69 results about "Real time model" patented technology

The invention discloses a traffic image labeling method and device based on a D-S evidence theory and a medium, and relates to the technical field of image labeling. According to the method, two different types of real-time models are simultaneously utilized to process the same image in parallel, and two uncertainty detection results of the same image are obtained; a KM algorithm is utilized to match targets in two uncertainty detection results, then a prior credibility score is utilized to reduce a conflict coefficient, and two different types of small model prediction results are combined through an uncertainty theory. According to the traffic image labeling method and device based on the D-S evidence theory and the medium, the strong generalization ability of the open set model is utilized to make up for the defects of a closed set model, the use of a large model to improve the hardware dependence is avoided, the precision problem of a single small model is made up under the condition that the efficiency is ensured, and the iteration duration of the model is shortened.

The invention relates to the technical field of new energy automobile batteries, in particular to a new energy automobile battery state prediction method, optimization control method and system, and the method comprises the steps: collecting the real-time operation data of a new energy automobile battery; uploading the real-time operation data to a cloud processing platform in which historical operation data of the battery is stored, and dynamically predicting the state of the battery; fusing the charge state data driven by the real-time model and the charge state data driven by the historical data to carry out charge state estimation prediction, and outputting charge state estimation parameters; a circulation attenuation model is combined with reinforcement learning correction, historical circulation data is used for fitting a capacity attenuation curve, and a dynamic battery health state estimation value is output; a fusion thermal model and an internal resistance model are adopted, a safety voltage / temperature boundary is predicted through digital twin simulation, and a dynamic battery power state estimation value is output; and generating a battery control strategy based on the battery state parameters and issuing the battery control strategy to an onboard BMS module for execution.

The invention relates to the technical field of artificial intelligence, can be applied to business scenes of financial science and technology, medical health and the like, and discloses a data processing and decision-making method, device, equipment and medium, and the method comprises the steps: obtaining multi-dimensional source data, carrying out the standardization processing, screening the standardization data based on a decision-making scene, and obtaining a core index set, the method comprises the following steps: establishing and optimizing indexes, mining association characteristics among the indexes, constructing and optimizing a dynamic statistical model, carrying out standardization processing on newly added multi-dimensional source data, obtaining a key decision index value by utilizing the dynamic statistical model, and carrying out comparison judgment on the key decision index value and a preset decision threshold value, and generating a decision scheme by combining the comparison judgment result with a preset scheme generation mechanism. According to the method, multi-scene adaptive support and real-time decision optimization are realized through unified multi-dimensional data standardization, dynamic scene-driven core index screening and association characteristic mining, real-time model updating and closed-loop decision scheme generation.

A structure inspection system includes an image conversion device located in a far edge of the structure inspection system that is proximate to a physical structure, and a near edge processor located in a near edge of the structure inspection system that is remote from the physical structure. The image conversion device receives an image of the physical structure, and renders a real time 3D model of the physical structure based on the first images. The near edge processor is configured to provide a baseline 3D model of the physical structure to the image conversion device. The image conversion device further compares the first real time 3D model with the baseline 3D model, and provides an indication when the first real time 3D model deviates from the baseline 3D model.

The application belongs to the technical field of vehicle software development, and particularly relates to a method for realizing hard real-time of a non-real-time system in hardware-in-the-loop simulation and electronic equipment, wherein the method for realizing hard real-time of the non-real-time system in the hardware-in-the-loop simulation comprises the following steps: creating a real-time model task and forming a task program; the task program automatically reads computer configuration and judges the number Z of processor cores of a current computer device; when the number Z of the processor cores does not exceed X, the number n of threads is set to 1 to execute the real-time model task, otherwise the number n of threads is set to (Z-X) / Y to execute the real-time model task, wherein Y represents the number of threads of one physical core of the processor.

The application discloses a kind of digital twin and stress prediction method and device of large component additive manufacturing process, and time sequence image is obtained by scanning in time sequence by slidable line laser head, corresponding three-dimensional real-time model is constructed and surface displacement difference in molten pool formation to cooling solidification process is obtained;It is input to three-dimensional real-time model as surface displacement load and is carried out numerical simulation, temperature field, stress field are obtained by numerical simulation, and the temperature measured by infrared thermal imager, the stress measured by stress sensor and the temperature, stress obtained by numerical simulation are compared, the data obtained are input into three-dimensional real-time model, and the digital twin of real-time molten pool and deposition layer is formed.The application can well reduce simulation time, improve the efficiency of printing process numerical simulation, and effectively monitor and predict the deformation and cracks generated in large component printing process.

A simulation modeling method and system for a real-time model of a silicon carbide IGBT module, which relate to the technical field of power electronics and are used for a silicon carbide IGBT module. On the basis of acquired module characteristic parameters and oscillation characteristic parameters, by means of a targeted data acquisition method, key information can be quickly and accurately provided for modeling, thereby greatly reducing unnecessary computational amount, and significantly improving the modeling efficiency. On the basis of a constructed initial real-time model, real-time oscillation simulation is performed by using the acquired oscillation characteristic parameters and module characteristic parameters, such that the oscillation characteristics at the moment of turning on / off a silicon carbide IGBT module in actual operation can be simulated more accurately; oscillation simulation data is compared to preset standard working condition data, if the two are consistent, the initial real-time model can serve as a target real-time model, thereby ensuring the accuracy of the model.

The invention relates to a QEC detection method and system based on AR. The method comprises the steps that a pre-constructed engine standard model is disassembled into part units, a three-dimensional model set in the installation process of all the part units is obtained, and the three-dimensional model obtained after installation of each part unit is finished is a part unit standard model; obtaining an engine real-time model based on AR equipment; when the difference degree between the engine real-time model and the standard model is higher than a first threshold value, a first part unit with the largest difference degree is obtained, whether the corresponding real-time model is matched with the three-dimensional model set or not is judged, if yes, the first part unit is maintained towards the corresponding standard model state, and if not, the first part unit is maintained; and when all the part units in contact with the first part unit are synchronously changed to the corresponding standard model states, the engine real-time model is obtained again to continue execution, and otherwise, a QEC request is sent out. Compared with the prior art, the method has the advantages that the safety and the efficiency of the whole QEC operation inspection and verification process are improved, and the like.

This invention relates to a lightweight adaptive method for testing non-stationary data streams. Its core lies in constructing a two-branch adaptive architecture that decouples observation and execution statistics. A shadow observation branch, independent of the main inference branch, maintains a statistical reference baseline in real time, unaffected by adaptive calibration actions, thus suppressing the closed-loop coupling problem between drift perception and statistical updates at the mechanism level. Based on this decoupled architecture, the shadow branch calculates a composite drift surrogate metric to map and generate dynamic momentum parameters, and introduces relaxed cumulative variables to perform three-level state determination and spatiotemporal misalignment mask generation. Finally, during inference forward propagation, dynamic online calibration of statistics is performed only for the key levels selected by the mask. This invention alleviates the adaptation lag and statistical oscillation risks caused by fixed momentum, suppresses the statistical degradation risk caused by single-sample noise, and is suitable for real-time model optimization in single-sample, non-stationary streaming scenarios.

The invention relates to the technical field of BIM-based construction management, in particular to a BIM-based construction progress control management method and system, and solves the technical problem of relatively large prediction deviation in BIM-based construction progress management establishment in the prior art. Factors occurring in the construction process are selected through field investigation, the influence degree is analyzed, the actual progress is extracted according to a 4D real-time model established through BIM, the influence coefficient is calculated through a regression model, and the predicted construction period and the actual construction period are compared to verify the effectiveness of the model; in addition, according to the technical scheme, the BIM technology is applied to the construction process, project actual conditions can be visually inquired through progress visual management based on a real-time 3D model, and progress deviation is analyzed; and meanwhile, the predicted construction period is used as a time parameter to be linked with the BIM three-dimensional model, so that the dynamic management of the subsequent project progress is realized.

The application discloses a kind of based on neural network and genetic algorithm's thermal power unit working condition dynamic optimization method, system, equipment and storage medium, it is related to thermal power unit intelligent dynamic optimization technical field, including determining fitness function parameter, constructs BP neural network model;Establish the boundary line equation of parameter to be optimized, the upper and lower boundary lines of parameter to be optimized under different loads are fitted out;Using genetic algorithm receives BP neural network model and boundary line equation, publishes working condition optimization AI model, configures non-optimized parameter;Boiler efficiency working condition optimization real-time model is constructed, and based on non-optimized parameter real-time optimization is carried out to parameter to be optimized, obtains efficiency deviation and corresponding parameter to be optimized deviation;The method described in the application realizes thermal power unit working condition dynamic optimization, parameter to be optimized reasonable boundary constraint and efficiency deviation and parameter deviation linkage output, improves the pertinence, coherence and executability of optimization result under complex load condition.

The invention belongs to the technical field of vehicle software development, and particularly relates to a system and method for a non-real-time system to achieve hard real-time in hardware-in-loop simulation, and the system comprises a computer device, a hardware-in-loop simulation device and a hardware-in-loop simulation device. The computer device is configured to execute a task creating module and a task thread setting module; wherein the task creating module is configured to create a real-time model task and form a task program; the task thread setting module is configured to enable a task program to automatically read computer configuration and judge the number Z of processor cores of the current computer device, when the number Z of the cores does not exceed X, the thread number n is set to be equal to 1 so as to execute a real-time model task, otherwise, the thread number n is set to be equal to (Z-X) / Y so as to execute the real-time model task, wherein Y represents the number of threads of one physical core of the processor.

The invention discloses a microgrid voltage and frequency stable adjustment method and system based on a distributed integral model, a medium and a processor, and relates to the technical field of microgrid control. The method comprises the following steps: independently identifying system parameters and states on line through each distributed energy node, and constructing a linear time-varying model; an integral effect is introduced and converted into an incremental state space model containing an integral item, an IIMPC framework is formed by combining control variable constraints, a control target is converted into a secondary optimization problem with constraints, a distributed energy reference power signal is solved, and a master controller and a slave controller cooperatively control dynamic load distribution according to the distributed energy reference power signal; and finally, comparing the real-time model in-loop simulation with the optimal PI controller to evaluate the performance. The method improves the control precision and robustness, and solves the problems of low response speed and poor flexibility in the prior art.

The invention relates to a QEC (Quality Error Correction) maintenance safety supervision method and system based on AR (Augmented Reality). According to the method, after a maintenance general task is generated, the maintenance general task is decomposed into a plurality of sub-tasks and sent to pre-obtained online AR equipment, and a maintenance piece where each sub-task is located and a target model are obtained; obtaining an associated sub-task set according to the maintenance piece associated sub-tasks; when the current sub-task is executed, acquiring a marked sub-task from the associated sub-task set of the current sub-task; acquiring a real-time model of a maintenance piece where the marked sub-task is located, and executing the next sub-task if the real-time model meets a first preset condition before and after the current sub-task is executed; if the real-time models of the maintenance pieces where all the sub-tasks are located in the current associated sub-task set meet a second preset condition, the associated sub-task set is qualified; and when the completion time of all the sub-tasks and the total task real-time model meet a third preset condition, the maintenance total task is qualified. Compared with the prior art, the method has the advantages of improving data analysis efficiency and accuracy in QEC maintenance work and the like.

This specification discloses a fragmented intelligent agent system and its processing method based on multi-model tentacle-like collaboration. The intelligent agent system, by setting a control central module, decomposes the original task context into multiple context fragments, and distributes these fragments to multiple isolated real-time model modules for processing, ensuring that no single real-time model module can obtain the complete task context during processing. The system further uses a fragment reconstruction module to verify and reconstruct the consistency of the processing results from multiple real-time model modules, thereby generating the final decision result. This specification also provides a detailed description of the intelligent agent system's structural composition, context fragmentation method, multi-model collaboration mechanism, processing result reconstruction process, and compatible operating modes, aiming to improve security, reliability, and scalability while ensuring processing power.

The invention belongs to the technical field of boiler furnace temperature field monitoring, and particularly relates to a boiler furnace three-dimensional temperature field monitoring method integrating physical measurement and online simulation, which comprises the following steps: S0, acquiring sound wave temperature measurement data and DCS data; the method comprises the following steps: S1, constructing a CFD high-precision mechanism model of hearth combustion reaction flow; s2, performing simulation calculation by using the model in the S1 to obtain basic characteristic data of the hearth temperature field; s3, combining the basic characteristic data in the S2 with the data in the S0 to construct a sample data set; s4, constructing an online real-time model based on the sample data set; s5, dynamic information of the three-dimensional temperature field of the hearth is obtained through calculation based on the online real-time model; and S6, monitoring a model calculation error in the step S4 in combination with sound wave temperature measurement data, and correcting and updating the model when the error exceeds a set threshold value. According to the method, a small number of sound wave temperature measuring points are combined, an online real-time model capable of accurately reflecting combustion dynamic characteristics and giving consideration to real-time performance is constructed, and real-time monitoring of the three-dimensional temperature field in the furnace is achieved.

The invention discloses a regional cathode protection potential balance algorithm. The algorithm specifically comprises the following steps: S1, real-time data acquisition; s2, real-time data preprocessing; s3, performing real-time state evaluation and adjustment object matching; s4, establishing and updating a real-time response model; s5, real-time adjusting quantity calculation and safety control are carried out; and S6, performing adjustment and iteration termination in real time. The method has the advantages that the real-time response speed is increased, the traditional hour-level adjusting period is shortened to the second-level adjusting period, adjustment is started after potential unbalance is carried out for 30 seconds or less, and the risk of short-term corrosion of a pipeline is avoided; due to the real-time model and coupling correction, the potential range is smaller than or equal to 0.08 V, the standard deviation is smaller than or equal to 0.03 V, and the method is superior to a traditional algorithm; the characteristics of a real-time acquisition control type intelligent potentiostat and a real-time data acquisition intelligent test pile are completely matched, and the real-time advantage of the equipment is exerted; and multi-dimensional real-time verification (step length, coupling, current and potential rate) is realized, and the risk of overshoot or equipment overload is avoided.

The invention belongs to the technical field of vehicle software development, and particularly relates to a method for realizing hard real-time by a non-real-time system in hardware-in-loop simulation and electronic equipment, and the method for realizing hard real-time by the non-real-time system in hardware-in-loop simulation comprises the following steps: creating a real-time model task, and forming a task program; the task program automatically reads computer configuration and judges the number Z of processor cores of the current computer device, when the number Z of the cores does not exceed X, the number n of threads is set to be equal to 1 to execute the real-time model task, otherwise, the number n of the threads is set to be equal to (Z-X) / Y to execute the real-time model task, and Y represents the number of the threads of one physical core of the processor.

The invention discloses a fully mechanized coal mining face roof broken area identification and hydraulic support supporting parameter optimization system and optimization method, and the system comprises an edge calculation unit which is used for collecting roof state data in real time through operating a lightweight online real-time model and a sensor disposed on a hydraulic support, performing feature extraction on the collected original data to obtain key feature vectors representing the state of the top plate, and calculating and outputting real-time support parameters of the hydraulic support through an online real-time model based on the key feature vectors; the cloud server is used for operating an off-line optimization model, is connected with the edge calculation unit through a data transmission channel, receives the original data, the key feature vectors and the real-time support parameters, and drives the off-line optimization model to perform iterative training by using the received data; and the network guarantee layer is used for guaranteeing the certainty and redundancy of communication between the edge computing unit and the cloud server. According to the invention, the support precision, the dynamic response capability and the intelligent level are improved.

The invention belongs to the technical field of aero-engine modeling, and particularly relates to an engine model solving method adopting Newton-Raphson and feedforward neural network methods. According to the method, a mixed algorithm of a feed-forward neural network agent model and a Newton-Raphson method is combined, and a rough value of an iteration independent variable is directly predicted by adopting a feed-forward neural network, so that meaningless and local extremal regions on an iteration path can be avoided to a great extent, and high-quality iteration variables are quickly provided for the Newton-Raphson method; the convergence capability of the Newton method is greatly improved, the iteration step number is reduced, and the calculation speed of the Newton method is improved. And a normalized training set processing mode is adopted, so that the training set scale and the network scale are greatly reduced, the network estimation precision is improved, and the application complexity of the hybrid solving method is reduced. According to the hybrid solving method, the simulation solving convergence of the violent dynamic process of the dynamic real-time model of the engine is remarkably improved, and compared with a gradient method, the calculation speed is increased.

The application discloses a kind of intelligent sports movement state analysis method, system, equipment and medium based on multi-module collaborative optimization.Method includes: using lightweight real-time model to video stream is human detection and initial key point positioning;Through radial basis function network, initial key point is nonlinear fine-tuning to improve accuracy;From the key point sequence after fine-tuning, extract high-dimensional kinematic feature, and utilize principal component analysis to reduce dimension, obtain core low-dimensional time sequence feature;Finally, through time sequence neural network, feature sequence is modeled, realizes action recognition and posture evaluation.The application passes through the collaborative optimization link of "perception-fine-purification-understanding", effectively solves the problems, such as imbalance of accuracy and speed, feature redundancy, time sequence modeling deficiency and error accumulation, in prior art, provides efficient, accurate, automated analysis solution for wisdom sports training.

A method to estimate a remaining of useful life of a DC link capacitor of a converter through determination of a minimum value Clim of its capacitance C under which performances of the converter are considered as insufficient. The method includes implementing a real time model including a motor drive model and a capacitor model using decreasing capacitor values to calculate a forecasted minimum capacitance Clim sufficient for keeping at least one of DC voltage ripple, motor current ripple, speed ripple or torque ripple under a specified value. The method further includes calculating a time difference between a time Tactual, for which the capacitor has an estimated capacitance based on an ageing curve Ccurve, and a time Tlim for which the capacitor will have the forecasted minimum capacitance Clim, with the use of an ageing curve Ccurve of the capacitor and estimating the remaining useful life of the capacitor Tactual−Tlim.

The application discloses a civil aviation tractor-oriented adaptive model predictive control method and system based on fusion perception, and belongs to the technical field of automatic driving and vehicle motion control. The application identifies load parameters such as the number, mass and geometric size of a baggage compartment through fusion perception of a vehicle-mounted laser radar and a visual sensor, dynamically constructs an adaptive variable-structure multi-body dynamics model, adaptively adjusts the weight and constraint boundary of a model predictive controller based on the real-time model, solves an optimal control sequence by rolling and executes the optimal control sequence through a drive-by-wire chassis, and simultaneously realizes safety monitoring through closed-loop feedback of a vehicle motion state. The application can adapt to 1-10 baggage compartment dynamic load working conditions, solves problems such as mismatch of a traditional fixed model, low trajectory tracking precision, high risk of folding and collision and the like, realizes high-precision trajectory tracking and all-condition safety control, adapts to a civil aviation apron operation scene, and has high safety, strong adaptability and good engineering landing value.

The invention relates to the field of artificial intelligence, and discloses a dynamic model detection method based on an AI algorithm, and the method comprises the steps: selecting at least one monitoring layer in an AI model; calculating a model state entropy of the monitoring layer based on the calibration data set, generating a state target gradient disturbance vector by calculating a gradient of the model state entropy relative to an input sample, and further calculating disturbance sensitivity of the monitoring layer so as to respectively establish a state entropy baseline and a disturbance sensitivity baseline; when a plurality of monitoring layers exist, establishing a cross-layer transmission consistency baseline; obtaining an online real-time input sample, and calculating the real-time model state entropy and the real-time disturbance sensitivity of the online real-time input sample; comparing the real-time indicator with the baseline to generate a single point anomaly score and a cross-layer consistency deviation; and determining the current health state of the AI model based on the score and the deviation degree. According to the method, the internal state of the model can be monitored, potential risks are warned in advance, and fault diagnosis is provided.

The invention discloses a lambda self-adaptive control method and system based on a DPF carbon-loaded model. The method comprises the steps that a real-time mileage model value of a vehicle, a real-time flow resistance model value of a particle catcher and a real-time original exhaust model value of an engine are obtained; the real-time flow resistance model value is compared with the real-time original discharge model value, the maximum value is obtained, and a real-time DPF carbon-loaded model value is output; outputting a real-time model deviation value according to the real-time mileage model value and the real-time DPF carbon-loaded model value; outputting a DPF carbon load model deviation coefficient value according to the real-time model deviation value and a DPF maximum allowable carbon load value of the particle catcher; the rotating speed of the engine and the deviation coefficient value of the DPF carbon load model serve as the x axis, the y axis and the z axis of the DPF carbon load model lambda correction MAP as lambda correction coefficients; and inputting the lambda correction coefficient of the DPF carbon-loaded model into the smoking limitation lambda main strategy for calculation and output. The smoking limit lambda coefficient is subjected to self-adaptive correction based on the DPF carbon-loaded model, and the transient dynamic property of the vehicle can be improved in the using process of the vehicle.

The invention provides a vehicle control method, a vehicle and a storage medium, and relates to the technical field of vehicle control. In the method, based on a bird's-eye view angle image, the current driving of the vehicle in a rare road surface area (generally a high-risk road surface area) is identified, and vehicle control resources can be focused on the road surface area where accidents occur frequently. A historical experience library is started in a rare road surface area, and a verified successful torque strategy under similar road conditions is provided for initial torque control information to serve as a reference. Initial torque control information can be determined through a real-time model, which can effectively make up for the judgment blind area of the real-time model to the unseen road features. And adjusting the initial torque control information based on the reference torque control information to generate target torque control information which responds to the instantaneous state and inherits historical success experience. Therefore, when the road surface characteristics suddenly change or the perception is uncertain, the stable and reliable torque control instruction can still be output, and the passing success rate of the vehicle in an unstructured scene is remarkably improved.

The invention relates to the technical field of energy management, in particular to an energy storage cabinet adaptive energy safety management system based on cloud edge collaboration, which is characterized in that a model construction module generates an initial energy state model and an energy storage unit state sub-model; the data updating module receives edge end energy monitoring data and threshold data, positions a state based on an energy fluctuation direction, updates a sub-model, and generates a real-time energy state model; the scheduling processing module responds to a cloud scheduling request, and analyzes scheduling information to determine a scheduling type and an attribute; and the strategy execution module calls the adaptive strategy according to the scheduling attribute, traverses the real-time model to position the target unit, the state layer and the node, generates a security management instruction and sends the security management instruction to the edge end. According to the system, real-time accurate sensing and dynamic updating of an energy state are realized through a cloud edge collaborative architecture, and the problems of state sensing lagging, scheduling strategy stiffness and passive safety response of a traditional energy storage system are solved.

A priori georeferenced vegetative index data is obtained for a worksite, along with field data that is collected by a sensor on a work machine that is performing an operation at the worksite. A predictive model is generated, while the machine is performing the operation, based on the georeferenced vegetative index data and the field data. A model quality metric is generated for the predictive model and is used to determine whether the predictive model is a qualified predicative model. If so, a control system controls a subsystem of the work machine, using the qualified predictive model, and a position of the work machine, to perform the operation.

The invention relates to the technical field of encrypted network traffic security detection, and discloses an encrypted malicious traffic detection system based on deep learning. According to the system, an intelligent data acquisition module, a feature extraction and filtering module, an intelligent data analysis module, an annotation quality verification module, a metadata enhancement detection module, an adversarial disturbance detection module, a dynamic cooperative control module, a real-time online feedback module and a full-scene validity verification module are constructed; a multi-level cooperative processing detection architecture is formed, so that the system can automatically collect metadata and generate pseudo labels, calculate and label data quality scores, average distinction degree of multi-dimensional features and quantitative indexes of feature disturbance threshold values of attack resisting judgment, and perform real-time model iterative optimization and adaptive decision adjustment. Dynamic evaluation of data quality, feature validity and security risk is realized, so that the problems of scarcity of labeled data, feature noise interference, vulnerability to attack resistance and poor environmental suitability are effectively solved.