920 results about "Feature Dataset" patented technology

The invention relates to a power distribution network distributed energy bearing capacity assessment method and device. The method comprises the following steps: carrying out topology analysis on a network structure of a power distribution network to obtain an initial network topology model; obtaining a node dynamic feature data set based on the initial network topology model and the distributed energy access point data of the power distribution network; wherein the node dynamic characteristic data set comprises operation parameters of each node of the power distribution network in different load scenes; generating a parameter incidence matrix according to the node dynamic characteristic data set, and obtaining a bearing capacity reference model of the power distribution network according to the parameter incidence matrix and real-time data of the power distribution network in an operation state; wherein the parameter incidence matrix is used for quantifying the coupling degree between the operation parameters; and obtaining a risk distribution mapping graph according to the bearing capacity reference model, and identifying a potential overload area of the power distribution network based on the risk distribution mapping graph. According to the invention, power distribution network operation risk assessment can be accurately realized.

The invention relates to the technical field of reservoir dam safety monitoring, and discloses a reservoir dam siltation dynamic monitoring and early warning system. Multi-dimensional sensing node arrays of the system are arranged at key positions of a dam body structure and a reservoir area, and sediment thickness distribution data, water flow velocity field data and sediment concentration gradient data are synchronously collected. And the edge computing node receives the original monitoring data, executes data cleaning and space-time alignment processing, and generates a standardized siltation feature data set. And the cloud analysis platform receives the data set, calculates a deposition evolution trend matrix through a space-time coupling prediction model, and outputs a reservoir area deposition risk level distribution map. And the dynamic visualization engine analyzes the risk level distribution map, generates a three-dimensional dynamic deposition situation model, and marks the space coordinates of the abnormal deposition area. And the early warning decision center generates a graded early warning instruction set according to the space coordinates of the abnormal region, and triggers a corresponding emergency response strategy.

The invention provides a method for identifying PFAS in an environment based on machine learning pseudo-targeted screening, which comprises the following steps: calling mass spectrum data containing PFAS compounds from a preset mass spectrum database, and performing interference peak elimination processing on the mass spectrum data to obtain a model training data set; extracting a feature data set for model training from the model training data set based on a feature extraction standard; training a plurality of machine learning classification models based on the feature data set, and performing performance evaluation on each machine learning classification model based on a training result to determine an optimal machine learning classification model; and analyzing the optimal machine learning classification model, determining key features when the PFAS is screened and identified, and carrying out PFAS screening identification verification on the optimal machine learning classification model according to the key features based on an actual environment sample. The method has the advantages of saving analysis cost, improving analysis efficiency and improving compound recognition accuracy.

The invention discloses a multi-dimensional enterprise qualification evaluation method and system, and the method comprises the steps: obtaining multi-dimensional original data, and obtaining a standardized enterprise multi-dimensional feature data set through employing a data cleaning and normalization preprocessing technology; aiming at a standardized enterprise multi-dimensional feature data set, performing grouping optimization and weight calculation on indexes by utilizing an improved WP-PVC algorithm, and constructing a multi-dimensional evaluation index system; based on the multi-dimensional evaluation index system, a double-standard WP-PVC algorithm is adopted to carry out inter-index correlation analysis and score calculation, and an enterprise qualification comprehensive score result is generated; according to an enterprise qualification comprehensive scoring result, in combination with a multi-logarithm algorithm of random online sorting, calculating a matching degree between the enterprise and various science and technology projects, and outputting a project matching recommendation list and application success rate prediction; and automatically generating an enterprise qualification diagnosis scoring report based on the enterprise qualification comprehensive scoring result and the project matching recommendation list. According to the invention, comprehensive evaluation of enterprise qualification, accurate project matching and scientific application guidance are realized.

The invention discloses a campus intelligent consumption analysis system and method based on multi-dimensional data fusion, and relates to the technical field of electric digital data processing. The system comprises a terminal feature fusion module, an anomaly identification analysis processing module and a classification verification and synchronization module. The terminal feature fusion module fuses terminal feature data and time feature data in the original multi-dimensional data to generate behavior association features, and data dispersion is reduced; the anomaly identification analysis processing module performs anomaly identification and processing on the behavior association features to form a feature data set and verify the validity, so that the detection accuracy is improved; and the classification verification and synchronization module uploads the verified classification result as a core data set to a specified service platform and synchronizes the core data set to a database of the specified service platform, so that the platform processing efficiency is improved. Through cooperation of multiple modules, efficient fusion, accurate anomaly detection and data synchronization of campus consumption data are realized, and comprehensiveness of campus consumption analysis and processing efficiency of a service platform are improved.

The invention relates to the technical field of power utilization monitoring, and discloses a photovoltaic user power utilization abnormity monitoring method and system based on artificial intelligence. The photovoltaic user electricity consumption abnormity monitoring system based on artificial intelligence comprises a data acquisition module which is used for acquiring photovoltaic power generation data, electricity consumption data and environment data of a user; the data preprocessing and feature engineering module is used for cleaning, aligning and normalizing the original data acquired by the data acquisition module and constructing a feature data set for model training and reasoning; and the artificial intelligence analysis engine module comprises an unsupervised learning unit, a supervised learning unit and a deep learning unit. According to the invention, the false alarm rate and the missing report rate can be effectively reduced, the accurate diagnosis of the abnormal type can be realized, and the intelligent and accurate operation and maintenance requirements of power grid enterprises on the power utilization monitoring of photovoltaic users are met.

The invention discloses a tightening process abnormal fluctuation analysis and early warning method and system, and the method comprises the steps: collecting original torque and angle signals of a tightening process in real time, carrying out the time synchronization processing, and carrying out the abnormal value filtering and feature extraction, thereby obtaining a feature data set; then comparing and analyzing the feature data set and a pre-generated standard tightening curve, and performing anomaly judgment through a trained anomaly recognition model based on a comparison result to generate an anomaly recognition result; and finally, determining an early warning level according to an abnormal recognition result and outputting corresponding early warning information. The system correspondingly comprises a data acquisition module, a feature extraction module, an abnormity identification module and an early warning output module. According to the method and system, the dynamic time warping algorithm and the long-short-term memory network model are adopted, early-stage tiny abnormal fluctuation in the tightening process can be effectively recognized, multi-stage early warning and preventive quality control are achieved, and the quality control level and production efficiency of the tightening process are remarkably improved.

The invention relates to the technical field of electron microscope sample pretreatment, and discloses a critical point drying control method and device based on online near-infrared monitoring, and the method comprises the steps: obtaining infrared spectrum data, and unmixing the spectrum data to obtain a feature data set of a solvent; outputting a to-be-corrected signal according to the feature data set, correcting the to-be-corrected signal with preset data, and determining a signal feature of the solvent; carrying out dynamic correction on the signal characteristics and a preset characteristic matrix, outputting concentration time sequence data of the solvent, carrying out steady state judgment by combining a preset convergence threshold value, and determining a critical state of the solvent; acquiring control strategy data according to the critical state, analyzing the control strategy data into execution response characteristics, performing multi-channel mapping and correction processing, and outputting a process control signal; and performing control gain correction on the process control signal to finish closed-loop feedback control of the final drying process. The method is based on multi-solvent spectrum sensing and concentration dynamic analysis, and self-adaptive closed-loop control of the critical point drying process is achieved.

The invention discloses a five-dimensional motion vector real-time construction and calibration method based on multi-sensor fusion, and relates to the technical field of multi-sensor information fusion and dynamic state estimation, and the method comprises the steps: collecting and preprocessing motion carrier data, obtaining a preprocessing data set and a feature data set, inputting the feature data set into a long short-term memory network model, and obtaining a multi-sensor fusion model; and outputting a sensor error offset prediction vector to the extended Kalman filtering model, outputting a preliminary five-dimensional motion vector, and performing consistency verification and correction on the preliminary five-dimensional motion vector through a kinematics constraint equation to obtain a corrected five-dimensional motion vector so as to drive a virtual model corresponding to a motion carrier to perform synchronous position and attitude updating. According to the method, the sensor error is predicted through the long-short-term memory network model, and the dynamic motion model constraint module is additionally arranged to perform physical constraint correction, so that the problems of inaccurate error compensation and lack of physical authenticity of the calculation result under the dynamic working condition are solved, and the construction precision and reliability of the five-dimensional motion vector are improved.

The invention discloses a multi-modal data fusion-based steel wire rope detection and real-time transmission method and system, and the method comprises the steps: starting a detection system, collecting a steel wire rope damage detection signal, collecting equipment position and operation state data, constructing an original data set, carrying out the preprocessing, building an incidence matrix, dividing a processing unit, and then extracting features, and forming an initial feature data set. And extracting damage features through a multi-branch network, strengthening the weight of a key region, fusing position and equipment state features through a multi-modal fusion module, generating fusion vectors, classifying and identifying, and outputting a structured detection result. And the acquisition equipment end performs grading processing on the data, distributes transmission channels according to priorities, dynamically adjusts parameters, adds integrity and time sequence identifiers, and the upper computer verifies the data integrity and triggers abnormal supplementary transmission. And the upper computer decodes the data, reconstructs the waveform, gives an alarm in real time in combination with preset parameters, and generates details of damage key information. Reliable technical support is provided for safe operation and maintenance of the steel wire rope in industrial scenes.

The invention relates to the technical field of circuit board conductivity detection, and discloses a circuit board conductivity detection method and system based on deep learning, and the method comprises the steps: obtaining high-resolution image data and circuit design parameters of a to-be-detected circuit board, and generating a circuit board feature data set; calculating theoretical impedance distribution of each area of the circuit board by using a physical constraint neural network based on the circuit board feature data set, and outputting an impedance prediction parameter set; inputting the impedance prediction parameter set and the circuit symbol rule into a neural symbol inference device to generate a conductance fault diagnosis result; and obtaining diagnosis results of a plurality of detection sites, executing cross-site diagnosis fusion, and outputting a unified circuit board conductivity detection report. The method overcomes the limitation that a traditional visual detection method cannot identify electrical characteristic defects, and solves the detection problem of hidden faults such as discontinuous impedance of the high-frequency circuit board.

The invention relates to a cache data processing method and device, computer equipment, a computer readable storage medium and a computer program product. The method is applied to a distributed storage system adopting a three-layer architecture design, and the three-layer architecture comprises a service application layer, a data processing layer and a persistence layer. Comprising the following steps: acquiring access request monitoring information through a service application layer to obtain an access feature data set, and performing pattern classification identification on the access feature data set to obtain a predicted access pattern; generating a prefetching task list and prefetching data according to the predicted access mode, and storing the prefetching data to a data processing layer for caching processing; and executing each prefetching task according to the sequence of the prefetching task list, and reading the prefetching data from the mechanical hard disk and writing the prefetching data into a cache space in a solid state disk resource pool under the condition that the prefetching data does not exist in the solid state disk resource pool of the persistence layer. By adopting the method, the cache performance can be improved.

The invention discloses a centrifugal pump flow dynamic detection system based on the Internet of Things, and relates to the technical field of intelligent sensing systems. Comprising a data acquisition module used for acquiring multi-source signals in real time and obtaining a sensing feature data set after preprocessing; the physical property parameter module is used for calculating the influence parameters of the physical property change of the medium on the pump performance based on the sensing characteristic data set to obtain a medium physical property parameter set; the flow prediction module is used for correcting a preset flow prediction model through the medium physical property parameter set to obtain a corrected flow prediction model; inputting a sensing characteristic data set acquired in real time into the corrected flow prediction model, and outputting to obtain a flow prediction value; the traffic detection module is used for judging whether the current traffic is abnormal or not based on the traffic predicted value, the historical traffic data and a preset operation threshold parameter set, and giving a traffic detection result; the influence of medium physical property changes on flow detection can be accurately captured, and the accuracy and stability of flow detection are improved.

The invention relates to a low-dimensional subspace clustering method based on projection matrix guidance, and the method comprises the steps: extracting a light response non-uniformity PRNU noise residual error from input image data through employing a denoising filter, and constructing a PRNU feature data set of an image; performing feature dimension reduction on the feature data set by adopting a projection matrix method, constructing a projection matrix maintaining a geometric structure, mapping the projection matrix to a low-dimensional potential subspace, and further constructing a model for the subspace by utilizing a sparse self-representation method; constraint is applied to sparse self-representation in the low-dimensional potential subspace, and joint optimization is carried out through an augmented Lagrange multiplier method ALM and an alternating direction minimization ADM strategy to be used for efficient clustering of data in the low-dimensional potential subspace. According to the method, the projection matrix maintaining the geometric structure is constructed, the high-dimensional PRNU features are mapped to the low-dimensional potential subspace, the local neighborhood relation and the global distribution structure are reserved in the dimension reduction process, the calculation cost is reduced, and the clustering robustness and performance are effectively improved.

The invention relates to the technical field of electrical fault positioning, in particular to an urban rail train relay state online monitoring method and system. The specific implementation process comprises the following steps: collecting coil current, contact voltage, load current and environment vibration signals by using an electrical sensor group; when the coil current is subjected to step change, intercepting an electromechanical characteristic data set; arc energy and contact resistance fluctuation characteristics are calculated, correlation analysis is carried out in combination with environment vibration signals, and passive fluctuation and active fluctuation are distinguished; and establishing a multi-factor health degree evaluation model, dynamically adjusting the arc energy and the weight coefficient of the contact resistance according to the magnitude of the load current and the vibration intensity, monitoring the health state of the relay in real time, and predicting the residual life. Through multi-dimensional data synchronous acquisition and vibration decoupling, the interference of mechanical vibration on electrical performance monitoring is effectively eliminated, the problem that a dynamic working condition cannot be reproduced in an off-line test is improved, and the accuracy of relay state evaluation and service life prediction is remarkably improved.

The invention discloses an encrypted domain name resolution protocol simulation and characterization system, and relates to the technical field of network security and network traffic analysis. The invention aims to simulate an encrypted domain name resolution process in a real network environment, collect the flow of the process and extract features to construct a data set, and ensure the quality of the generated data set through data enhancement and a data set evaluation scheme. The system comprises a traffic simulation module, a traffic representation module, a data enhancement module and a data set evaluation module. The flow simulation and characterization module simulates and encrypts domain name resolution flow and extracts a structured feature vector containing 34 side channel features; and the data enhancement and data set evaluation module is used for enhancing a feature set based on a conditional table generative adversarial network CTGAN so as to construct a feature data set which is closer to traffic in a real network environment, and ensuring that the constructed data set has engineering availability and theoretical rationality through evaluation. The system can be used for constructing a current scarce encrypted domain name resolution protocol side channel feature data set, and provides data support for related security detection and research.

The invention discloses a parameter identification method for an electrochemical-thermal-micro short circuit coupling model. The method comprises the following steps: constructing the electrochemical-thermal-micro short circuit coupling model; collecting multi-working-condition data of battery operation, and clustering the multi-working-condition data; performing feature extraction on the clustered data to obtain a feature data set; performing sensitivity analysis on parameters in the electrochemical-thermal-micro short circuit coupling model, and estimating the influence degree of the parameters on the characteristics; and parameters with high influence degrees are preferentially considered, probability influence evaluation is carried out on the parameters, and a parameter identification result is obtained. According to the method, on the premise of accurately judging the potential influence of the parameters, the accurate identification of the operation characteristic parameters of the battery system when the micro short circuit occurs can be realized.

The invention discloses a computing power distribution method based on AI intelligent scheduling, and relates to the field of data analysis, and the method comprises the steps: collecting computing power demand parameters of AI analysis and diagnosis tasks of medical images, and carrying out the structural processing of the computing power demand parameters, and forming a feature data set; computing resources distributed at different medical nodes are integrated, hardware attributes and performance indexes of the computing resources are extracted and converted into standardized computing power characterization parameters, and a shared computing power pool is constructed; according to the method, medical image AI task computing power demand parameters are collected and subjected to structural processing, multi-medical node computing resources are integrated to construct a shared computing power pool, computing power required by tasks and processing duration are accurately predicted through model training, and a scheduling decision model is established in combination with task priorities to realize optimal allocation of computing power. And the task completion condition can be quantitatively evaluated, and double models can be fed back and optimized.

The invention relates to the technical field of industrial equipment fault detection, and discloses a generator iron core lamination equipment fault detection method and system, and the method comprises the steps: obtaining an original vibration signal, and obtaining a first vibration data set through synchronous calibration and feature extraction; a second vibration data set is obtained through time-frequency analysis, decomposition and denoising; constructing a model to enhance the periodic weak impact signal, and obtaining an enhanced signal feature set; performing matching quantification with a preset reference frequency template library, and determining a potential fault signal distribution range; and when the threshold value is exceeded, a fault feature data set is obtained through depth feature extraction, a hyperplane decision model is constructed based on the fault feature data set, and a feature local evolution trend is analyzed to output an early warning result. The method can accurately capture early fault signals, reduces the missing detection and false detection rate, and improves the operation reliability of equipment.

The invention discloses a brain disease risk prediction method and system based on big data analysis, and belongs to the technical field of brain disease risk prediction. The method comprises the following steps: carrying out standardized preprocessing and tagged classification on brain disease related big data to generate a feature data set; mining specific disease characteristics and risk factors in the set, and carding an association rule; training a risk prediction sub-model for each disease type based on the data, and building a multi-sub-model hierarchical prediction system; and collecting to-be-predicted object data, matching a disease type, and calling the corresponding sub-model to complete risk assessment. The system comprises multiple modules for collaborative operation, and a full-process closed loop of data storage, feature processing, model management and result output is realized. According to the scheme, the pertinence, the accuracy and the efficiency of risk prediction are improved, the traceability of the whole process and the dynamic optimization of the model are realized, and reliable technical support is provided for early screening and risk early warning of brain diseases.

The invention relates to the technical field of pressure-bearing structure welding defect identification, and discloses a pressure-bearing structure welding defect automatic identification and risk prediction method and system, and the method comprises the steps: obtaining an initial defect feature data set, carrying out the serialization analysis according to the initial defect feature data set, determining the initial defect extension trend, and carrying out the risk prediction of the initial defect extension trend; carrying out stress response analysis according to the defect initial expansion trend to obtain a defect stress response sequence, carrying out microscopic difference extraction analysis based on the defect stress response sequence to obtain a defect microcrack expansion trend, and carrying out time sequence prediction processing according to the defect microcrack expansion trend to obtain an expansion prediction vector; and carrying out characteristic evolution analysis and fatigue life calculation on the extended prediction vector, determining a potential failure time point, and carrying out risk index fusion calculation according to the potential failure time point to obtain a final failure risk prediction report. The method can solve the problem of insufficient dynamic monitoring in the prior art.

The invention discloses a slurry shield slurry discharge pipeline stagnant discharge early warning method based on multi-source data fusion and dynamic risk evolution, and the method comprises the steps: collecting multi-source monitoring data of a slurry discharge pipeline, and obtaining a comprehensive feature data set through multi-scale feature extraction; analyzing and calculating a time-varying weight based on the parameter coupling degree, constructing a risk evolution graph, and quantifying a risk evolution index; dynamically calculating a self-adaptive early warning threshold value in combination with the construction working condition; and performing multi-level early warning and risk traceability analysis according to the risk level. According to the method, the slurry discharging system is regarded as a nonlinear power system, the risk evolution trajectory is tracked through the phase-space reconstruction technology, the problems of unclear parameter coupling mechanism, risk evolution process deficiency, early warning threshold staticization and the like of a traditional method are solved, early accurate early warning and traceability diagnosis of the stagnant discharging risk are achieved, and the method is suitable for large-scale popularization and application. And the safety and efficiency of slurry shield construction are improved.

The invention provides an intelligent night lamp control method, device and equipment based on multi-mode AI and a medium, and is suitable for the technical field of digital dimming, and the method comprises the steps: carrying out the feature extraction of sensor data, and obtaining an environment feature data set, performing spatio-temporal behavior identification of each user and segmentation of an activity area on the environment feature data set to obtain a target area and a non-target area, calculating a preliminary illumination parameter set according to the environment feature data set, the target area and the non-target area, and performing light pollution simulation evaluation on the preliminary illumination parameter set according to the non-target area to obtain a light pollution index, and performing multi-target constraint optimization on the preliminary illumination parameter set according to the light pollution index to obtain an adjustment illumination parameter set, and performing signal modulation on the adjustment illumination parameter set according to the equipment data to generate a night lamp control signal. According to the method and the device, directional, low-interference and energy-saving safe illumination for the user getting up at night is realized through region segmentation based on user behavior recognition and regional preliminary dimming.

The invention discloses a hydropower station time-sharing power generation income dynamic control method fusing a power grid load demand, and the method comprises the steps: obtaining hydropower station multi-dimensional operation data and power grid time-sharing load and electricity price data, building a data association model based on an attention mechanism, generating a fusion feature data set through dynamic weight distribution, and carrying out the dynamic control of the hydropower station time-sharing power generation income. Time-sharing power generation income maximization serves as an objective function, reservoir water balance, unit output limitation and downstream ecological flow constraint are combined, a deep reinforcement learning algorithm is introduced to construct a dynamic control model, historical and simulated data are utilized to train the model and optimize parameters, and then real-time data are input to generate an optimal time-sharing power generation control scheme. Issuing an instruction and monitoring output deviation to realize dynamic control; according to the method, the power grid load and the time-of-use electricity price can be accurately matched, the power generation income is improved while multiple constraints are met, and the method is suitable for various hydropower station dispatching scenes needing to consider power grid response and income optimization.

The invention relates to the technical field of water conservancy control, and discloses a water conservancy gate cooperative control method and system, and the method comprises the steps: fusing the real-time operation parameters and environmental hydrological parameters of a water conservancy gate, and obtaining a dynamic feature data set; identifying the interaction relationship between the water level and the flow velocity parameter, and constructing a hydraulic association network; determining a leading gate and an auxiliary gate based on the network edge weight, and generating an initial control instruction sequence; a control effect data set is formed by monitoring actual data and an operation state of opening degree adjustment; analyzing a control deviation between the control effect data set and an expected control target to correct an action time sequence and an opening parameter in the gate coordination rule to obtain an optimal control strategy; a gate cooperation rule in the optimization control strategy is fed back to the construction process of the hydraulic association network, and the judgment standards of the main gate and the auxiliary gate are updated; the cooperative control efficiency of the water conservancy gate can be improved.

The application discloses a vehicle test boundary scene generation method and system based on a pre-boundary scene, comprising the following steps: obtaining a first feature data set; obtaining a real boundary scene and a real pre-boundary scene based on the first feature data set and a preset risk discrimination criterion; constructing a second feature data set based on real pre-boundary scene data and the real boundary scene data; obtaining generated pre-boundary scene data by using a generation model based on the real pre-boundary scene data; training a prediction model based on the real pre-boundary scene data and the real boundary scene data; and obtaining test prediction boundary scene data based on the generated pre-boundary scene data and the prediction model. The method provided by the application obtains test prediction boundary scene data based on real pre-boundary scene data, and uses a generation model and a prediction model to expand real boundary scene data, thereby solving the technical problem that existing test scene data is sparse due to the limited space of generated scene data.

The invention belongs to the technical field of hydraulic performance data analysis, and discloses a hydraulic performance data analysis method for a high-temperature liquid metal circulating pump. The method comprises the steps that time domain features and frequency domain features are extracted based on a standardized data set, a feature data set is obtained and stored in a database, health state detection is conducted based on the real-time and historical feature data set, and a real-time health state report is obtained. The method comprises the steps of predicting a hydraulic performance trend and equipment degradation based on a real-time and historical feature data set to obtain a performance trend prediction report, analyzing based on a real-time health state report and the performance trend prediction report to obtain an adaptive control instruction set, and managing, checking and outputting the adaptive control instruction set. In general, the method has the remarkable advantages that the reliability of the basic data is high, the deep analysis capability of the data is high, and the intelligent effect of the system is good.

The invention is suitable for the technical field of computers, and provides a training method and device of a detection model and a service quality detection method and device.The training method of the detection model comprises the steps that a first feature data set is obtained, and the first feature data set comprises multiple first input features and corresponding service quality; taking all the first input features and the corresponding service quality as a model input sample and a model output sample respectively, and training the detection model to obtain an intermediate detection model; the intermediate detection model is used for learning effective features related to the service quality in the first input features; determining a second feature data set based on the first feature data set and an intermediate detection model; and training the intermediate detection model based on the second feature data set to obtain a target detection model. Through the training method, the target detection model accurately predicts the service quality of the monitoring data, and cluster resource scheduling in the cloud system is ensured to be matched with the service quality demand.

The invention provides an electric power system carbon potential tracking and predicting method based on physical and data dual drive. The method comprises the following steps: constructing a time sequence characteristic data set; constructing a source-side unit double-layer collaboration map, and generating a normalized adjacent matrix; constructing a physically guided source side graph neural network model, and outputting a predicted source side dynamic carbon emission factor and a node carbon injection amount sequence; constructing a high-dimensional input tensor through a space-time diagram attention mechanism; constructing a whole-network space-time diagram neural network model to realize dynamic tracking of carbon potential; the method comprises the following steps: introducing a Kirchhoff carbon flow conservation law as a physical regularization term in a space-time law deduction process, constructing a mixed loss function containing the physical regularization term, and executing an optimal sentinel mechanism in a back propagation process of whole-network space-time diagram neural network model training, and finally, outputting a dynamic carbon potential prediction result of the load side after physical verification. According to the invention, the precision and credibility of carbon potential prediction can be improved.

The invention provides an egg product defect detection and grading method and system based on a deep learning technology. The method comprises the following steps: acquiring a visible light image data set and a near-infrared data set of an egg product to be detected, and fusing the visible light image data set and the near-infrared data set to form an original data set; establishing a three-dimensional coordinate transformation model, and mapping the original data set to a standardized ellipsoid coordinate system; processing data in the mapped data set to obtain a standardized feature data set; constructing a double-branch convolutional neural network fusing channel attention and space attention, and extracting features from the feature data set by using two branches; fusing the features of the two branches, and outputting a comprehensive feature vector; based on the comprehensive feature vector, performing hierarchical decision to obtain a detection result; and associating the detection result with the metadata of the egg product to generate a quality detection report. According to the method, the defect detection accuracy is improved, the inconsistency of manual subjective judgment is avoided, the whole-process traceable management of egg product quality is realized, and a powerful guarantee is provided for food safety.