81 results about "Model dynamics" patented technology

The invention discloses a personal obesity risk prediction system and method based on AI of big data, and belongs to the field of medical health, and the system comprises a data collection module, a multi-dimensional feature construction module, a risk label dynamic generation module, a model training and risk prediction module, a credibility evaluation and calibration module and the like. The system collects multi-source heterogeneous data through wearable equipment, a biochemical interface and a health platform API (Application Program Interface), uniformly encodes the multi-source heterogeneous data into a standard time sequence and then constructs behavior-metabolism-environment coupling characteristics. And performing joint modeling on the dynamic features and the labels by adopting a graph neural network in combination with causal factorization, and outputting an individual obesity risk prediction result. The result credibility is improved through Monte Carlo Dropout and a confidence interval calibration mechanism, and calibration information is fed back to a feature construction link to optimize a modeling strategy. Finally, the key risk factors are presented in the form of a visual thermodynamic diagram and a causal path diagram, and an individualized intervention suggestion vector is generated. The method has the beneficial effects of improving prediction accuracy and enhancing individual intervention pertinence.

The invention relates to the technical field of data center dynamic modeling, and discloses a twin model dynamic modeling method for a data center DCIM platform, which comprises the following steps: constructing a discrete state space model containing a thermal coupling matrix and a system matrix, collecting real-time power and temperature time sequence data, and calculating a cross-correlation function to lock hot air dynamic transmission lag time; calculating cut-off frequency based on physical attributes of the cabinet and decomposing data into high and low frequency components by using a complementary filter; according to the method, the model parameters are made to return to a physical source through a frequency domain decoupling mechanism, the problem of aliasing of airflow coupling and structural thermal inertia parameters in a traditional single-scale identification method is solved, and the method is suitable for large-scale identification. And the physical authenticity and prediction robustness of the twin model under a complex working condition are improved.

The invention relates to the field of power load prediction, and particularly provides a multi-model dynamic fusion load prediction method and system, a terminal and a medium, and the method comprises the steps: constructing a plurality of different types of load prediction models, carrying out the independent training of each load prediction model through a training set, and carrying out the verification of a verification set on a verification set; calculating a dynamic weight corresponding to each load prediction model by adopting a Monte Carlo algorithm on the basis of similar day data similar to the prediction target day in the test set in feature; acquiring historical load data in a preset time period before the current moment, and meteorological data and time characteristic data at the corresponding moment to form a model input data set; preprocessing the input data set, and inputting the preprocessed input data set into each trained load prediction model to obtain an initial load prediction value corresponding to each load prediction model; and according to the dynamic weight, performing weighted fusion on each initial load prediction value, and outputting a final load prediction value. The accuracy of load prediction is improved.

The invention discloses a complex scene target detection method based on a model dynamic joint mechanism, and the method comprises the steps: carrying out the preliminary detection through an RT-DETR model, retaining more potential targets through dynamic threshold adjustment, and projecting a generated detection frame to a feature space of an improved YOLOv12 model through dual-mode feature mapping; the improved YOLOv12 model integrates an SEAM attention module and a rejection loss function so as to enhance feature representation and positioning compactness of an occluded target. Then, a model joint mechanism is adopted to process preliminary results of the two models; through difficult case mining and online learning, missing detection targets are supplemented, and the RT-DETR model is optimized; and intelligently fusing the detection results of the two models through dynamic weight distribution based on scene complexity and hierarchical fusion of a decision tree. And finally, post-processing is carried out by using an improved non-maximum suppression algorithm, and mistaken deletion is reduced. According to the method, the problems of missing detection, false detection and inaccurate positioning of the target in a complex scene are effectively solved, and the recall rate and the accuracy rate of detection are remarkably improved.

The invention provides a tunnel blasting vibration reduction method and system based on asymmetric slotting, relates to the technical field of geotechnical engineering numerical simulation, and provides an integrated intelligent blasting vibration reduction system of asymmetric slotting structure optimization-quantum genetic algorithm collaborative optimization-model dynamic regulation. Triple fusion of a physical mechanism, a mathematical algorithm and an information technology of proximity tunnel blasting vibration control is creatively achieved, the bottleneck that a traditional method depends on experience trial and error, model static lag and vibration control and construction efficiency are difficult to consider at the same time is broken through, vibration speed control precision is improved, tunneling efficiency is improved, and meanwhile the construction efficiency is improved. The dynamic adaptability to complex geological conditions is remarkably enhanced, and a new generation of integrated solution is provided for safe, efficient and intelligent construction of the adjacent tunnel.

The invention relates to the technical field of artificial intelligence, in particular to a browser-side AI reasoning system and method based on TensorFlow.js. The browser-side AI reasoning system and method based on TensorFlow.js comprises the following steps of model loading and initialization, equipment performance detection and model selection, input data collection and preprocessing, reasoning task scheduling and execution, result analysis output and visual export. Model cache updating maintenance and performance monitoring dynamic adjustment and optimization are carried out; the method has the beneficial effects that a set of complete front-end AI reasoning flow control mechanism is provided; realizing model dynamic adaptation based on equipment performance; various input forms are supported; a main thread is prevented from being blocked by utilizing multi-thread scheduling; a GPU acceleration and mixing precision calculation mechanism is introduced to improve the reasoning efficiency; establishing a model caching mechanism to improve the loading speed and the offline availability; performance monitoring and dynamic tuning functions are provided, and reasoning stability is ensured; providing a structured result output and visual display interface; a model version control and background hot update mechanism is realized; and the security, compatibility and expansibility of the system are improved.

The invention discloses a tethered satellite system dynamics modeling method considering perturbation factors. The tethered satellite system dynamics modeling method comprises the steps that a geocentric coordinate system, an orbital coordinate system of a Kepler orbit where a tethered satellite system is located and a body coordinate system of a unit satellite are established; according to the movement of the centroid of the tethered satellite system on the orbit, the generalized coordinates are the distance from the centroid of the system to the earth center and the true anomaly of the orbit, the Lagrange function of the centroid of the system is obtained; according to the movement of the unit star relative to the centroid, the generalized coordinates are the distance between the unit star and the centroid, the relative first rotation angle and the relative second rotation angle, the relative Lagrange function of the unit star is obtained; comprehensively considering perturbation factors to obtain non-spherical earth, third gravitational force and solar pressure perturbation potential energy; establishing a system kinetic equation considering comprehensive perturbation factors; and obtaining a complete Lagrange equation of the tethered satellite system. The error problem caused by the fact that the master satellite directly serves as the mass point of the system and the influence of comprehensive perturbation on system motion is not considered is solved, and the number of the slave satellites is not limited.

The invention discloses an intelligent acquisition and analysis method and system based on a big data model. The method comprises the following steps: 1, carrying out multi-modal identity verification and health state analysis; 2, construction safety real-time monitoring and behavior prediction are carried out; 3, special equipment dynamic security area modeling and risk early warning are carried out; and step 4, carrying out intelligent assessment on the leaving article loss risk. According to the invention, full-process closed-loop management and intelligent collaboration of construction site supervision can be realized, a full-process safety management control closed loop of entrance, on-duty and departure is constructed, a complete chain from risk early warning to disposal feedback is realized, a model dynamic optimization mechanism based on big data continuously improves the adaptability of the system to complex construction scenes, and the safety of construction sites is improved. And the manual intervention dependence is reduced.

The invention discloses a three-dimensional model dynamic rendering method and a storage medium, and the method comprises the steps: obtaining three-dimensional data of a target scene, obtaining a plurality of voxels, and enabling each voxel to correspond to a plurality of voxel attributes and an initial comprehensive displacement vector; in any voxel in any time step, obtaining a comprehensive displacement vector of the voxel in the current time step according to a preset physical deformation rule and / or animation curve deformation rule, and performing coordinate transformation on the voxel; after the coordinate transformation, updating the voxel attribute of the voxel corresponding to the current time step length; and performing real-time three-dimensional rendering on the target scene according to the vertex normal of each voxel at the current time step length and the voxel attribute. The physical consistency is ensured in the dynamic deformation process by synthesizing the displacement vector; by updating the voxel attributes, the voxel attributes are kept in smooth transition after deformation; and high-quality real-time rendering of the target scene is realized by recalculating the deformed vertex normal and combining the illumination model.

The invention discloses an automatic driving vehicle dynamics modeling method based on a time-varying parameter neural network, and belongs to the field of vehicle dynamics. Comprising the steps of obtaining operation data of a vehicle under different working conditions, performing preprocessing, and constructing a sample set according to a time sequence slicing mode; constructing a neural network model based on the physical constraints and the time-varying parameters; training and verifying the neural network model by adopting the sample set to obtain a neural network model after parameter optimization for state prediction; the neural network model comprises an input layer, a GRU network, a physical constraint layer and an output layer; input data enters the neural network model through the input layer, and dimension transformation and time sequence packaging are carried out through the linear embedding layer; then time sequence features are extracted through the GRU network, bounded mapping is carried out on output vectors of the GRU network through a physical constraint layer, a time-varying parameter estimation value is obtained, the time-varying parameter estimation value is substituted into a vehicle kinetic equation for calculation, and a physical prediction state is obtained.

The invention discloses an LLMOps-based large model efficient operation and maintenance method and system, belongs to the technical field of artificial intelligence, and aims to solve the technical problems of how to realize large model efficient operation and maintenance, ensure that the large model quickly responds to business requirements in the application process and maintain the performance and stability of the large model. Comprising the steps that an LLMOps tool chain is constructed based on a model deployment tool, a model detection tool, a model upgrading tool, a model running tool and a model iteration tool, and the model deployment tool deploys a large model based on containerization and standardization and CI / CD assembly line automation; the model detection tool is integrated with a verification framework and a real-time monitoring system; the model upgrading tool applies a service network technology to implement gray level upgrading and A / B testing; the model running tool realizes elastic expansion and load balancing of the model service based on an automatic capacity expansion and contraction mechanism of the container arrangement platform; and a model dynamic iteration mechanism based on feedback driving is configured in the model iteration tool.

The invention discloses a dynamic load response adjustment method and system for a hot water boiler, and relates to the field of boiler load adjustment, and the method comprises the steps: carrying out the deep learning of boiler and weather data through LSTM, and precisely predicting the thermal load change trend in a period of time in the future; on the basis, the current system state of the boiler is estimated in real time in combination with a Kalman filter, and the predicted load, the real-time state and a dynamic system model of the boiler are incorporated into a model predictive control (MPC) framework. And the MPC calculates an optimal gas valve opening sequence considering the physical time lag of the boiler through rolling optimization in a prediction time domain. The prospective prediction is combined with optimization control based on model dynamic characteristics, so that the boiler can adjust fuel supply in advance, response delay caused by system thermal inertia is effectively overcome, rapid and accurate response to dynamic loads is achieved, and the defect that traditional control is always slow by half beat is overcome.

The invention provides a large model routing method and device combined with a reward model, and relates to the technical field of artificial intelligence model dynamic routing, and the method comprises the steps: obtaining a question inputted by a user, and determining a candidate model set according to the user authority and a preset priority rule; performing sensitive information judgment on the question through a sensitive detection model, directly returning a preset answer if sensitive information is detected, and otherwise, entering the next step; carrying out reward evaluation on each model in the candidate model set based on a routing function, training the routing function through a knowledge distillation technology, and optimizing a model selection strategy by taking KL divergence as a loss function; and according to an evaluation result of the routing function, dynamically distributing the problem to a target model with the highest reward value for reasoning, and outputting a processing result of the target model. According to the method, the calculation overhead of large model integration can be effectively reduced, the intelligent level of routing decision is improved, and the system resource utilization rate and the response efficiency are remarkably improved.

The invention provides a CO emission prediction method based on robust kernel feature space semi-supervised drift detection. The method comprises the following steps: calculating a first-order difference component of a historical sample through a pre-constructed historical model to construct a historical data set; calculating feature space statistics of the real-time data, and performing drift detection through the feature space statistics to obtain a drift sample; union set taking operation is carried out on the historical samples and the drift samples, and a new training set is obtained; and performing data updating on the historical model through the new training set, and predicting the CO emission through the updated prediction model. According to the method, through semi-supervised drift detection, the problems of concept drift identification and model dynamic updating under truth value deficiency during CEMS fault are solved, and the CO emission prediction precision and the robustness in a complex industrial scene are improved.

The application provides a triaxial six-direction impact test simulation method and system based on F-DEM, the triaxial six-direction impact test simulation method based on F-DEM comprises a full-size model frame establishing step, an F-DEM coupling step, a model dynamic solving step and an automatic data post-processing step, and the application also provides a triaxial six-direction impact test simulation system based on F-DEM, a three-dimensional full-scale numerical model is established based on the basic parameters of a waveguide rod and a rock sample in a TEHB test system, and the three-dimensional full-scale numerical model comprises a discrete element rock sample model and a finite element waveguide rod model. The application realizes efficient coupling modeling of the finite element waveguide rod and the discrete element rock sample, has good stress wave propagation accuracy, can accurately simulate the dynamic response behavior of the rock sample under triaxial six-direction loading, and provides efficient and verifiable technical support for impact response analysis and engineering application research of rock mass in a complex stress environment.

This invention belongs to the field of ship formation control technology, specifically relating to a neural network-based anti-interference control method for multi-ship formations at sea. It utilizes neural networks to approximate and compensate for uncertainties in the model, while simultaneously constructing a disturbance observer to achieve effective estimation of coupled marine environmental disturbances. To overcome the problem of traditional observers' dependence on accurate models, this method parameterizes the ship motion equations and disturbance terms to establish a canonical model of the disturbance, and uses a disturbance filter to achieve exponential convergence estimation of the disturbance state vector, thus transforming the disturbance suppression problem into an adaptive control problem. Furthermore, this method fully considers input quantization, effectively reducing propeller wear and communication burden by slowing down the rate of change of actuator output. Compared with existing technologies, the anti-interference formation control strategy proposed in this invention can achieve online estimation and suppression of unknown disturbances without relying on any prior model dynamics knowledge.

The invention discloses a financial power transmission right transaction decision optimization method and system based on multi-dimensional dynamic modeling, and relates to the technical field of power market optimization, and the method comprises the steps: collecting multi-source data of a power transmission market, carrying out the preprocessing, extracting a multi-dimensional feature matrix, carrying out the multi-model dynamic collaborative modeling, and generating a joint prediction result; risk index assessment is carried out based on the joint prediction result, and a financial power transmission right transaction decision target expression is generated in combination with constraint conditions by taking revenue maximization and risk minimization as double targets; and solving the financial power transmission right transaction decision target expression by adopting an improved particle swarm optimization algorithm to obtain an optimal financial power transmission right transaction decision scheme. According to the method, the joint prediction result is realized through multi-model dynamic collaborative modeling, the transaction decision target expression is obtained by taking benefit maximization and risk minimization as double targets, and the improved particle swarm optimization is introduced for adaptive optimization, so that the accuracy and robustness of transaction decision can be effectively improved.

The invention discloses a special effect processing method and device, a storage medium and equipment. The special effect processing method comprises the steps of obtaining skeleton binding data of a skeleton binding model corresponding to a target static model; generating a plurality of particles according to the target static model and the skeleton binding data, wherein the plurality of particles correspond to skeleton points in the skeleton binding data; performing special effect processing on the plurality of particles based on a preset special effect, and determining particle dynamic data of the plurality of particles corresponding to the special effect processing; determining model dynamic data of the target static model according to the particle dynamic data; the target static model is controlled to generate the preset special effect of the target static model according to the model dynamic data change, the special effect can be used for the static model, the special effect generation effect is improved, and the special effect generation mode is enriched.

This invention provides a method and system for dynamic adaptation and longitudinal-lateral cooperative control of a multi-degree-of-freedom model, comprising: acquiring vehicle driving state parameters and road environment feature parameters; determining the scene category of the current working condition using a scene classifier based on the vehicle driving state parameters and road environment feature parameters; dynamically selecting a matching vehicle dynamics model from a preset multi-degree-of-freedom model library based on the scene category; constructing a prediction model for a model prediction contour controller based on the selected vehicle dynamics model; and optimizing the model prediction contour controller using a multi-objective cost function that includes contour error, path progress, and stability penalty terms to obtain the cooperative longitudinal and lateral control quantities, and performing longitudinal-lateral cooperative control on the vehicle. This invention solves the technical problem that traditional trajectory tracking MPC easily exceeds the stability boundary under extreme working conditions.

The invention relates to a dynamic project cost intelligent prediction method, system and equipment, and the method comprises the steps: anchoring a prediction dimension through a cost motivation map through a closed-loop process of cost motivation quantification, multi-source data structuring, feature engineering and hybrid model dynamic prediction, and carrying out the directional mapping from multi-source data to a cost motivation dimension, performing corresponding feature processing according to feature types on the basis of feature classification defined by the cost motivation quantization atlas to form a structured feature set for model training; project data of different stages are input into a trained mixed cost prediction model, dynamic project cost prediction is achieved, and the mixed cost prediction model comprises an XGBoost model used for extracting static features and an LSTM model based on extracted historical time series data features. Compared with the prior art, the method has the advantages of realizing more accurate and more adaptive project full-stage dynamic cost prediction and the like.

The invention relates to the technical field of railway engineering construction, and discloses a sky-ground fusion live-action three-dimensional modeling method and system based on domestic data and application, which are used for existing railway line transformation. According to the modeling method, a three-dimensional coarse model of a no-fly zone is constructed through a satellite image, and a high-precision point cloud is generated in combination with ground panoramic photography for supplementation; automatically extracting a building contour based on a Mesh model, and realizing monomer modeling by using a neural network trained by a roof texture sample; and developing a segmented projection algorithm to convert the BIM model to a GIS platform in a lossless manner, thereby realizing multi-source data fusion. According to the method, a multi-source data seamless fusion technology, a BIM and three-dimensional model dynamic mapping method and a fusion scheme of a whole-process localization autonomous controllable architecture are adopted, the problem that data collection of a no-fly zone is difficult can be effectively solved, and the construction design precision and efficiency of complex line positions such as Fuhuai high-speed rails are remarkably improved.

This invention provides an energy-aware intelligent model dynamic pruning and scheduling method and system, relating to the field of model lightweighting technology. By introducing an energy-aware dynamic pruning mechanism, a real-time scheduling strategy, and a closed-loop feedback structure, this invention achieves dynamic optimization of energy consumption, accuracy, and latency during deep learning model inference, offering significant advantages over existing static pruning or fixed model inference schemes. The dynamic pruning decision mechanism of this invention can perform fine-grained structural adjustments based on the real-time energy budget during inference. Its runtime control overhead is far lower than traditional dynamic networks, and it does not significantly increase inference time. The extremely low runtime overhead allows it to operate stably in ultra-low-power scenarios such as wearable devices, IoT nodes, and energy harvesting equipment.

The invention provides a signal quantization and event-triggered unmanned ship trajectory tracking multi-constraint control method based on a neural network observer, and the method comprises the steps: carrying out the linear description of an input quantization process in a controller through a linear analysis model, and enabling the controller to not need the prior information of any quantization parameter; aiming at the problem that the ship speed cannot be directly measured in navigation practice, a neural network adaptive observer is designed, and the motion state of the unmanned ship is estimated; a constraint processing mechanism based on a logarithm barrier function is adopted, and the state variable is kept within a preset safety boundary; a neural network approximates unknown model dynamics in a system, a low-frequency gain learning method is introduced to suppress control signal high-frequency oscillation caused by external disturbance, consistent final boundaries of all closed-loop signals are proved, and boundaries of internal signals and quantization errors are clarified; the effectiveness of the signal quantization and event-triggered unmanned ship trajectory tracking multi-constraint control method based on the neural network observer is verified through a simulation experiment.

The invention discloses a model reasoning method, a model dynamic encryption method and a model dynamic encryption device, and belongs to the technical field of data processing. The model reasoning method comprises the steps of obtaining a model reasoning request, wherein the model reasoning request is used for requesting a first model to perform model reasoning; obtaining a target model fragment related to the model reasoning request and fragment attribute information of the target model fragment from a trusted execution environment (TEE) storage space, wherein the fragment attribute information is used for indicating at least one of the following items of a calculation layer included in the corresponding model fragment: a layer type, a calculation amount and sensitivity to private data; determining a first model fragment and a second model fragment according to fragment attribute information of the target model fragment and system state information of the electronic equipment; transmitting the second model fragment to a rich execution environment (REE); and based on a scheduler between the TEE and the REE, calling the first model fragment in the TEE, and calling the second model fragment in the REE to execute a processing process corresponding to the model reasoning request.

The invention relates to the field of teaching and demonstration appliances for emergency drilling, and discloses an airport emergency plan deduction system fused with a three-dimensional sand table, which comprises an emergency plan management module, a three-dimensional electronic sand table module, a deduction execution and fusion module and an evaluation and backtracking module. Comparing the model dynamic data with preset core teaching constraints in real time during deduction; when deviation is detected, a simulator pause instruction is executed immediately, a cognition guide interaction interface is forcibly activated, a user is forced to make cognition selection in mutually exclusive teaching options, and then deduction is restored. The problem of deduction separation teaching feedback lagging is solved, and by conducting immediate intervention and cognition guide at the moment when a student makes an error, the deduction efficiency is improved. The system is changed from a passive assessment tool to an active teaching guidance tool, and qualitative change from behavior recording to decision logic recording is realized.

The invention discloses a mechanical system dynamics modeling method, and relates to the technical field of mechanical dynamics, and the method comprises the steps: decomposing a target mechanical system into a set of a plurality of structural components; performing finite element modeling on each structural component to obtain a finite element model of each structural component; performing dynamic order reduction processing on the finite element model to obtain a polycondensation model of each structural component; integrating and coupling the polycondensation model into a system body dynamic model based on the motion parameters of each motion axis of the target mechanical system and the system assembly relationship; and coupling the dynamical model of the end effector with the system body dynamical model to obtain a target system dynamical model. The whole machine is decomposed into components, order reduction processing is carried out on the components respectively, dynamic integration is carried out according to real-time motion parameters, real dynamic behaviors of a mechanical system in different working spaces are accurately reflected, and limitation of traditional static modeling is overcome.

The application discloses a self-evolution multi-agent event semantic understanding system and method based on a dynamic scalable model, which comprises a multi-agent cooperation framework modeling module, a demand understanding and task planning module, a capability verification module, a model dynamic expansion module, a task execution module and a completion module connected in sequence, wherein the multi-agent cooperation framework modeling module comprises a central control type multi-agent, a task execution type multi-agent and a model expansion type multi-agent. The method focuses on a news event semantic understanding scene, constructs a dynamic scalable system architecture through a multi-agent cooperation mechanism, introduces a capability verification link to optimize a task execution process, and realizes on-demand iteration of model capability by using an active learning framework, so that adaptive evolution of agent capability is realized, thereby solving the problems of model rigidity and poor adaptability of a traditional event semantic understanding system in a complex news event scene, and significantly improving the dynamic adaptability and accuracy of news event semantic understanding.

The invention discloses a gas sensor dynamic response modeling method based on a liquid continuous time network, and belongs to the technical field of gas sensors, and the method specifically comprises the following steps: testing mixed gas through a gas sensor array, generating a response curve group, forming a sample point with a concentration sequence group, and intercepting the response sequence group; constructing an initial sample set comprising a concentration sequence group, a response sequence and a real time interval sequence; a continuous time dynamic response neural network model is constructed and trained, and a continuous time closed-form solution neural network unit is specifically used as a core calculation unit. According to the method, the real time interval sequence is used as input, so that the modeling precision and the physical authenticity can be improved. The invention further discloses a self-training data enhancement strategy, an effective way is provided for solving the model training problem under the small sample condition, and the generalization ability and robustness of the model on the aspect of unseen data are improved.

Apparatuses, systems, and methods dynamically model intrinsic parameters of a camera. Methods include: collecting, using a camera having a focus motor, calibration data at a series of discrete focus motor positions; generating, from the calibration data, a set of constant point intrinsic parameters; determining, from the set of constant point intrinsic parameters, a subset of intrinsic parameters to model dynamically; performing, for each intrinsic parameter of the subset of intrinsic parameters, a fit of the point intrinsic parameter values against focus motor positions; generating a model of the intrinsic parameters for the camera based, at least in part, on the fit of the point intrinsic parameter values against the focus motor positions; and determining a position of a fiducial marker within a field of view of the camera based, at least in part, on the model of the intrinsic parameters for the camera.

The invention provides a power inspection model dynamic scheduling cooperation method and system based on a hierarchical strategy, and relates to the technical field of power system inspection, and the method comprises the steps: receiving a power inspection task, collecting multi-modal power inspection data, and carrying out the preprocessing; based on a multi-scale convolution feature segmentation algorithm, key features of the electric power inspection task are extracted; evaluating an electric power inspection task level label corresponding to the electric power inspection task in combination with the multi-dimensional task level evaluation index; a corresponding dynamic model scheduling strategy is selected according to the electric power inspection task level label, and models in the dynamic model scheduling strategy comprise a light-weight small model and a cloud large model; and inputting the key features of the electric power inspection task into a model in the dynamic model scheduling strategy, outputting an electric power inspection identification result and pushing the result to the inspection monitoring platform, so that the dynamic model scheduling strategy can be adopted, dynamic distribution and efficient collaboration of model resources are realized, and the real-time performance, the energy efficiency performance and the intelligent level of the electric power inspection system are improved.