48 results about "Bridging model" patented technology

The invention relates to the technical field of bridge management, in particular to a bridge informatization management system based on digital twinning. Comprising a stress acquisition device and a traffic flow acquisition device, the stress acquisition device is arranged at a key stress point of a bridge, and the server comprises a bridge modeling module, a data receiving module, a model prediction module and a state analysis module; the bridge modeling module is used for establishing a bridge high-precision three-dimensional model through BIM and constructing a digital twinborn body of the bridge; the data receiving module is used for receiving the sensor measured value uploaded by the stress acquisition device and the traffic flow information uploaded by the traffic flow acquisition device; the model prediction module is used for converting the traffic flow data into a dynamic load and inputting the dynamic load into a digital twinborn body of the bridge for transient analysis to obtain a model prediction value of each bridge key stress point; and the analysis module is used for judging the bridge state according to an error value of the model prediction value and the sensor actual measurement value, and generating bridge state early warning when the error value exceeds an error threshold value.

The embodiment of the invention provides a BIM-based bridge model intelligent construction method and system, and belongs to the field of BIM-based bridge model intelligent construction. The method comprises the steps that bridge model construction related data are collected and preprocessed, and a component library is established according to the preprocessed bridge model construction related data; defining geometric linkage rules and topological connection logic of components in the component library, and generating an initial BIM bridge model through a generative adversarial network; performing lightweight finite element calculation on the initial BIM bridge model to obtain an optimized BIM bridge model; and performing standard conflict detection and conflict parameter adjustment on the optimized BIM bridge model to obtain a final bridge model. According to the method, the parameter relation matrix and the dynamic linkage mechanism are constructed, the generative adversarial network is introduced to generate the complex bridge type, and the graph convolutional network is introduced into the generative adversarial network, so that reasonable bridge type topology and dynamic parameter linkage are guaranteed.

According to the image reconstruction method and system of the knowledge distillation-based diffusion bridge model and the terminal provided by the invention, the diffusion bridge model is utilized to directly generate the high-resolution image from the low-resolution image, the traditional Gaussian noise initial state is abandoned, and the credibility and anatomical consistency of the reconstruction result are improved. According to the method, the training strategy of knowledge distillation is utilized, an image reconstruction teacher network model needing multi-step reasoning is compressed into an ultimate image reconstruction student network model needing only one-step reasoning, the image fidelity is guaranteed, meanwhile, the calculation time is shortened, and the clinical real-time requirement is met. Moreover, some extra data without pairing are introduced in the training process of the image reconstruction student network model, so that the model learns more data samples, and the generalization of the model is improved through the semi-supervised training strategy.

The invention provides an annotation generation method for a multi-party collaborative design bridge model, a medium and equipment, and the method comprises the steps: obtaining annotation information inputted by an auditor, and the annotation information comprises the opinion content of a part of components in the bridge model; extracting a component number pointed by the opinion content from the annotation information, and determining a target component corresponding to the component number from the bridge model; calling an API (Application Program Interface), and generating an annotation containing corresponding opinion content at a target component in the bridge model; and determining a target designer responsible for the target component based on the component number of the target component, and generating task prompt information to prompt the target designer. According to the scheme, the annotation information of the auditor is automatically analyzed, the component number is identified, the target component is positioned in the model, the API is called to automatically generate the annotation, and the associated designer is determined by using the component number, so that the task prompting function is realized, and the working efficiency of multi-party collaborative design of the bridge model is effectively improved.

The invention discloses a structural semantic guided bridge point cloud Gaussian splash modeling and rendering method, and belongs to the technical field of computer vision and bridge engineering digitization. The method comprises the steps of collecting a multi-view image of a bridge, and obtaining a sparse point cloud and a camera pose through a motion recovery structure technology; dense point clouds are generated, deep learning semantic segmentation is carried out, and key components such as piers, bridge floors and cables are identified; mapping the semantic tag back to the sparse point cloud; performing differential Gaussian element initialization on the sparse point cloud based on semantic tags, and endowing different parts with initial shapes conforming to geometric characteristics of the different parts; a geometric constraint loss function is introduced in the optimization process for joint optimization; and finally, outputting a Gaussian splash model of a high-fidelity and accurate geometric structure. According to the method, the problems of model redundancy and geometric distortion when a traditional Gaussian splashing technology is used for processing a bridge scene are solved, and lightweight and high-precision reconstruction and real-time rendering of the bridge model are realized.

The invention relates to an assembly type bridge modeling method based on a large model and an RAG technology. The method comprises the steps that a bridge standard component library is established based on an assembly type bridge standard image set; calling a large language model, and generating standardized knowledge statements based on the bridge standard component library; a word embedding method is adopted to convert the normalized knowledge statement into vectors, and the vectors are stored in a vector database; converting an input natural statement query into a vector by adopting a word embedding method, and retrieving in a vector database to obtain a normalized knowledge statement related to the query; performing parameter analysis and optimization on the retrieved knowledge statements based on a large language model by utilizing an RAG technology so as to generate component parameters; and calling a component modeling algorithm based on each component parameter to generate a component three-dimensional model, and assembling the component three-dimensional model to obtain a fabricated bridge model. According to the method, component parameter sorting, knowledge statement intelligent combination, efficient semantic retrieval, parameter optimization, three-dimensional model automatic generation and the like are organically integrated, and the modeling efficiency and quality are improved.

The invention relates to a bridge prestressed reinforcement elastic compression loss prediction method and system and a storage medium. The method comprises the following steps: simulating a bridge model through finite element software; the sum of all the prestressed reinforcement units is calculated, and temperature loads applied to all the reinforcement units are calculated; according to a prestressed reinforcement tensioning sequence and a life-death unit method, temperature loads are applied to reinforcement units one by one, and calculation is carried out; the method comprises the following steps: constructing an original data set, preprocessing, and dividing to obtain a training set, a verification set and a test set; constructing an MLP multilayer perceptron neural network model as a prediction model of the elastic compression loss of the bridge prestressed reinforcement to the concrete, and training the prediction model; and utilizing the trained prediction model to predict concrete elastic compression prestress loss caused by tensioning other prestressed reinforcements by the prestressed reinforcements. According to the invention, the elastic compression loss prediction of the bridge prestressed reinforcement can be quickly and accurately realized.

The invention relates to the technical field of infrastructure operation and maintenance, and provides a bridge digital twin modeling system for structure detection, which comprises a parametric modeling rule base, an input interface, a parameter processing module, a geometry generation engine and an output module, and is characterized in that a user only needs to provide a bridge type and a small amount of key geometry parameters through the input interface; the parameter processing module can automatically derive and supplement all parameters required for generating a complete model according to a pre-stored rule base; and the geometric generation engine then programmatically generates a component section and a three-dimensional component according to the complete parameter set and performs integral assembly, and finally outputs a complete bridge three-dimensional model. According to the method, the defects that in the prior art, all accurate parameters are input manually, and modeling efficiency is low are overcome, the high-fidelity bridge model can be rapidly and automatically generated on a detection site only by means of limited data, modeling efficiency and feasibility are remarkably improved, and iterative optimization and integrated application of the model are naturally supported.

The invention discloses a visual navigation method and system based on a denoising diffusion bridge model. The method comprises the following steps: acquiring initial image data; performing feature coding processing on the initial image data to generate a context vector; performing linear modulation processing on the context vector to generate a conditional variable; according to the context vector and a preset motion rule, a prior action is generated; inputting the priori action and the conditional variable into a denoising diffusion bridge model for diffusion training, and generating a target navigation action; wherein the priori action is used as an initial state of the de-noising diffusion bridge model during reverse de-noising processing. Redundant iteration in the diffusion process can be effectively reduced, accumulative errors are reduced, the efficiency and stability of action generation are greatly improved, and the method can be widely applied to the technical field of computers.

The embodiment of the application provides a kind of based on BIM bridge model intelligent construction method and system, belong to based on BIM bridge model intelligent construction field;The method includes collecting bridge model construction related data and pre-processing, according to the bridge model construction related data after pre-processing, component library is established;The geometric linkage rule and topological connection logic of component in the component library are defined, and the initial BIM bridge model is generated through the generation of adversarial network;Lightweight finite element calculation is carried out on the initial BIM bridge model, and the optimized BIM bridge model is obtained;After the specification conflict detection and conflict parameter adjustment of the optimized BIM bridge model, the final bridge model is obtained.The application constructs parameter relation matrix and dynamic linkage mechanism, introduces the generation of complex bridge type in adversarial network, and introduces graph convolution network in adversarial network, to ensure that bridge topology is reasonable and parameters are dynamically linked.

The invention provides a BIM-based bridge model conflict coordination method and system and electronic equipment, and the method comprises the steps: carrying out the multi-party collaborative design of a bridge model through a BIM platform, and carrying out the collision conflict detection of the designed bridge model; the collision conflict data are exported, and each piece of collision conflict data comprises a collision component ID, a collision component attribute and a collision component coordinate; performing feature extraction on each piece of collision conflict data, and determining a corresponding input feature vector; and inputting the input feature vector into the collision conflict coordination model to obtain an output result of the collision conflict coordination model, namely a conflict coordination scheme. According to the scheme, the technology of an existing BIM platform can be fully utilized to detect collision conflicts, the collision conflict data can be utilized, feature extraction can be performed, input feature vectors can be formed, the collision conflict coordination model of external training can be utilized to generate the conflict coordination scheme, dependence on artificial experience is effectively reduced, and the intelligent collision conflict coordination scheme is provided.

The application discloses a GIS environment bridge scheme parameterization model high-precision and high-efficiency modeling and assembling method, relates to the technical field of bridge engineering, and comprises the following steps: acquiring bridge span configuration data in a GIS environment, arranging a static bridge model according to the bridge span configuration data, and determining an initial bridge arrangement strategy; determining engineering mileage coordinates according to the initial bridge arrangement strategy, converting the engineering mileage coordinates, and determining the accurate geographic coordinates of each hole beam and the corresponding control points in the bridge arrangement control points; calling a geometry engine based on a target beam type corresponding to the initial bridge arrangement strategy to obtain a dynamic bridge model; and assembling the static bridge model and the dynamic bridge model based on the accurate geographic coordinates to obtain a bridge three-dimensional model. The method can construct a multi-scale precision bridge model, adjust appropriate static bridge models and dynamic bridge models, assemble and deploy the bridge models according to high-precision positioning coordinates, and obtain a bridge parameterization model with high precision and high efficiency.

The application provides a bridge process information generation method and device, electronic equipment and storage medium, and relates to the technical field of bridge information processing.The method comprises the following steps: cutting a pre-constructed bridge model based on a preset segmentation strategy to obtain multiple segments of a rod piece; determining the contact information between the parts in each segment of the rod piece based on the preset constraint information corresponding to the bridge model; obtaining the segmentation process information corresponding to each segment of the rod piece according to the contact information corresponding to each segment of the rod piece, and obtaining the bridge process information based on each segmentation process information; wherein the segmentation process information comprises welding information determined based on the line contact information and / or coating information determined based on the surface contact information.The application determines the contact information between the parts based on the preset constraint information, provides accurate and reliable reference for the generation of process information, and is beneficial to avoiding the problems of low efficiency and the like when relying on manual acquisition of bridge process information.

The present invention discloses a method for analyzing the seismic time-history response of bridges driven jointly by physics and data, which relates to the technical field of seismic response analysis methods. The present invention includes the following steps: S1: Nonlinear element simulation of support and restraint components; S2: Determination of highway bridge model parameters; S3: Establishment of a parametric finite element representation method for bridges; S4: Establishment of a highway bridge database; S5: Establishment of finite element models for each bridge sample; S6: Design of a deep network for mapping the input and output layers of the bridge; S7: Establishment of a surrogate model for analyzing the seismic time-history response of bridges. The present invention strengthens the physical basis of machine learning methods and makes full use of their powerful nonlinear characterization ability to accurately and efficiently analyze the seismic time-history response of bridges, which has important engineering value for the evaluation methods of seismic performance of new-generation single and regional bridge structures, and also has important significance for the development of basic innovation theories of interdisciplinary intersections.

The present disclosure relates to a bridge information processing method, device, equipment and storage medium. The method comprises: determining a plurality of structure information corresponding to a plurality of bridge structures in the bridge, and a plurality of component positions corresponding to a plurality of bridge components in each bridge structure; for each bridge structure, generating a component identifier corresponding to each component type in the bridge structure according to the assembly information and the component detail information, and assigning the component identifier to the corresponding bridge component according to the component type and the component position; determining a plurality of component models according to the component types corresponding to a plurality of bridge components, and adding the plurality of component models to the corresponding component positions to obtain a bridge model corresponding to the bridge; establishing a component tree diagram of the bridge according to the component identifier, and associating and displaying the component tree diagram and the bridge model. Thus, the belonging relationship of the bridge components is intuitively displayed through the component tree diagram, and the unified collaborative management of information is realized.

The application belongs to the field of point cloud data, and discloses a bridge real scene three-dimensional point cloud dataset production method, device and medium, the method comprises the following steps: obtaining dataset production demand; when the dataset production demand is the point cloud dataset production demand of the bridge component, obtaining a plurality of bridge components in the bridge model; extracting the component parameters of each bridge component, and constructing the point cloud dataset of the bridge component based on the component parameters of each bridge component and the parameters of the bridge model; and / or when the dataset production demand is the point cloud dataset production demand of the bridge model, obtaining the entity object corresponding to each bridge component in the bridge model; performing triangulation and laser simulation scanning on the surface of the entity object to obtain the point cloud space information of the entity object, and constructing the point cloud dataset of the bridge model according to the point cloud space information of the entity object. The dataset production method can efficiently generate the dataset meeting the deep learning demand.

The present invention provides a method and system for simplifying a stand-alone device based on a three-dimensional model, including: a bridge model creation step, a base model creation step, an internal component analysis processing step, an exposed component analysis processing step, an independent component designation step, and a model merging and deriving step: geometrically merging the base model with the exposed component set model to generate a body model, and assembling independent components to form a simplified model, and obtaining simplified information of a stand-alone device based on a three-dimensional model. The present invention realizes the compression of the satellite three-dimensional stand-alone model file size, the filtering of internal components, and the unified standard simplification of three-dimensional models in multiple software formats, which provides conditions for the satellite overall design unit to carry out satellite design and calculation, and also improves the design iteration efficiency.

This invention discloses a method for constructing a bridge digital twin health monitoring system, belonging to the field of civil engineering and smart infrastructure technology. The method includes: constructing a three-dimensional geometric model of the bridge; establishing an initial finite element model based on the bridge's three-dimensional geometric model, and using a genetic algorithm to optimize and correct its key parameters to obtain a physical simulation model; acquiring structural response sample data based on the physical simulation model, and constructing an RBF surrogate model using the response surface methodology or radial basis function neural network method to obtain a digital twin bridge model; integrating the digital twin bridge model into a three-dimensional visualization platform to realize visual perception, interactive operation, and remote management of the bridge structural status, thereby constructing a bridge digital twin health monitoring system.

The application provides a bridge data conversion method and system of BIM and a finite element model, and relates to the technical field of bridge model optimization. Based on Revit API, component traversal extraction and attribute information extraction are performed on the BIM model to generate a structured data table; a first classification model is used to automatically identify the component type, attribute supplementation is performed in combination with a second rule mapping process, and an optimized structured data table is obtained; subsequently, mesh division and quality optimization processing are performed to obtain high-quality mesh division information; finally, a target finite element model is constructed through material attribute mapping and information addition. The application realizes seamless connection of the BIM model and finite element analysis, provides efficient and accurate technical support for bridge structure design and analysis, effectively solves the problems of low model conversion efficiency and modification difficulty in bridge structure analysis, and improves the informatization and intelligentization level of bridge engineering.

The present application relates to a kind of three-way change approach bridge model construction method, device, equipment and storage medium, belong to water transport engineering technical field, wherein, the method includes by approach bridge axis to the multiple row frames of approach bridge simulation model is carried out coordinate calculation to obtain the three-dimensional coordinates of each row frame, so that the change of the plane and the elevation of approach bridge can be determined;According to the preset cross section data corresponding to the horn mouth of wharf side, middle section and shore side horn respectively, the slope of three-dimensional coordinates is calculated, to obtain the height of each row frame on cushion stone, so as to determine the change of approach bridge in cross section;The approach bridge simulation model can also be optimized according to the height of cushion stone and the three-dimensional coordinates of each row frame, so as to construct three-way change approach bridge model according to the change of approach bridge in plane, elevation and cross section, realize the quick and accurate parameter adjustment of three-way change special-shaped approach bridge, without manual parameter adjustment, save a lot of human resources and cost.

The invention provides a terrestrial heat development scene-oriented domain decomposition type multi-scale intelligent simulation method, and belongs to the field of intelligent simulation, and the method comprises the steps: obtaining terrestrial heat resource data according to a terrestrial heat development scene, constructing a multi-scale simulation model, and obtaining the deep feature representation of the multi-scale simulation model through a deep neural network; carrying out feature fusion on the deep feature representation to obtain a common subspace; constructing an unsupervised clustering neural network model, performing expansion processing on the public subspace, and constructing an inter-scale bridging model by using a transfer learning algorithm; and utilizing the inter-scale bridging model to obtain a domain decomposition type multi-scale simulation model, carrying out simulation experiment verification, and building a simulation platform oriented to a geothermal development scene. According to the method, the problems that deep features of simulation models with different scales cannot be fully analyzed and utilized by an existing simulation method, and cross-scale transition of the simulation models is difficult due to the fact that simulation areas with different scales coexist and are mutually overlapped in a domain decomposition task are solved.

The present invention provides a domain decomposition-based multi-scale intelligent simulation method for geothermal development scenarios, belonging to the field of intelligent simulation. The method includes: obtaining geothermal resource data according to the geothermal development scenario, constructing a multi-scale simulation model, and using a deep neural network to obtain the deep feature representation of the multi-scale simulation model; performing feature fusion on the deep feature representation to obtain a common subspace; constructing an unsupervised clustering neural network model to perform expansion processing on the common subspace, and using a transfer learning algorithm to construct an inter-scale bridging model; using the inter-scale bridging model to obtain a domain decomposition-based multi-scale simulation model and performing simulation experiment verification, and building a simulation platform for geothermal development scenarios. The present invention solves the problems that existing simulation methods cannot fully analyze and utilize the deep features of different-scale simulation models themselves, and in the domain decomposition task, due to the simultaneous existence and overlapping of different-scale simulation regions, it is difficult for the simulation model to cross-scale transitions.

This invention discloses a heterogeneous federated cross-domain recommendation system based on hypernetworks and its implementation method. This method is designed for heterogeneous federated cross-domain recommendation scenarios, allowing participating clients to flexibly deploy personalized recommendation models based on their local data scale. The method includes performing multi-objective collaborative pre-training within each single domain to obtain high-quality initial embedding representations; and constructing a hypernetwork-based bridging model generator on the server side. By fusing the architecture description of the client's local bridging model with specific user preference information, it dynamically generates weight parameters that match the client's model structure. The client uses these parameters for local fine-tuning and sends gradient updates back to the server for global parameter iteration. This invention effectively overcomes the challenge of directly aggregating model parameters due to heterogeneous embedding layer dimensions, and achieves efficient knowledge sharing and cross-domain collaborative training among heterogeneous clients while strictly protecting user privacy.

The invention discloses an urban bridge model generation method based on parameterization rule driving, and belongs to the technical field of bridge model generation. Comprising the following steps: S1, judging a design mode of a workflow, and generating a skeleton model by adopting different generation modes based on different design modes; s2, inputting the skeleton model into urban three-dimensional modeling software, and using a CGA rule to drive the skeleton model to generate a bridge superstructure three-dimensional model; s3, inputting the skeleton model into urban three-dimensional modeling software, and using a CGA rule to drive the skeleton model to generate a bridge substructure three-dimensional model; s4, dynamically adjusting parameters of the bridge superstructure three-dimensional model and the bridge substructure three-dimensional model by using an interactive parameter adjustment interface; and S5, outputting the adjusted three-dimensional model of the bridge superstructure, the adjusted three-dimensional model of the bridge substructure and the adjusted topographic map layer as scene assets. According to the method, the CGA rule code is adopted to realize automatic modeling, so that the model generation efficiency and precision are improved, and meanwhile, the modeling cost and the data delivery difficulty are reduced.

The invention relates to the technical field of crossing of bridge engineering and machine learning, in particular to a bridge network anti-seismic toughness evaluation method based on a GAN-Stacking model, and the method comprises the following steps: S1, generating a parameterized bridge model and carrying out the vulnerability analysis; s2, constructing a bridge network sample library based on the GAN; s3, bridge network feature calculation; s4, establishing a toughness index; s5, establishing and training a toughness evaluation model based on Stacking; and S6, performing practical application and verification. According to the method, parametric modeling and the GAN technology are combined, a large number of different bridge network samples can be rapidly generated, rich data are provided for a toughness evaluation model, it is guaranteed that sample features conform to an actual network, and the generalization ability of the model is improved. According to the comprehensive toughness index combining the structural toughness and the functional toughness, the recovery capability of the post-earthquake structure and the traffic function of the bridge network is comprehensively considered, so that the evaluation result is more accurate and comprehensive.

A method, system, device and computer-readable storage medium for solving a bridge model, which relates to the field of engineering structure tests, includes determining a first function value based on the target parameters of a target bridge damper and a preset lower bound of spring displacement, where the target parameters include target total displacement, displacement of the damping section at the previous moment, step size, damping coefficient, reference velocity and spring stiffness; determining a second function value based on the target parameters and a preset upper bound of spring displacement; determining spring displacement based on the bisection method, target parameters, first function value and second function value, and determining the nonlinear damping force based on the spring displacement and spring stiffness; constructing a balance equation based on the mass matrix, proportional damping matrix, stiffness matrix, nodal acceleration, nodal velocity, nodal displacement and nonlinear damping force; performing modal decomposition on the balance equation to obtain modal equations of each order, and solving the modal equations of each order based on the explicit integration method to obtain modal displacements of each order. This application improves the efficiency of solving the bridge model.

The invention discloses a monitoring data and natural language driven bridge finite element model rapid construction method, system and device and a storage medium, and belongs to the technical field of bridge models.The method comprises the following steps that a bridge initial finite element model is constructed based on Midas Civil NX and a natural language; optimized and corrected bridge finite element model parameters are obtained based on the bridge initial finite element model and the PSO optimization model architecture, the bridge initial finite element model is corrected, and a bridge reference finite element model capable of reflecting the real state of the structure is obtained; and importing the obtained bridge key monitoring data into the bridge reference finite element model in real time, calculating an accurate bridge structure response and verification coefficient, and carrying out rapid verification and visual display on the safety state of the bridge structure. According to the method, efficient construction of the bridge finite element model is achieved, the method is particularly suitable for large-quantity and wide-range bridge groups with similar structural forms, and technical support is provided for scheme design and safety inspection of a bridge monitoring system.

The invention discloses a miniaturized 3dB bridge building method. The method comprises the steps that S1, a traditional # imgabs0 # branch line 3dB bridge model is acquired; s2, under the constraint of meeting specific impedance and electrical length, selecting a specific loading position by loading branches with different widths, so that the distribution of the characteristic impedance and the electrical length meets a relational expression; s3, fragmenting the loaded branches, distributing the fragments into a specified design space, and obtaining design parameters of the miniaturized # imgabs1 # branch line 3dB bridge model; s4, obtaining an approximation model of the miniaturized # imgabs2 # branch line 3dB bridge model and the traditional # imgabs3 # branch line 3dB bridge model; s5, obtaining a minimum # imgabs4 # branch line 3dB bridge model; and S6, carrying out rapid construction on the minimum # imgabs5 # branch line 3dB bridge model. On the premise that the 3dB bridge designed by the invention has an extremely small structural size, the scattering coefficient index is not deteriorated relative to a classical structure.

The application discloses a kind of composite material multiscale analysis method and application considering manufacturing defect, belong to composite material performance analysis technical field.The method is by constructing microcosmic, mesoscopic and macro scale model introducing manufacturing defect;Damage model of each scale model is constructed, the damage model includes the constitutive relation of describing material linear elastic behavior, initial failure criterion judging damage initiation, and damage evolution criterion describing stiffness degradation after damage occurs;Through bridging model, the stress, strain and damage information between microcosmic and mesoscopic scale are bidirectionally transferred, and the equivalent performance of composite material is obtained;Equivalent performance is endowed to macro scale model for structure analysis.The application can be applied to performance prediction, defect tolerance evaluation and structure design optimization of composite material by multiscale coupling and defect introduction, accurately predicts the mechanical properties and failure mechanism of composite material containing manufacturing defect.

The embodiment of the specification provides a method for training a student model, a behavior prediction method and device, the method trains the student model based on a teacher model and a bridge model, the method comprises the following steps: inputting original feature data into a first embedding layer to obtain first embedding features and inputting the first embedding features into the trained teacher model and the bridge model to obtain first and second prediction results and updating the bridge model according to the first and second prediction results; inputting the first embedding features into a bridge feature extraction network included in the bridge model and a student feature extraction network included in the student model to obtain a first bridge extraction vector and a student extraction vector, and determining an interactive distillation loss according to the first bridge extraction vector and the student extraction vector; inputting the original feature data into a second embedding layer to obtain second embedding features and inputting the second embedding features into the bridge feature extraction network to obtain a second bridge extraction vector; determining an embedding distillation loss according to the first and second bridge extraction vectors; and determining a comprehensive loss according to at least the interactive distillation loss and the embedding distillation loss and updating the student model according to the comprehensive loss.