44 results about "Universal model" patented technology

The invention belongs to the technical field of crossing of remote sensing and computer vision, and particularly relates to a change general model construction method for multi-task interpretation of a remote sensing multi-temporal scene. Comprising the steps that a CNN encoder and a Vi T encoder are adopted for encoding, and a double-time-phase visual token is generated by an input double-time-phase image; the LoRA technology is adopted to introduce learnable parameters on the basis of the frozen multi-modal large model to achieve visual language alignment in the remote sensing change field, and the multi-modal large model is further used for generating text output containing multi-modal information for input double-time-phase images and text instructions; a visual decoder is added for performing high-dimensional decoding on the obtained double-time-phase visual token and the obtained text output to output a change graph; adding learnable parameters by adopting an Adapter fine tuning mode according to the constructed variable general model, and realizing a multi-task unified processing function of the variable general model; and continuously training based on the text generation loss and the segmentation mask loss, and optimizing the change general model.

The present invention belongs to the field of lifting equipment control technology, and specifically relates to a dynamic modeling method and control method for double-bridge lifting equipment. The dynamic equations of a five-degree-of-freedom double-bridge crane are constructed using the Lagrangian method, and the geometric constraint relationships between trolley I and trolley II in the direction parallel to the bridge and the vertical direction are constructed. The dynamic model of the double-bridge crane is constructed by combining the dynamic equations and the geometric constraint relationships, and the dynamic model is transformed into a general model. Based on the general model, a first error function is constructed based on the difference between the actual position and the expected position, and a virtual controller is constructed based on the first error function. A second error function is constructed based on the virtual controller, and an actual controller is constructed based on the first error function, the virtual controller, and the second error function. This technical solution models the system as a whole through the constraint relationship between the two trolleys, taking the mass of the hook into consideration during the modeling process, and solving the singularity problem.

The invention relates to the technical field of computers, and discloses an application development method based on a large model technology, and the method comprises the steps: 1, carrying out the demand analysis, and converting a fuzzy business description into an executable technical scheme through natural language understanding and a structured modeling technology; 2, carrying out architecture design and planning, and constructing a code-free and low-code dual-mode collaborative architecture; 3, establishing a visual arrangement engine, and establishing a dual-mode development interface of a code-free development mode and a low-code development mode; step 4, aiming at different requirements of a code-free scene and a low-code scene, a dynamic adaptation strategy is adopted, and differentiated resource management of lightweight adaptation and elastic expansion is carried out; and step 5, performing field adaptation, and obtaining the industry exclusive capability according to the general model and the field knowledge through code-free rapid adaptation and low-code deep customization. By means of the scheme, the development threshold can be lowered, the efficiency can be improved, and the accuracy of cross-industry application is improved.

The invention relates to the technical field of glacier monitoring, in particular to a glacier jump prediction method, system and terminal based on an intelligent algorithm, and the method comprises the steps: data collection: collecting multi-source data related to glacier jump through a sensor and a remote sensing technology, the meteorological data comprises temperature, humidity, air pressure and water content in ice, and the glacier movement data comprises flow velocity, ice layer thickness and displacement; according to the method, a simple, convenient and efficient real-time prediction scheme is constructed by adopting a multi-source data and deep learning technology for medium and small-scale glaciers, so that the problem that glacier data cannot be predicted in real time in a multi-scale physical constraint method adopted by existing glacier research is solved, and the prediction precision and timeliness are ensured; meanwhile, a simple and universal model is constructed, and the process and engineering automation deployment are simplified; in addition, disaster early warning is supported in result visualization, and the method is easy to embed into a GIS (Geographic Information System).

The invention provides a universal model heterogeneous federated learning privacy protection method, which comprises the following steps that: a cloud server randomly selects an online client, and obtains a model generation parameter union set of the client through a privacy protection union set protocol; generating heterogeneous models with different sizes and architectures according to the union set, and distributing the models to edge nodes for storage; the client downloads the local model from the corresponding edge node, and disturbs and encrypts the trained model parameters; the edge node receives the encryption model parameters uploaded by the client, and performs secure aggregation through a threshold vector aggregation protocol to obtain an aggregation model meeting the differential privacy requirement; and the cloud server receives the aggregation models from all the edge nodes, and globally updates the aggregation models by adopting a heterogeneous model aggregation method to obtain a new global model. Through an innovative privacy protection protocol and system architecture, the privacy protection problem in a model heterogeneous federated learning scene is effectively solved, and the practicability and security of federated learning are improved.

The invention discloses a universal model for voice separation and target voice extraction, which relates to the field of voice and comprises a separation network, an EDA (electronic design automation) module and a clue network. The separation network is based on a frequency band division recurrent neural network, namely a BSRNN, and is a frequency domain separation network. The EDA module is introduced into a frequency domain speech separation model BSRNN, so that the BSRNN has the capability of processing mixed speech with unknown and variable speaker numbers; and the RawNet3 is used as the clue network to realize a target voice extraction function. The model can be flexibly switched among different tasks according to different inputs, the'universal 'effect is realized, the flexibility and the universality of the model are greatly improved, the additional overhead caused by using different models for different tasks is avoided, and the cost is reduced.

A universal model-based modeling and detection method and system for defect detection in panel production lines, a device, and a storage medium. The method comprises: step 1, collecting panel production line sample data and performing defect annotation to obtain a sample data set; step 2, using the data set for modeling and training to obtain a localization model; step 3, on the basis of the data set, cropping images each containing a defect and classifying the images to form defect image data sets; step 4, using the defect image data sets for modeling and training to obtain an image classifier, and cascading the image classifier with the localization model; and step 5, on the basis of factory attributes and manufacturing process attributes, mapping defects to specific defect names by means of the image classifier and the localization model. In the method, accurate defect name mapping can be performed according to specific production environments and requirements. The overall solution reduces communication and checking time, increases sales, and also improves the applicability and accuracy of detection results.

This invention discloses a method for maintaining the service performance of machine tool feed systems based on sensitivity analysis using a twin model and a particle swarm optimization algorithm. The overall method encompasses a five-layer framework for digital twin modeling, electromechanical wear mechanism fusion and encapsulation, time-varying element sensitivity analysis, and multi-objective maintenance decision optimization. The modeling part employs an object-oriented five-layer architecture, deeply encapsulating physical mechanisms such as electromechanical coupling wear evolution based on spatial location distribution into a quasi-physical model layer, achieving accurate description and real-time updates of implicit time-varying elements such as localized guideway wear. The decision-making part uses the Sobol sensitivity analysis method to quantify the impact weight of each element on service performance, constructs a full-effect exponential-driven adaptive tightening mechanism for early warning thresholds, and combines it with a particle swarm optimization algorithm (PI-PSO) that introduces degradation inertia and physical boundary penalties to iteratively calculate the optimal maintenance scheme with the goal of minimizing maintenance costs, downtime losses, and performance degradation. This method not only solves the problem that general models are difficult to describe the underlying electromechanical-dynamic coupling mechanism but also achieves deep integration of optimization algorithms and physical degradation laws, significantly improving the service reliability and economic benefits of machine tool feed systems.

The application provides a UVM verification platform reference model hardware architecture supporting multistage dynamic configuration, belongs to the technical field of UVM verification platforms, and comprises the following steps: S1, a hierarchical reference model construction step, which comprises a model splitting link and a cascaded connection link; S2, a dynamic configuration mechanism implementation step, which comprises a centralized configuration link, a tag distribution link and a real-time reconfiguration link; and S3, a hierarchical independence processing step, which comprises a universal model template link and a differential injection link. The traditional verification platform is generally a single-stage reference model, and for some hardware designs with multistage similar structures, directly integrating all the structures leads to the problem of low efficiency. The application constructs a verification platform with multistage reference models through a multistage hardware design structure, and solves the problems of code reusability and low efficiency of the traditional platform.

The present disclosure relates to a large-scale model hardware and software integrated machine tuning method and device for rapid deployment, which can improve the overall deployment efficiency of the large-scale model hardware and software integrated machine. The method includes: in response to a user powering on the large-scale model hardware and software integrated machine, obtaining hardware information of each hardware module, and generating a first deployment plan based on the hardware information of each hardware module and preset general model requirements through an artificial intelligence model; in response to a user configuring the business requirements of the large-scale model hardware and software integrated machine in a visual interface, determining a target large model from multiple pre-trained large models built into the large-scale model hardware and software integrated machine through an artificial intelligence model according to the business requirements, and adjusting the first deployment plan according to the hardware conditions required by the target large model to obtain a second deployment plan for the large-scale model hardware and software integrated machine; and deploying the target large model in the large-scale model hardware and software integrated machine based on the second deployment plan.

The application discloses an online roll temperature and thermal crown prediction method based on finite difference and polynomial regression, aiming at the technical pain point that the accuracy of the traditional analytical model is insufficient and numerical simulation cannot meet the online real-time calculation demand, a plurality of groups of working conditions are designed through an orthogonal experiment method, and a finite difference method is used to calculate roll steady-state temperature field data to construct a training data set, based on the data set, a polynomial regression algorithm with elastic network regularization is used to fit an axial temperature distribution general model with the roll temperature field as an input variable, after obtaining the online surface axial temperature distribution of the roll, the general model is substituted into the internal temperature field to restore the internal temperature field and calculate the radial thermal expansion amount at each position, so that compensation information is provided for a shape control system of a rolling mill. The application simplifies complex physical field simulation into lightweight algebraic formula operation, retains the high analytical accuracy of the finite difference method, realizes real-time prediction, and can significantly improve the rolling quality of silicon steel products.

The invention discloses a construction method of an oil and gas production large model platform, and relates to the technical field of oil and gas production intellectualization, and the method comprises the following steps: pre-constructing an oil and gas field knowledge graph, employing a BERT-BiLSTM-CRF model to carry out the automatic extraction of entities and attributes of massive unstructured engineering documents in oil and gas production, the extraction result is used for filling and perfecting the knowledge graph; a WellGPT model focusing on professional knowledge reasoning and a WellGPT-VL model focusing on view fusion understanding are constructed respectively. According to the method, on the basis of Qwen2.5 double bases and the dynamic weight hybrid LoRA technology based on intention recognition, efficient fine adjustment of a universal base model is achieved through precise classification of an intention recognition scheduling center and special training of an independent LoRA adapter, the generalization ability of the universal model is reserved, professional knowledge of the oil and gas industry is deeply injected, and the method has the advantages of being high in universality, high in practicability and high in practicability. And professional terms, physical laws and multi-modal data characteristics in the field can be accurately understood, and the professional adaptability and output reliability of the model in an oil and gas production scene are remarkably improved.

The invention belongs to the technical field of wheat plant modeling, and discloses a wheat plant tillering geometric model general modeling method based on a discrete element method.The wheat plant tillering geometric model general modeling method comprises the steps that statistical analysis is conducted on shape parameters of wheat tillering plants, and single tillering modeling parameters and root system modeling parameters are obtained; constructing a single tillering model framework which comprises a stalk model and a wheat ear model; calculating expressions of spherical coordinates of the stalk model and the wheat ear model according to the single tillering modeling parameters to obtain a single tillering general model; taking one single-tillering general model as a main stem model, taking a plurality of single-tillering general models as tillering plant models, and combining to obtain a plant general model; constructing a root system model framework which comprises a tillering joint model, a primary rooting model and a plurality of secondary rooting models; calculating expressions of spherical coordinates of the tillering joint model, the primary rooting model and the secondary rooting model according to the root system modeling parameters to obtain a root system universal model; and combining the root system general model with the plant general model to obtain a wheat plant tillering geometric general model.

The invention discloses a large language model equipment relation extraction method based on two-way retrieval and context learning, and relates to the technical field of relation extraction in natural language processing. The large language model equipment relation extraction method based on two-way retrieval and context learning comprises the following steps: S1, fine adjustment of an embedded model; s2, constructing a retrieval example library; s3, performing double-path retrieval and correlation sorting; and S4, prompt word construction and relation extraction. According to the large language model equipment relation extraction method based on two-way retrieval and context learning, global semantics and local instance relations are captured at the same time through two-way retrieval, three types of overlapping relations of Normal, SEO and EPO in the equipment field can be effectively recognized, and an F1 value is remarkably improved compared with an existing optimal baseline model GPT-RE; through comparison learning and a BGE-M3 model optimized by a domain adaptation layer, semantic information of technical terms and technical description in the field of equipment can be accurately captured, and the problem of insufficient domain adaptation of a general model is solved.

The invention provides a model reasoning method, device and system based on a storage and calculation integrated hybrid calculation architecture, and the method comprises the steps: carrying out the prediction of to-be-processed target domain data on a simulation storage and calculation integrated unit based on a deployed universal model, and determining a prediction result; under the condition that the prediction result does not meet the preset condition, retrieving the to-be-processed target domain data based on a deployed database on the digital storage and calculation integrated unit, and determining a retrieval result; and determining a reasoning result corresponding to the to-be-processed target domain data according to the prediction result and the retrieval result. According to the method, hybrid reasoning of the prediction result of the universal model and the retrieval result of the database can be realized, the reasoning precision of the universal model for personalized data of the user is improved, storage and calculation are integrated, and the reasoning speed and energy efficiency are improved by reducing the data carrying frequency.

The invention provides a transformer substation GIM model intelligent retrieval and BIM general format conversion method, relates to the technical field of transformer substation data processing, and solves the problem that an existing transformer substation GIM model is difficult to realize cross-modal accurate retrieval, high-fidelity BIM format conversion and dynamic standard knowledge fusion. The method comprises the steps of firstly constructing a power field knowledge graph and extracting key features in user multi-modal query intention data, then understanding the key features by a multi-modal retrieval model through a mode of combining a graph neural network and a vision-language pre-training model, and carrying out joint reasoning in the power field knowledge graph; a target GIM model is retrieved; the geometric topology of the target GIM model is reconstructed, professional attributes of the target GIM model are mapped, then a generative adversarial network is adopted to convert material codes into PBR materials, and a corresponding general BIM model is generated together; and after compliance verification and optimization of the general BIM model are completed, the model can be output to realize business application.

The invention relates to the technical field of artificial intelligence model servitization, and particularly discloses a universal model reasoning service framework supporting multi-type heterogeneous algorithm packaging. According to the framework, by designing a unified algorithm adaptation layer, a standardized model management module, a universal inference engine and a standardized interface interaction module, unified packaging and management of different types of heterogeneous algorithms such as Sciit-learning (sklearning for short), PyTorch, YOLO and the like are achieved. An external system can call any model reasoning service in the framework through a standardized API interface without paying attention to implementation details of an underlying algorithm, so that the development cost of multi-algorithm collaborative application is greatly reduced, and the efficiency and flexibility of model servitization deployment are improved. The method is suitable for various scenes requiring integration of multiple artificial intelligence algorithms, and has a wide application prospect.

The invention discloses a gravity compensation method and equipment of force feedback equipment, a medium and a product, and belongs to the technical field of robots. According to the gravity compensation method of the force feedback equipment, a universal model skeleton diagram of the force feedback equipment is constructed, standardized joint characteristic parameters are collected, an accurate dynamic model is established according to the parameters, and the force feedback equipment is obtained by combining set universality constraint conditions. According to the method, the optimal spring parameters adaptive to master manipulators of different structure types can be automatically calculated by utilizing an iterative optimization algorithm, so that smooth and stable gravity balance is realized in the whole working space of various force feedback equipment, and the operation portability and control precision of the force feedback equipment are remarkably improved.

The invention relates to the technical field of micro-energy collection, particularly discloses a surface-mounted magnetic field energy collection system modeling method considering magnetic core structures and materials, and aims to solve the problems that an existing model is insufficient in universality and applicability, and particularly, system design under different magnetic core structures and materials is lack of theoretical guidance. General modeling is realized through four core steps: establishing an equivalent circuit model of a system; constructing a general mutual inductance model considering the structure and the material of the magnetic core based on a magnetic core shape equivalent method; establishing a system loss model including coil loss, magnetic core loss and diode conduction voltage drop loss; and combining the models to form a complete magnetoelectric conversion general model. According to the method, shape coefficient correction mutual inductance calculation is introduced, material characteristics and circuit loss are comprehensively considered, the adaptability and accuracy of the model in different application scenes are remarkably improved, and a reliable theoretical basis and design guidance are provided for optimization design of the surface-mounted magnetic field energy collection system.

A Life-cycle Component Management (LCM) system associated with Kubernetes can employ a universal operator to aid in the management of any software system, no matter the size or complexity. Specifically, the LCM system includes a universal operator, which eliminates the need for a human software developer to implement a set of workflows. The universal operator dynamically computes and runs the disclosed workflow set, based upon the description of the current state of the system (i.e., the “begin state”) and the new desired state of the system (i.e., the “end state”). Thereby, this universal model enables LCM for large Kubernetes software systems, which require excessive workflow sets due to the intrinsic complexity of large software systems and the plurality of vendors that contribute to such systems.

Systems, methods, apparatuses and non-transitory computer executable media configured to unify preprocessing, configuration, training, monitoring, and evaluation of multiple neural network based object detection algorithms under a singular development environment / platform (i.e., a “unified training platform”). The unified training platform may include a neural network agnostic model training environment that addresses the deficiencies described above and may allow for unified data annotation formatting. In addition to incorporating a wide variety of state-of-the-art neural networks into the unified training platform, the unified training platform may also provide full accessibility to available network optimizations. The present disclosure may also include a universal model converter.

The invention relates to the technical field of medical instruments, in particular to an osteotomy distraction stabilizer distraction angle calibration system based on medical big data, and the system analyzes the skeleton creep law of a current patient based on the data of the current patient through a universal model generation module, and generates a digital twin universal model; a practical model generation module is used for carrying out data assimilation digital twinning general model calibration, a personalized digital twinning practical model capable of simulating the skeleton creep behavior of the current patient is generated, the unique biological dynamic characteristics of the skeleton of the patient can be accurately reproduced, and then the accurate comparison of the individual conditions of the current patient and historical patients is realized; finally, the deviation between the distraction angle of the current patient and the optimal distraction angle interval is evaluated through accurate comparison, the correction angle is dynamically calculated, then self-adaptive closed-loop control over the distraction angle is achieved, and system stability is maintained.

This invention relates to the field of multi-axis CNC machining and industrial robot machining technology, specifically disclosing an asymmetric five-axis toolpath smoothing method based on path weighted synthesis. The method includes: constructing a universal corner smoothing model based on path weighting; constructing an analytical expression for a high-order continuous weighted function; Cartesian space smoothing of the tool tip position path; unit spherical space smoothing of the tool axis direction path; synchronous adjustment of C3 continuous parameters for position and direction paths; interference verification and asymmetric adjustment of adjacent corner transition intervals; continuous speed planning and real-time interpolation; and motion control of the machining equipment. This invention overcomes the shortcomings of existing five-axis path transition smoothing technologies, such as poor geometric versatility, incoordination between tool tip and tool axis movements, easy interference in short path segments, and inconsistent bidirectional machining trajectories. It improves geometric versatility and processing efficiency, achieves high-order continuity and consistent machining surfaces, and can be ported across platforms and extended to higher orders.

An aircraft electrical structure comprises equipment items that are interconnected by cables. A processing system constructs (400) a digital representation of the electrical structure in the form of nested graphs. A first graph represents the electrical structure by first arcs representing the cables and first nodes representing the equipment items. A second graph represents an internal interconnection structure between at least one input and at least one output of each equipment item which has such an internal interconnection structure, second arcs representing interconnections between inputs and outputs, second nodes representing the inputs and the outputs concerned. The first graph is obtained from contextual data (411) describing the electrical structure and each second graph is obtained from a generic model (412) for each type of equipment item concerned. An analysis is carried out by performing a search (404) for equipotentials by browsing through the nested graphs. Thus, the analysis is facilitated.

The invention relates to the technical field of aerospace, in particular to a landing gear dynamic characteristic simulation generalization model construction method based on a virtual buffer, which comprises the following steps: defining an airframe coordinate system and an undercarriage coordinate system, and determining a connecting point of an undercarriage and an airframe and coordinates of each key node; based on the connecting points and the key nodes, an undercarriage system connected by an equivalent buffer is constructed, the undercarriage system is subjected to modularization processing, and the undercarriage system comprises an undercarriage inelastic mass module, an oil buffer module, an elastic tire module and a connecting rod module; carrying out vectorization modeling on each module, solving the force and moment of force of the corresponding module, and constructing a Simulink dynamics simulation model. According to the method, the strut type landing gear configuration and the rocker arm type landing gear configuration are unified through the virtual buffer, and a dynamic simulation model is established through modular modeling and Simulinnk.

The invention relates to the technical field of power grid harmonic source modeling, in particular to a harmonic source uncertainty general model building method and device, and the method comprises the steps: sampling and collecting harmonic data at a harmonic source port, and carrying out the normalization processing of a harmonic data set; establishing a harmonic source deterministic general model based on XGBoost, and obtaining a prediction error of the verification set; fitting prediction error distribution of the verification set by using a KDE method, and calculating upper and lower bounds of a prediction error interval under a specified confidence level through a cumulative distribution function; inputting the test set into an established XGBoost-based harmonic source deterministic universal model, obtaining a predicted value of harmonic current, and calculating a harmonic current prediction interval result in combination with a prediction error interval; and evaluating the interval prediction result of the harmonic source uncertainty general model based on the XGBoost-KDE. The method can describe the uncertainty of the harmonic characteristics of the harmonic source, and provides powerful support for power grid harmonic analysis and treatment.

In order to reduce the time of model training and increase the possibility of large-scale deployment of traffic light control, the present invention provides a general model system for traffic light control, including a feature extraction module and an action prediction module. The feature extraction module is used to convert the traffic flow information of the intersection into an intersection information feature vector, wherein the traffic flow information includes the average traffic flow under each traffic flow signal, the maximum lane occupancy within K seconds, the average occupancy, the direction of movement, the number of lanes included, the duration of the movement, whether the duration is the minimum green light time, and whether the current signal is green. The action prediction module is used to predict the duration of the current traffic light state of the intersection based on the intersection information feature vector. Using "setting the current phase duration" as an action not only retains the concept of the traffic cycle, but also ensures the safety of traffic participants such as drivers and pedestrians.

The invention discloses an electrocardiosignal personalized classification method based on quantum features and meta learning and a medium. The electrocardiosignal personalized classification method comprises the steps that S1, electrocardiosignals are preprocessed; s2, extracting electrocardiosignals by adopting quantum global features to obtain a first feature vector; meanwhile, extracting local waveform features of the electrocardiosignal to obtain a second feature vector; s3, channel splicing is carried out on the first feature vector and the second feature vector, normalization processing is carried out, and unified double-flow fusion feature representation is formed; s4, on the basis of the model-independent element learning framework, performing element training on the electrocardiogram analysis model by using the double-current fusion features to obtain an electrocardiogram analysis universal model; s5, acquiring electrocardiogram data of the target user to obtain personalized double-current fusion features of the target user; and inputting the personalized double-current fusion features into the electrocardio analysis general model, and finally generating a personalized classification model. According to the method, the problems of low adaptation speed and large individual difference of the electrocardio model under the small sample condition are solved, and the accuracy and the individuation level are improved.

Systems, methods, apparatuses and non-transitory computer executable media configured to unify preprocessing, configuration, training, monitoring, and evaluation of multiple neural network based object detection algorithms under a singular development environment / platform (i.e., a “unified training platform”). The unified training platform may include a neural network agnostic model training environment that may allow for unified data annotation formatting. In addition to incorporating a wide variety of state-of-the-art neural networks into the unified training platform, the unified training platform may also provide full accessibility to available network optimizations. The unified training platform may also include a universal model converter.