Power grid model construction method, system and equipment based on transmission power information and medium
By using multimodal feature extraction and environmental perception technologies, the physical parameters of the power grid model are dynamically updated, solving the problems of information fragmentation and insufficient response in the power grid model, and realizing high-precision simulation and adaptive control.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-03
- Publication Date
- 2026-03-10
AI Technical Summary
Existing power grid models cannot effectively integrate unstructured data, lack near real-time sensing and dynamic update mechanisms for model parameters, cannot automatically associate with specific equipment models, and are insufficient in responding to changes in the physical characteristics of equipment.
By collecting real-time data from power grid equipment and extracting multimodal features, combined with environmental meteorological data, the physical parameters of the equipment are dynamically updated. Simulation and calibration are then performed using a digital twin of the power grid, enabling real-time parameter correction and model self-improvement.
It achieves deep integration of unstructured data and physical models, improving the dynamic accuracy and adaptability of the models, enabling preventative control of equipment state changes, and enhancing simulation accuracy and long-term operational reliability.
Smart Images
Figure CN121637247A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power grid modeling technology, and specifically to a method, system, equipment, and medium for constructing power grid models based on transmitted electricity information. Background Technology
[0002] Traditional power grid modeling systems mainly rely on structured electrical parameter databases for modeling and simulation, such as voltage, current, and power collected by SCADA systems. They typically focus on core electrical measurements and topology connections, and perform core functions such as power flow calculation, short-circuit analysis, and state estimation.
[0003] However, existing technologies have significant shortcomings, including the separation of electrical parameters from unstructured data and insufficient information integration. Current systems fail to effectively integrate the large amount of unstructured data generated during power grid operation and maintenance. For example, inspection reports contain key qualitative information such as equipment appearance descriptions, abnormal phenomenon records, and preliminary judgments, which cannot be automatically linked to specific equipment models or quantitative parameters. Infrared spectra reflect equipment temperature distribution and potential overheating fault points, but the thermal anomaly information they contain is difficult to directly convert into electrical or state parameters that can be identified by the model. Most existing power grid models are static or quasi-static models, which are insufficient for modeling the time-varying characteristics of equipment physical parameters that change significantly with the external environment, especially temperature. For short-term, drastic environmental changes and the dynamic changes in equipment physical characteristics they cause, such as the increase in sag caused by significant thermal expansion of conductors in extreme high-temperature weather, there is a lack of near-real-time sensing, recording, and dynamic updating mechanisms for model parameters. Summary of the Invention
[0004] In view of the above-mentioned existing problems, the present invention provides a method, system, device and medium for constructing a power grid model based on transmitted power information, in order to solve the problems in the prior art that cannot be automatically associated with specific equipment models or quantitative parameters, and lack a near real-time sensing, recording and dynamic updating mechanism for model parameters.
[0005] To address the aforementioned technical issues, a method for constructing a power grid model based on transmitted electricity information is proposed, including: Real-time data acquisition of power grid equipment is performed, and multimodal feature extraction is conducted on the acquired data to obtain feature vectors of equipment anomalies and overheating index based on temperature gradient distribution. Based on the feature vectors and overheating index, the electrical parameters of the equipment are corrected through parameter fusion. Real-time acquisition of environmental meteorological data is also performed, and temperature change curves are predicted. Based on the predicted temperature change curves and the corrected electrical parameters, the physical parameters of the equipment are dynamically updated. The updated physical parameters are injected into the multiphysics coupling equations to update the dynamic physical model library and generate a power grid digital twin. Power grid simulation is performed using the power grid digital twin, and the simulated values are compared with the actual measured values. When the error exceeds the threshold, parameter backtracking calibration is triggered.
[0006] As a preferred embodiment of the power grid model construction method based on power transmission information described in this invention, the multimodal feature extraction includes: performing text analysis on the inspection report text, parsing the text content through natural language processing technology, identifying equipment abnormal keywords, and generating feature vectors representing the abnormal state of the equipment. The infrared thermal imaging spectrum is processed to extract temperature gradient distribution features, and the local overheating index of key areas of the equipment is calculated based on the temperature distribution. In this process, text analysis and graph processing are performed simultaneously. When the difference between the timestamps of the text and graph data is less than a set threshold, the feature vector fusion enable signal is activated.
[0007] As a preferred embodiment of the power grid model construction method based on power transmission information described in this invention, the step of correcting the electrical parameters of the equipment through parameter fusion includes converting the feature vector into an equipment status score through a mapping function, representing the probability that the equipment is in an abnormal state. Based on the equipment status score and local overheating index, it is determined whether to trigger electrical parameter correction. When the equipment status score exceeds the preset threshold or the local overheating index exceeds the safety threshold, the electrical parameters of the equipment are dynamically corrected. The corrected electrical parameters are packaged together with the ambient temperature prediction curve and used as input for time-varying parameter updates.
[0008] As a preferred embodiment of the power grid model construction method based on power transmission information described in this invention, the physical parameters of the dynamically updated equipment include: real-time collection of environmental meteorological data through environmental sensing technology, and prediction of future temperature change curves based on historical meteorological sequences; Based on the predicted temperature change curve and the corrected electrical parameters, the physical parameters of the dynamic computing device are updated using time-varying parameters, and the updated parameters are injected into the dynamic physical model library.
[0009] As a preferred embodiment of the power grid model construction method based on power transmission information described in this invention, the environmental perception technology includes: using a temperature and humidity sensor network to collect environmental data at a frequency of 5Hz, deploying the sensors at key nodes of the power transmission corridor, using a long short-term memory network to predict temperature change curves, inputting historical time series data, outputting future temperature prediction values, and cleaning, normalizing and spatiotemporally aligning the original data, constructing training samples using historical sequences within a sliding time window, and predicting future temperature curves; The method of correcting the electrical parameters of the device through parameter fusion also includes using the Sigmoid function to map the feature vector into a device status score, and triggering electrical parameter correction based on the device status score and the local overheating index. The formula for obtaining the equipment status score is expressed as follows: in, Rate the equipment status. For the Sigmoid function, As weight, This is the feature vector of the equipment malfunction. For bias; Electrical parameter corrections include adjusting the wire resistance based on the highest temperature and the material temperature coefficient, expressed by the formula: in, This is the corrected wire resistance value. This is the standard resistance value. The temperature coefficient of the material. This is the highest temperature measured. This is a reference temperature.
[0010] As a preferred embodiment of the power grid model construction method based on power transmission information described in this invention, the physical parameters of the dynamically updated equipment include: dynamically calculating conductor sag based on time-varying parameters, which is calculated based on the instantaneous value of conductor tension, weight per unit length, span, coefficient of thermal expansion and predicted temperature change curve. During the calculation process, environmental sensing data is associated in real time, and the updated parameters are injected into the admittance matrix of the power grid digital twin. The formula for calculating conductor sag is expressed as: in, Let sag of the conductor be at time t. This refers to the weight per unit length of the conductor. The span is the horizontal distance between the two suspension points of the conductor. Let be the instantaneous value of the conductor tension at time t. The coefficient of thermal expansion of the conductor. The predicted temperature change curve; The dynamic physical model library stores the multi-physics field coupling equations of the device. The model equations are linked in real time to the ambient temperature data collected by environmental sensing, and the winding temperature state is updated through numerical solution. The multiphysics coupling equation is expressed as follows: in, This represents the rate of change of winding temperature over time. This refers to the winding temperature of the transformer. The current flowing through the winding, The resistance of the winding, For the heat capacity of the transformer, Here is the ambient temperature, and here is the thermal resistance of the transformer.
[0011] As a preferred embodiment of the power grid model construction method based on power transmission information described in this invention, the generation of the power grid digital twin includes: generating dynamic topology and electrical parameters based on the updated model library, performing measurement and simulation error verification, acquiring real-time switch status and measurement values, dynamically adjusting the model topology connection relationship, using the dynamic parameterized model to perform power system flow calculation, solving the voltage of network nodes and the power flow of branches, and comparing the measurement values and simulation values at the same time and physical location point by point. When the relative error between the simulation value and the measurement value exceeds 5%, parameter backtracking calibration is triggered. The calibration trigger command includes the specific measurement point location, time point, measured value and simulated value, error value, identification of the currently used equipment model, and dynamic parameter value where the error exceeds the limit.
[0012] The beneficial effects of this preferred technical solution are as follows: by predicting future temperature change curves through an LSTM network and directly using the predicted values to calculate parameters such as conductor sag, the physical impact of environmental changes on power grid equipment can be reflected in advance, supporting preventive control. Furthermore, by introducing a correction formula for conductor resistance based on the material temperature coefficient and the measured highest temperature, the influence of temperature on electrical parameters is characterized, making electrical calculations more consistent with the actual physical state of the equipment and improving simulation accuracy.
[0013] As a preferred embodiment of the power grid model construction system based on power transmission information described in this invention, it is characterized by including a data acquisition module, a multimodal feature extraction module, an environmental perception module, a parameter fusion module, and a power grid maintenance and update module.
[0014] The data acquisition module is used to receive structured inspection report text through the API interface of the mobile inspection terminal, acquire raw thermal imaging video stream through the fiber optic interface of the infrared thermal imager, perform protocol conversion and encoding processing on the text data, extract key frames and compress the format of the video stream, and bind spatiotemporal tags to the text and image data through the data transmission controller, and transmit them to subsequent processing through a high-speed network.
[0015] The multimodal feature extraction module is used to parse the inspection report using natural language processing technology, identify abnormal keywords of equipment, and generate feature vectors representing the status of the equipment. The graph processing unit uses computer vision technology to analyze the infrared thermal imaging graph, extract temperature gradient distribution features, and calculate the local overheating index of key areas of the equipment. The synchronization controller ensures the temporal consistency of the text and graph data. When severe overheating is detected, the key fault mode scan is triggered to perform deep feature mining and fusion of multi-source data.
[0016] The environmental perception module is used to collect meteorological data such as temperature, humidity, wind speed, and air pressure in real time through a distributed sensor network deployed at key nodes of the power transmission corridor. It also uses a long short-term memory network to perform deep learning on historical meteorological sequences, establish a temperature change prediction model, and output temperature change curves for future time scales. The module also performs cleaning, normalization, and spatiotemporal alignment processing on the raw data.
[0017] The parameter fusion module is used to convert feature vectors into equipment status scores through a mapping function, determine whether to trigger electrical parameter correction based on the score results and local overheating index, and dynamically adjust the electrical parameters of the equipment according to material properties, temperature data and physical laws when the triggering conditions are met, and package the corrected parameters with the environmental prediction curve for output.
[0018] The maintenance and update grid module is used to dynamically calculate physical parameters that change over time based on environmental data, equipment status, and prediction curves through a time-varying parameter updater. The dynamic physical model library stores the multi-physics coupling equations of the equipment, updates the equipment status in real time through numerical solution methods, and injects the updated high-fidelity parameters into the grid digital twin.
[0019] A computer device includes a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of a method for constructing a power grid model based on transmitted electricity information.
[0020] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method for constructing a power grid model based on transmitted electrical information.
[0021] The beneficial effects of this invention are as follows: By real-time acquisition of inspection report text and infrared thermal imaging spectra and multimodal feature extraction, this invention transforms unstructured qualitative descriptions and image information into quantifiable equipment anomaly feature vectors and local overheating indices, achieving deep fusion of unstructured data and physical models and solving the problem of information fragmentation. Furthermore, by mapping feature vectors to equipment status scores through parameter fusion, and combining the overheating index with preset thresholds to intelligently trigger electrical parameter corrections, it automates the process from feature recognition to maintenance decision-making. By leveraging sensor networks and LSTM networks deployed in transmission corridors to predict temperature change curves, it provides the model with forward-looking environmental input. The time-varying parameter updater, based on predicted temperature and correction parameters, dynamically calculates conductor sag and transformer winding temperature using sag calculation formulas containing thermally induced deformation components and multiphysics coupling equations, achieving high-fidelity simulation of equipment physical states and forward-looking response to environmental factors. The power grid digital twin uses the updated model to perform power flow calculations and compares the simulated values with SCADA measurements point by point. When the error exceeds a threshold, it triggers parameter backtracking calibration with detailed information, significantly improving the model's dynamic accuracy, adaptability, and long-term operational reliability. Attached Figure Description
[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0023] Figure 1 The overall flowchart of a power grid model construction method based on power transmission information provided in one embodiment of the present invention is shown.
[0024] Figure 2 This is a schematic diagram of the data acquisition process for a power grid model construction method based on power transmission information, provided in one embodiment of the present invention.
[0025] Figure 3 This is a schematic diagram of the multimodal feature extraction process of a power grid model construction method based on power transmission information provided in an embodiment of the present invention.
[0026] Figure 4 A system scheme flowchart for a power grid model construction system based on power transmission information provided in one embodiment of the present invention. Detailed Implementation
[0027] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.
[0028] Example 1, referring to Figure 1 As an embodiment of the present invention, a method for constructing a power grid model based on transmitted electricity information is provided, comprising: S100: Real-time acquisition of power grid equipment data, multi-modal feature extraction of the acquired data, and obtaining feature vectors of equipment anomalies and overheating index based on temperature gradient distribution.
[0029] S200: Based on feature vectors and overheat index, the electrical parameters of the equipment are corrected through parameter fusion. Environmental meteorological data is collected in real time, and temperature change curves are predicted. Based on the predicted temperature change curves and the corrected electrical parameters, the physical parameters of the equipment are dynamically updated.
[0030] S300: Injects the updated physical parameters into the multiphysics coupling equation, updates the dynamic physical model library, and generates a digital twin of the power grid. It then uses the digital twin to perform power grid simulation and compares the simulated values with the actual measured values. When the error exceeds the threshold, it triggers parameter backtracking calibration.
[0031] It should be noted that this invention, through multimodal feature extraction, integrates qualitative and quantitative data, overcomes the limitations of a single information source, significantly improves the accuracy of abnormal state identification, and transforms the equipment health status into calculable electrical parameter corrections through parameter fusion, enabling the model to reflect real working conditions in real time. It also dynamically updates physical parameters in conjunction with environmental and meteorological forecasts, and achieves self-improvement and continuous evolution of the model through a closed-loop calibration mechanism of simulation and measurement.
[0032] Example 2, refer to Figures 1-3 This is a second embodiment of the present invention, which provides a method for constructing a power grid model based on transmitted electricity information, including: In this embodiment of the application, step S100, the acquisition of power grid equipment data includes real-time acquisition of inspection report text and infrared thermal imaging spectra of power grid equipment, specifically including steps S101~S104: S101: Receives structured report text via the mobile inspection terminal API, converts it according to the protocol, and stores it in the text cache queue.
[0033] S102: The raw thermal imaging video stream is acquired through the fiber optic interface of the infrared thermal imager, and key frame images are extracted by the frame extractor and stored in the image buffer pool.
[0034] S103: The data transmission controller polls the two buffers. Once new data is found, it triggers the data packer to bind spatiotemporal tags containing device ID, GPS coordinates and timestamps to text and images.
[0035] S104: The packaged data is transmitted to the feature extraction module via Gigabit Ethernet. The text is stored in a Redis queue in UTF-8 encoding, while the images are compressed into H.264 format for storage.
[0036] In an optional implementation, in step S100, the data collection of power grid equipment further includes the drone flying along a preset transmission line, capturing infrared thermal imaging spectra through a multispectral sensor, collecting verbal reports from inspection personnel through a voice recorder and automatically converting them into text, transmitting the data to a cloud platform in real time via a 5G network, performing spatiotemporal alignment and format standardization in the cloud, and storing it in a distributed database.
[0037] In another optional implementation, in step S100, the collection of power grid equipment data may further include: a camera periodically capturing infrared thermal images; a microphone array collecting ambient sound and the voice of inspection personnel; voice recognition and text generation through edge computing nodes; and low-power transmission of the data to a central server via the LoRaWAN protocol, and storage in a time-series database.
[0038] Furthermore, in step S100, the multimodal feature extraction includes steps S111~S113: S111: In terms of text analysis, the inspection report is preprocessed by word segmentation, stop word filtering and entity annotation to generate a standardized text sequence, which is then input into the BERT model encoding layer to generate equipment anomaly feature vectors that can quantify the weights of various fault keywords.
[0039] S112: Image processing performs non-uniformity correction, radiometric calibration and temperature matrix reconstruction on infrared thermal imaging images, and feeds the preprocessed images into a pre-trained convolutional neural network based on the ResNet-50 architecture. The convolutional neural network outputs a temperature gradient matrix that reflects the overall temperature distribution of the device through a fully connected layer with 512 neurons and global average pooling.
[0040] S113: The system loads the preset key area of interest (ROI) according to the device ID, and uses a mask to extract the temperature value in the region from the matrix. By calculating the sum of squares of the deviations between the temperature of each pixel in the region and the average temperature, the local overheating index is obtained to quantify the severity of overheating in a specific region. The formula for obtaining the local overheating index is expressed as: in, The Local Overheating Index indicates the severity of overheating within the Area of Critical Concern (ROI) of the equipment. For pixels, Key areas of concern The temperature value of pixel p. This represents the average temperature.
[0041] Furthermore, in this embodiment of the application, in step S111, the text analysis includes steps A1~A2: A1: Preprocess the text, including word segmentation, stop word filtering, and entity annotation, to generate a standardized text sequence.
[0042] A2: Input the BERT model's encoding layer, extract the weights of the "device anomaly" keywords, and generate a feature vector. ,in, This represents the weight of the keyword for the i-th type of fault. This represents the total number of faults.
[0043] In an optional implementation, in step S111, the text analysis further includes extracting term frequency-inverse document frequency (TF-IDF) features from the inspection report, classifying the text as normal or abnormal using an SVM model, and generating feature vectors based on keyword frequency and category probability.
[0044] In another optional implementation, in step S111, the text analysis may further include: pre-constructing a knowledge graph of power grid equipment faults, extracting entities and relationships through rule matching when parsing inspection reports, and generating feature vectors based on the weights of graph nodes.
[0045] In this embodiment of the application, step S200, the correction of the electrical parameters of the device through parameter fusion, includes steps S201 to S203: S201: Using the Sigmoid function obtained through training on historical expert-annotated samples, the multidimensional feature vector generated by text analysis is mapped to a device status score between 0 and 1. The score comprehensively reflects the probability that the device is in an abnormal state. The equipment status scoring formula is expressed as follows: in, Rate the equipment status. For the Sigmoid function, As weight, This is the feature vector of the equipment malfunction. For bias.
[0046] S202: The parameter correction process is driven by the local overheating index obtained from the scoring and spectrum processing. When either the score is not lower than 0.5 or the overheating index exceeds the preset safety threshold, the correction operation is triggered.
[0047] S203: When the correction operation is triggered, the electrical parameters are dynamically adjusted according to the real-time overheating of the equipment. Based on the measured maximum temperature of the key area obtained from the infrared spectrum and combined with the material properties, the resistance value of the conductor is increased. The dynamic correction mechanism ensures that when the equipment heats up due to fault or overload, the parameters in the power grid model can be updated synchronously, thus maintaining a high degree of consistency with the physical reality of the equipment. The corrected parameters and the ambient temperature prediction curve are packaged and output to the time-varying parameter updater.
[0048] Correction formula for wire resistance: in, This is the corrected wire resistance value. This is the standard resistance value. The temperature coefficient of the material. This is the highest temperature measured. This is a reference temperature.
[0049] In an optional implementation, in step S200, the correction of the electrical parameters of the device through parameter fusion further includes using the device status score and overheat index as fuzzy inputs, and outputting correction coefficients through a fuzzy rule base (when the score is high and the overheat index is high, the resistance correction amount is large) to dynamically adjust the electrical parameters.
[0050] In another optional implementation, in step S200, the correction of the electrical parameters of the device through parameter fusion may further include directly predicting the parameter correction value using a machine learning regression model, training a random forest or gradient boosting model, and outputting the corrected electrical parameter value using the device status score, overheat index and environmental data as features.
[0051] Furthermore, in this embodiment of the application, in step S200, the environmental perception technology includes steps S211-S212: S211: Through a distributed sensor network deployed at key nodes of the power transmission corridor, multi-dimensional meteorological data of temperature, humidity, wind speed, wind direction and air pressure are collected in real time at a high frequency of 5Hz.
[0052] S212: It adopts an LSTM network, inputs historical weather sequences (data from the past 2 hours), and outputs future temperature change curves through two stacked LSTM hidden layers (128 neurons per layer). After the data is cleaned, normalized and spatiotemporally aligned, it is pushed to the time-varying parameter updater via the OPC UA protocol.
[0053] In an optional implementation, in step S200, the environmental sensing technology further includes, after collecting historical meteorological data, fitting the temperature change pattern through an autoregressive integrated moving average (ARIMA) model and predicting future curves, with the data transmitted through a wireless sensor network and stored in a local database.
[0054] In another optional implementation, in step S200, the environmental sensing technology may further include integrating macroscopic meteorological data provided by NWP with local sensor data, performing data fusion through Kalman filtering, and outputting a temperature change curve.
[0055] Furthermore, in step S200, the dynamic updating of the physical parameters of the device includes steps S221-S222: S221: The corrected electrical parameters and the ambient temperature prediction curve are packaged together and sent as input to the time-varying parameter updater. Combined with the inherent properties of conductor tension, weight, span and thermal expansion coefficient, the conductor sag that evolves over time is dynamically calculated. The formula for calculating conductor sag is expressed as: in, Let sag of the conductor be at time t. This refers to the weight per unit length of the conductor. The span is the horizontal distance between the two suspension points of the conductor. Let be the instantaneous value of the conductor tension at time t. The coefficient of thermal expansion of the conductor. This is the predicted temperature change curve.
[0056] S222: The multi-physics coupling equations stored in the dynamic physical model library, such as the thermoelectric coupling model of the transformer, will be linked to environmental sensing data in real time, dynamically calculate the winding temperature, and inject the high-fidelity physical parameters calculated by the time-varying parameter updater into the dynamic physical model library to jointly drive the power grid digital twin to evolve in real time and accurately. The transformer thermoelectric coupling model is represented as follows: in, This represents the rate of change of winding temperature over time. This refers to the winding temperature of the transformer. The current flowing through the winding, The resistance of the winding, For the heat capacity of the transformer, Here is the ambient temperature, and here is the thermal resistance of the transformer.
[0057] In step S300, generating the digital twin of the power grid includes steps S301 to S303: S301: Obtain real-time topology connection relationships from the SCADA system, load matching device models from the dynamic physical model library, input updated dynamic parameters, and dynamically adjust the model's topology connections according to changes in switch status.
[0058] S302: After the model is built, the system uses the dynamic parameterized model to perform power flow calculations and solves for the simulated values of voltage, phase angle and power flow of branches of all nodes in the network.
[0059] S303: The system compares the simulated values with the real-time measurement values received from the SCADA system point by point. When the relative error at any point exceeds 5%, the parameter backtracking calibration mechanism is triggered to ensure that the digital twin and the physical power grid remain highly consistent. The relative error is expressed as: in, This is a relative error. For measurement value, These are simulation values. This is the reference power.
[0060] Furthermore, in step S300, the running steps include continuously running in 30-second cycles, receiving the latest environmental data in each cycle, calculating and updating the device parameters, injecting the parameters into the admittance matrix to drive the model evolution, continuously comparing and verifying the model simulation results with real-time measurement values, and adding confidence scores to the successfully verified dynamic parameter set.
[0061] When the confidence score of a certain set of parameters accumulates to exceed a preset threshold, it will be marked as a high-confidence parameter set and archived into the validated parameter library, realizing the self-improvement of model parameters and knowledge accumulation.
[0062] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.
[0063] Example 3, referring to Figure 4 This is the third embodiment of the present invention. This embodiment provides a power grid model construction system based on power transmission information, including a data acquisition module, a multimodal feature extraction module, an environmental perception module, a parameter fusion module, and a power grid maintenance and update module.
[0064] The data acquisition module is used to receive structured inspection report text through the API interface of the mobile inspection terminal, acquire raw thermal imaging video stream through the fiber optic interface of the infrared thermal imager, perform protocol conversion and encoding processing on the text data, extract key frames and compress the format of the video stream, and bind spatiotemporal tags to the text and image data through the data transmission controller, and transmit them to subsequent processing through a high-speed network.
[0065] The multimodal feature extraction module is used to parse the inspection report using natural language processing technology, identify abnormal keywords of equipment, and generate feature vectors representing the status of the equipment. The graph processing unit uses computer vision technology to analyze the infrared thermal imaging graph, extract temperature gradient distribution features, and calculate the local overheating index of key areas of the equipment. The synchronization controller ensures the temporal consistency of the text and graph data. When severe overheating is detected, the key fault mode scan is triggered to perform deep feature mining and fusion of multi-source data.
[0066] The environmental perception module is used to collect meteorological data such as temperature, humidity, wind speed, and air pressure in real time through a distributed sensor network deployed at key nodes of the power transmission corridor. It also uses a long short-term memory network to perform deep learning on historical meteorological sequences, establish a temperature change prediction model, and output temperature change curves for future time scales. The module also performs cleaning, normalization, and spatiotemporal alignment processing on the raw data.
[0067] The parameter fusion module is used to convert feature vectors into equipment status scores through a mapping function, determine whether to trigger electrical parameter correction based on the score results and local overheating index, and dynamically adjust the electrical parameters of the equipment according to material properties, temperature data and physical laws when the triggering conditions are met, and package the corrected parameters with the environmental prediction curve for output.
[0068] The maintenance and update grid module is used to dynamically calculate physical parameters that change over time based on environmental data, equipment status, and prediction curves through a time-varying parameter updater. The dynamic physical model library stores the multi-physics coupling equations of the equipment, updates the equipment status in real time through numerical solution methods, and injects the updated high-fidelity parameters into the grid digital twin.
[0069] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.
[0070] Example 4, the fourth embodiment of the present invention, differs from the previous three embodiments in that: If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0071] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.
[0072] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.
[0073] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.
Claims
1. A method for power grid model construction based on power transmission information, characterized in that: The method comprises the following steps: Real-time acquisition of power grid equipment data, multi-modal feature extraction of collected data, obtaining feature vector of equipment anomaly and overheat index based on temperature gradient distribution; Based on the feature vector and the overheat index, the electrical parameters of the equipment are corrected by parameter fusion, the environmental meteorological data are collected in real time, and the temperature change curve is predicted, based on the predicted temperature change curve and the corrected electrical parameters, the physical parameters of the equipment are dynamically updated; The updated physical parameters are injected into the multi-physical field coupling equation, the dynamic physical model library is updated, and the power grid digital twin is generated, the power grid simulation is carried out by using the power grid digital twin, and the simulation value is compared with the actual measured value, when the error exceeds the threshold, the parameter backtracking calibration is triggered.
2. The power grid model building method based on delivered power information according to claim 1, characterized in that: The multi-modal feature extraction includes text analysis of inspection report text, text content analysis by natural language processing technology, identification of equipment abnormal keywords, and generation of feature vector representing equipment abnormal state; The infrared thermal imaging spectrum is processed by spectrum processing, the temperature gradient distribution characteristics are extracted, and the local overheat index of the key area of the equipment is calculated based on the temperature distribution; Wherein, the text analysis and spectrum processing are carried out synchronously, when the time stamp difference of text and spectrum data is less than the set threshold, the fusion enable signal of feature vector is activated.
3. The power grid model construction method based on delivered power information according to claim 2, characterized in that: The electrical parameters of the equipment are corrected by parameter fusion, which includes: The feature vector is converted into equipment state score by mapping function, which represents the probability of equipment being in abnormal state; Based on the equipment state score and the local overheat index, it is judged whether to trigger the electrical parameter correction, when the equipment state score exceeds the preset threshold or the local overheat index exceeds the safety threshold, the electrical parameters of the equipment are dynamically corrected; The corrected electrical parameters and the environmental temperature prediction curve are packaged synchronously as the input of time-varying parameter update.
4. The power grid model building method based on delivered power information according to claim 3, characterized in that: The dynamic updating of the physical parameters of the equipment includes real-time acquisition of environmental meteorological data by environmental perception technology, and prediction of future temperature change curve based on historical meteorological sequence; Based on the predicted temperature change curve and the corrected electrical parameters, the physical parameters of the equipment are dynamically calculated by time-varying parameter update, and the updated parameters are injected into the dynamic physical model library.
5. The power grid model building method based on delivered power information according to claim 4, characterized in that: The environmental perception technology includes using temperature and humidity sensor network to collect environmental data at a frequency of 5Hz, the sensors are deployed at key nodes of power transmission corridor, using long short-term memory network to predict temperature change curve, inputting historical time series data, outputting future temperature prediction value, and performing cleaning, normalization and space-time alignment processing on original data to construct training sample, using historical sequence in sliding time window to predict future temperature curve; The electrical parameters of the equipment are corrected by parameter fusion, which includes using Sigmoid function to map feature vector into equipment state score, and triggering electrical parameter correction based on equipment state score and local overheat index; The equipment state score formula is represented as: wherein, score a device state, sigmoid function, weight, feature vector of a device anomaly, bias; The electrical parameter correction includes adjusting the conductor resistance according to the maximum temperature and the material temperature coefficient, and the formula is represented as: wherein, is the corrected wire resistance value, is the standard resistance value, is the material temperature coefficient, is the measured maximum temperature, is the reference temperature.
6. The power grid model building method based on delivered power information according to claim 5, characterized in that: The dynamic updating device physical parameter includes, time-varying parameter updates dynamic calculation of conductor sag, based on conductor tension instantaneous value, unit length weight, span, thermal expansion coefficient and predicted temperature change curve calculation, real-time correlation of environmental perception data in the calculation process, and the updated parameter is injected into the admittance matrix of the power grid digital twin; The conductor sag formula is calculated as: wherein, is the sag of the conductor at time t, is the weight per unit length of the conductor, is the span, i.e. the horizontal distance between the two suspension points of the conductor, is the instantaneous value of the tension of the conductor at time t, is the thermal expansion coefficient of the conductor, is the predicted temperature profile; The dynamic physical model library stores the multi-physical field coupling equation of the device, the model equation is real-time correlated with the environmental temperature data collected by environmental perception, and the winding temperature state is updated by numerical solution; The multi-physical field coupling equation formula is represented as: wherein, is the rate of change of the winding temperature with time, is the winding temperature of the transformer, is the current flowing through the winding, is the resistance of the winding, is the thermal capacitance of the transformer, is the ambient temperature, and is the thermal resistance of the transformer.
7. The power grid model building method based on delivered power information according to claim 6, characterized in that: The generation of the power grid digital twin includes, based on the updated model library, generating dynamic topology and electrical parameters, performing measurement and simulation error verification, and obtaining real-time switch state and measurement value, dynamically adjusting the model topology connection relationship, using the dynamic parameterized model to perform power system load flow calculation, solving network node voltage and branch power flow, and comparing the measurement value with the simulation value at the same time and the same physical location point by point, when the relative error between the simulation value and the measurement value exceeds 5%, triggering parameter backtracking calibration; Wherein, the calibration trigger instruction includes, the specific measurement point position, time point, measurement value and simulation value, error value, current used device model identifier and dynamic parameter value of the out-of-limit error.
8. A power grid model building system based on power transmission information, which applies the power grid model building method based on power transmission information according to any one of claims 1 to 7, characterized in that, It includes a data acquisition module, a multi-modal feature extraction module, an environmental perception module, a parameter fusion module, and a power grid maintenance and update module; The data acquisition module is used for receiving structured inspection report text through the API interface of the mobile inspection terminal, and obtaining original thermal imaging video stream through the optical fiber interface of the infrared thermal imager, performing protocol conversion and encoding processing on the text data, performing key frame extraction and format compression on the video stream, and binding the text and image data with space-time tags through the data transmission controller, and transmitting to the subsequent processing through the high-speed network; The multi-modal feature extraction module is used for analyzing the inspection report by using natural language processing technology, identifying device abnormal keywords, and generating feature vectors representing device state, analyzing infrared thermal imaging atlas by using computer vision technology through the atlas processing unit, extracting temperature gradient distribution features and calculating local overheating index of device key area, ensuring the time sequence consistency of text and atlas data through the synchronous controller, and triggering key fault mode scanning when serious overheating is detected, and performing deep feature mining and fusion on multi-source data; The environmental perception module is used for collecting temperature, humidity, wind speed, air pressure meteorological data in real time through the distributed sensor network deployed at the key nodes of the power transmission corridor, and using long short-term memory network to perform deep learning on historical meteorological sequence, establishing a temperature change prediction model, outputting temperature change curve at future time scale, and processing the original data through cleaning, normalization and space-time alignment. The parameter fusion module is configured to convert the feature vector into a device state score through a mapping function, determine whether to trigger electrical parameter correction based on a score result and a local overheating index, dynamically adjust electrical parameters of the device according to material characteristics, temperature data and physical laws when a trigger condition is met, and pack and output the corrected parameters and an environmental prediction curve. The maintenance and update power grid module is configured to dynamically calculate time-varying physical parameters based on environmental data, device states and a prediction curve through a time-varying parameter updater, store a multi-physical field coupling equation of the device in a dynamic physical model library, update the device state in real time through a numerical solution method, and inject updated high-fidelity parameters into a power grid digital twin. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-8 when the computer program is executed by the processor. The processor executes the computer program to implement the steps of the power grid model construction method based on transmission power information according to any one of claims 1 to 7.
10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the power grid model construction method based on transmission power information according to any one of claims 1 to 7.