equipment specifications, engineering standards, historical project data, and market prices, and storing the data in memory. The processor determines parameters including electrical, lighting, HVAC, and plumbing parameters, and performs Mechanical, Electrical, and Plumbing (MEP) calculations and building design analyses. The system generates outputs such as MEP design reports, bills of quantities, product comparison reports, and engineering service layouts. Additionally, the invention supports diverse input formats, including CAD and 3D Revit models, and enables calculations for equipment sizing, material requirements, and cost estimation. Outputs are verified for compliance with regional and international standards and can be displayed via a user interface, allowing real-time adjustments.

A system enables agile model development to speed up innovation by data scientists. Model training and deployment are coordinated and standardized to reduce redundancy. Data is obtained for feature generation and reformatted and de-sensitized for storage. The features are stored in locations available to all models and training modules of a system so data does not need to be adjusted for new models. To generate a machine learning model, the system establishes a cohort for evaluation by the model. A model template and features for use by the model are identified. The selected template and features are used for experimentation and evaluation. Model training artifacts, such as model weights are subsequently recorded in a model store and the model scripts and settings can then be registered in a centralized database where it can be accessed for execution.

This invention relates to a method and system for adaptability analysis of impact scenarios and CGE models based on multimodal fusion. By extracting impact scenario features and model structural features separately, it achieves comprehensive acquisition of scenario textual information and model structural information, avoiding analytical bias caused by single-dimensional information. By encoding both into the same space, it eliminates the heterogeneity between different modal data. By employing a masked cross-attention mechanism for fusion, it can adaptively focus on the key correlation between scenario features and model features, achieving dynamic and accurate modeling of their matching relationship. By directly outputting the adaptability analysis results based on the fused joint features, the entire analysis process does not rely on human experience intervention, avoiding the uncertainty caused by subjective judgment, and thus accurately outputting the adaptability analysis results.

The application provides a large language model knowledge fine-tuning method, which aims at professional fields to improve the understanding and reasoning ability of the model. The method first extracts knowledge from professional literature, generates question and answer pairs, including correct and incorrect answers, and conducts comparative analysis to enhance the depth of understanding of the model. Further, the question restatement mechanism is introduced to improve the clarity of question expression and the critical thinking of the model. The application of ChoiceBoost technology expands the data set by adjusting the answer order, reduces the position bias, and improves the data utilization. Overall, the application reduces the model reasoning time, enhances the generalization ability, improves the processing efficiency and accuracy of professional field tasks such as network communication tasks, and overcomes the current lack of high-quality training data problem, providing a new fine-tuning paradigm for the development of large language models.

A kind of deformation detection method of flexible constant force clamping mechanism under displacement input, after the two-dimensional grid model of the flexible constant force clamping mechanism to be measured is constructed, based on the Mooney-Rivlin second-order strain energy function model of super-elastic material and initialization model parameters are generated according to boundary constraint condition and load setting;Based on the virtual work principle discretization equation under the static equilibrium state of continuum, using the displacement control method based on displacement increment iteration, the action point of load application, i.e. control point is handled by displacement application and the displacement increment of corresponding non-constrained point is solved, then the node position coordinates and model coefficients are updated until the convergence criterion is met;Finally, the node deformation solution meeting the equilibrium equation is obtained, and the node displacement, internal stress and other data of the process are processed and charted.The present application can quickly and conveniently model similar examples with simple parameter setting, and obtain the structure deformation condition and force-displacement output characteristic results with high precision.

The application discloses an assembly method, device and medium of an industrial part, and relates to the technical field of robot control in industrial automation. The method comprises the following steps: obtaining the part category and the detection frame of each industrial part based on an RGB image and a depth image; obtaining the segmentation mask of each industrial part based on the detection frame and the RGB image, and obtaining the 3D model corresponding to each part category; inputting the segmentation mask, the RGB image, the depth image and the 3D model into a pose estimation model to obtain the 6D pose of each industrial part output by the pose estimation model, and determining the grabbing pose of a robot based on the 6D pose; determining the grabbing pose and the grabbing path of the robot based on the assembly sequence of the industrial part and the 6D pose, and controlling the robot to grab and assemble based on the grabbing path and the grabbing pose. The application is used to solve the problem of poor assembly precision caused by poor part recognition precision in the prior art, and realizes accurate recognition and accurate assembly of industrial parts.