Power risk prediction method, device and equipment for power transmission and transformation construction site and medium

Abstract

The application relates to a power transmission and transformation construction site electric risk prediction method, device, equipment and medium, and the method comprises the following steps: constructing each to-be-detected sample data into a graph node of a first graph structure; according to parameter attribute information and a preset risk correlation rule, it is judged whether the parameter attribute information between any two graph nodes in the first graph structure satisfies the preset risk correlation rule, if yes, an edge connection is established between the two graph nodes to construct a second graph structure; the second graph structure is input into a second electric risk prediction model trained by using a historical sample data set, and the electric shock risk probability, the overload risk probability and the short-circuit risk probability of a to-be-detected power transmission and transformation line are output; according to the electric shock risk probability, the overload risk probability and the short-circuit risk probability, it is judged whether the to-be-detected power transmission and transformation line has the electric shock risk, the overload risk and the short-circuit risk respectively. The application can realize the active early warning mode from the risk prediction to the early prevention and control of the power transmission and transformation construction site.

Description

Technical Field

[0001] This application relates to the field of power transmission and transformation safety monitoring, and in particular to a method for predicting power consumption risks at power transmission and transformation construction sites, a corresponding device, electronic equipment, and a computer-readable storage medium. Background Technology

[0002] The electrical environment at power transmission and transformation line construction sites is complex and variable, with dense temporary power lines, frequent equipment start-ups and shutdowns, and long-term exposure to harsh environments such as rainy season humidity and strong winds and dust, resulting in persistently high electrical safety risks. These risks mainly manifest as three core categories: electric shock, overload, and short circuit. Electric shock accidents are often caused by insulation failure and improper operation at heights, directly endangering the personal safety of construction workers. Overload risks stem from large fluctuations in temporary power loads and non-standard wiring, easily causing equipment damage and accelerated line aging. Short circuit accidents are often induced by excessive ambient humidity and damaged line insulation, potentially leading to electrical fires and significant property damage.

[0003] Traditional power transmission and transformation construction site electrical protection mainly relies on conventional devices such as fuses and ordinary circuit breakers. These devices can only provide passive protection against simple faults such as overcurrent and short circuits, and have the following technical defects, including: Firstly, the response is slow, and the speed of identifying risks such as instantaneous overload and hidden leakage is slow, making it difficult to meet the millisecond-level protection requirements. Secondly, the monitoring dimensions are singular, focusing only on electrical operating parameters and failing to consider the coupling influence of construction condition parameters, environmental parameters, and electrical operating parameters, resulting in a large number of monitoring blind spots; Third, it has a high rate of false alarms and is easily interfered with in harsh environments. It cannot accurately distinguish between real risks and parameter fluctuations, which affects the continuity of construction and may lead to safety accidents due to missed judgments.

[0004] In summary, existing technologies for power transmission and transformation construction site electrical protection can only achieve passive protection against simple faults such as overcurrent and short circuits, and suffer from problems such as delayed response, limited monitoring dimensions, and high false tripping rate. The applicant has made corresponding explorations to address these issues. Summary of the Invention

[0005] The purpose of this application is to solve the above-mentioned problems by providing a method for predicting power consumption risks at power transmission and transformation construction sites, a corresponding device, electronic equipment, and a computer-readable storage medium.

[0006] To achieve the various objectives of this application, the following technical solution is adopted: A method for predicting electricity consumption risks at power transmission and transformation construction sites, proposed to meet one of the purposes of this application, includes: Obtain a historical sample dataset of the target transmission and transformation line. The historical sample dataset includes multiple historical sample data and their corresponding sample labels. The historical sample data includes time-series data constructed from electrical operating parameters, construction condition parameters and environmental parameters corresponding to multiple consecutive time steps. The sample labels represent whether there is a risk of electric shock, overload risk and short circuit risk. Multiple test sample data from the test sample dataset are input into the first electricity risk prediction model trained using the historical sample dataset to extract the hidden correlation features between electrical operating parameters, construction condition parameters, and environmental parameters at each time step. Each sample data in the dataset to be tested is constructed as a graph node of the first graph structure. The parameters in the sample data to be tested are used as the feature vectors of the graph node. The hidden correlation features between the electrical operation parameters, construction condition parameters and environmental parameters at each time step in each sample data are used as the parameter attribute information of the graph node. Based on the parameter attribute information and the preset risk association rule, determine whether the parameter attribute information of any two graph nodes in the first graph structure satisfies the preset risk association rule. If it does, establish an edge connection between the two graph nodes to construct the second graph structure. The second graph structure is input into the second electricity risk prediction model trained using the historical sample dataset, and the corresponding electric shock risk probability, overload risk probability, and short circuit risk probability of the transmission and transformation line to be detected are output. Based on the probability of electric shock risk, the probability of overload risk, and the probability of short circuit risk, it is determined whether the transmission and transformation line to be tested has the risk of electric shock, overload risk, and short circuit risk, respectively.

[0007] Optionally, the steps for training the first electricity consumption risk prediction model include: The historical sample dataset is divided into a training set, a test set, and a validation set according to a preset ratio; Initialize the network parameters of the first electricity risk prediction model, setting the number of network layers to 3, the number of neurons in the hidden layer to 128, the dropout rate to 0.2, and the learning rate to 0.001; The multiple historical sample data are input into the initialized first electricity risk prediction model. The sample labels are used as supervision signals. The prediction error is calculated using the cross-entropy loss function. The model network parameters are adjusted by backpropagation through the Adam optimization algorithm. The model performance is evaluated using a validation set after a certain number of iterations. If the loss value on the validation set does not show a decreasing trend over multiple consecutive iterations, training is stopped. The test set is input into the first electricity risk prediction model after training, and the accuracy of the model in extracting hidden related features is calculated. If the accuracy is greater than or equal to a preset threshold, it is determined that the first electricity risk prediction model has been trained to a convergent state.

[0008] Optionally, the step of determining whether the parameter attribute information of any two graph nodes in the first graph structure satisfies the preset risk association rule based on the parameter attribute information and the preset risk association rule, and establishing an edge connection between the two graph nodes to construct the second graph structure, includes: If the parameter attribute information of the first node includes a residual current greater than or equal to 30mA and a hidden associated feature of decreased insulation performance, and the parameter attribute information of the second node includes a worker being in a high-altitude working state and a hidden associated feature of increased worker contact risk, then the electric shock risk association rule is satisfied. If the parameter attribute information of the first node includes a loop current greater than or equal to 130% of the rated value and a hidden associated feature of continuous overload, and the parameter attribute information of the second node includes a line temperature greater than or equal to 60℃ and a hidden associated feature of poor line heat dissipation, then the overload risk association rule is satisfied. If the parameter attribute information of the node in the first figure includes a sudden change in short-circuit current and the hidden associated feature is line insulation damage, and the parameter attribute information of the node in the second figure includes line humidity greater than or equal to 85% and the hidden associated feature is enhanced environmental conductivity, then the short-circuit risk association rule is satisfied.

[0009] Optionally, the step of inputting the second graph structure into the second electricity risk prediction model trained using the historical sample dataset, and outputting the corresponding electric shock risk probability, overload risk probability, and short circuit risk probability of the transmission and transformation line to be detected, includes: The second graph structure is input into the second electricity risk prediction model, which has been trained to convergence using the historical sample dataset. Feature aggregation is performed on the feature vectors and parameter attribute information of each graph node in the second graph structure. Differentiated attention weights are dynamically assigned to the neighboring nodes of each node by calculating the correlation between nodes, focusing on the combination of nodes closely related to electric shock risk, overload risk, and short circuit risk. By iteratively updating the aggregated features through a multi-layer attention mechanism, a high-level feature representation that can reflect the core characteristics of the three types of electricity consumption risks is gradually extracted. The high-level feature representation is input into the classifier in the second electricity risk prediction model. The high-level feature representation is then mapped using the softmax function to output the electric shock risk probability, overload risk probability, and short circuit risk probability of the transmission and transformation line to be detected.

[0010] Optionally, the step of determining whether the transmission and transformation line under test has the risk of electric shock, overload, and short circuit based on the probability of electric shock risk, the probability of overload risk, and the probability of short circuit risk, respectively, includes: The probability of electric shock risk, the probability of overload risk, and the probability of short circuit risk of the transmission and transformation line to be tested are compared with the corresponding judgment thresholds respectively. If the probability of electric shock exceeds the electric shock risk determination threshold, the transmission and transformation line under test is determined to have an electric shock risk; if the probability of overload risk exceeds the overload risk determination threshold, the transmission and transformation line under test is determined to have an overload risk; if the probability of short circuit risk exceeds the short circuit risk determination threshold, the transmission and transformation line under test is determined to have a short circuit risk.

[0011] Optionally, the electrical operating parameters include incoming line voltage, loop voltage, equipment terminal voltage, incoming line current, loop current, equipment terminal current, residual current, short-circuit current, and grounding resistance; the construction condition parameters include high-altitude operation status and equipment start-stop status; and the environmental parameters include line temperature, line humidity, and wind speed.

[0012] Optionally, the basic network architecture of the first electricity risk prediction model is a long short-term memory neural network; the basic network architecture of the second electricity risk prediction model is a graph attention neural network model.

[0013] A power supply risk prediction device for power transmission and transformation construction sites, provided for another purpose of this application, includes: The dataset acquisition module is configured to acquire the historical sample dataset of the target transmission and transformation line. The historical sample dataset includes multiple historical sample data and their corresponding sample labels. The historical sample data includes time-series data constructed from electrical operating parameters, construction condition parameters and environmental parameters corresponding to multiple consecutive time steps. The sample labels represent whether there is a risk of electric shock, overload risk and short circuit risk. The hidden feature extraction module is configured to input multiple test sample data from the test sample dataset into the first electricity risk prediction model trained using the historical sample dataset, so as to extract the hidden correlation features between electrical operating parameters, construction condition parameters, and environmental parameters at each time step; The graph node construction module is configured to construct graph nodes of the first graph structure from each sample data in the sample dataset to be detected. The parameters in the sample data to be detected are used as the feature vectors of the graph nodes, and the hidden correlation features between the electrical operation parameters, construction condition parameters and environmental parameters at each time step in each sample data are used as the parameter attribute information of the graph nodes. The graph structure construction module is configured to determine whether the parameter attribute information between any two graph nodes in the first graph structure satisfies the preset risk association rule based on the parameter attribute information and the preset risk association rule. If it does, an edge connection is established between the two graph nodes to construct the second graph structure. The risk probability prediction module is configured to input the second graph structure into the second power consumption risk prediction model trained using the historical sample dataset, and output the corresponding electric shock risk probability, overload risk probability, and short circuit risk probability of the transmission and transformation line to be detected; based on the electric shock risk probability, overload risk probability, and short circuit risk probability, it determines whether the transmission and transformation line to be detected has electric shock risk, overload risk, and short circuit risk, respectively.

[0014] An electronic device provided for another purpose of this application includes a central processing unit and a memory, wherein the central processing unit is used to invoke and run a computer program stored in the memory to perform the steps of the power transmission and transformation construction site power risk prediction method described in this application.

[0015] A computer-readable storage medium is provided for another purpose of this application, which stores, in the form of computer-readable instructions, a computer program implemented according to the power supply risk prediction method for power transmission and transformation construction sites, which, when called by a computer, executes the steps included in the corresponding method.

[0016] Compared to existing technologies, this application addresses the shortcomings of traditional power transmission and transformation site electrical protection, which only provides passive protection against simple faults such as overcurrent and short circuits. These traditional technologies suffer from issues such as delayed response, limited monitoring dimensions, and high false tripping rates. This application offers the following advantages, including but not limited to: Firstly, this application utilizes a long short-term memory neural network to mine the hidden correlation features of electrical operating parameters, construction condition parameters, and environmental parameters in the time dimension, completely breaking through the limitations of traditional devices that only focus on electrical parameters, and filling the monitoring gap under harsh environments and complex working conditions.

[0017] Secondly, traditional devices rely on passive triggering after a fault occurs. This application trains a long short-term memory neural network and a graph attention neural network using historical sample datasets to discover the evolution patterns of risks in advance and output the probabilities of electric shock, overload, and short circuit risks, thus realizing an active early warning mode of "risk prediction to early prevention".

[0018] Thirdly, by constructing a second graph structure, the scattered sample data is transformed into a structured risk association network. Then, through a graph attention neural network, attention weights are dynamically allocated to focus on high-risk node combinations, accurately distinguishing between real risks and parameter fluctuations, effectively reducing the false alarm rate in harsh environments, and balancing construction continuity with the reliability of safety protection. Attached Figure Description

[0019] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a flowchart illustrating the method for predicting electricity consumption risks at power transmission and transformation construction sites in this application. Figure 2 This is a schematic diagram of the process for training the first electricity risk prediction model in an embodiment of this application; Figure 3 This is a flowchart illustrating the process of constructing the second graph structure in an embodiment of this application; Figure 4 This is a flowchart illustrating the process of determining the probability of electric shock, overload, and short circuit of the transmission and transformation line to be tested in this application embodiment. Figure 5 This is a flowchart illustrating the process of determining whether a transmission and transformation line under test has the risk of electric shock, overload, or short circuit, as described in this application embodiment. Figure 6 This is a schematic diagram of the power risk prediction device for power transmission and transformation construction sites in the embodiments of this application; Figure 7 This is a schematic diagram of the structure of the computer device in the embodiments of this application. Detailed Implementation

[0020] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application.

[0021] Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the term “comprising” as used in this application means the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when we say an element is “connected” or “coupled” to another element, it can be directly connected or coupled to the other element, or there may be intermediate elements. Furthermore, “connected” or “coupled” as used herein can include wireless connections or wireless coupling. The term “and / or” as used herein includes all or any units and all combinations of one or more associated listed items.

[0022] Those skilled in the art will understand that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0023] Those skilled in the art will understand that the terms "client," "terminal," and "terminal device" as used herein include both devices that receive wireless signals, devices that only possess wireless signal receiver capabilities without transmission capabilities, and devices with receiving and transmitting hardware, devices that have receiving and transmitting hardware capable of bidirectional communication over a bidirectional communication link. Such devices may include: cellular or other communication devices such as personal computers or tablets, having single-line displays, multi-line displays, or cellular or other communication devices without multi-line displays; PCS (Personal Communications Service) that can combine voice, data processing, fax, and / or data communication capabilities; PDAs (Personal Digital Assistants) that may include radio frequency receivers, pagers, internet / intranet access, web browsers, notebooks, calendars, and / or GPS (Global Positioning System) receivers; and conventional laptops and / or handheld computers or other devices that have and / or include radio frequency receivers. As used herein, "client," "terminal," and "terminal device" can be portable, transportable, installed in a means of transportation (air, sea, and / or land), or suitable and / or configured to operate locally and / or in a distributed manner, operating in any other location on Earth and / or in space. "Client," "terminal," and "terminal device" as used herein can also be a communication terminal, an internet access terminal, or a music / video playback terminal, such as a PDA, a MID (Mobile Internet Device), and / or a mobile phone with music / video playback capabilities, or a smart TV, set-top box, etc.

[0024] The hardware referred to by the names "server," "client," and "service node" in this application is essentially an electronic device with the equivalent capabilities of a personal computer. It is a hardware device with the necessary components revealed by the von Neumann architecture, such as a central processing unit (including an arithmetic logic unit and a control unit), memory, input devices, and output devices. The computer program is stored in its memory, and the central processing unit loads the program stored in the secondary storage into the main memory to run it, execute the instructions in the program, and interact with the input and output devices to complete specific functions.

[0025] It should be noted that the concept of "server" used in this application can also be extended to the case of server clusters. Based on the network deployment principles understood by those skilled in the art, the servers should be logically divided. Physically, these servers can be independent of each other but accessible through interfaces, or they can be integrated into a single physical computer or a computer cluster. Those skilled in the art should understand this flexibility and should not use it to constrain the implementation of the network deployment method in this application.

[0026] One or more of the technical features of this application, unless explicitly specified herein, can be deployed on a server and accessed by a client remotely calling the online service interface provided by the server, or can be directly deployed and run on a client for access.

[0027] Unless otherwise specified, the neural network models referenced or potentially referenced in this application may be deployed on a remote server and invoked remotely on the client, or deployed on a client with the capability to invoke directly. In some embodiments, when running on the client, the corresponding intelligence may be acquired through transfer learning in order to reduce the requirements on the client's hardware resources and avoid excessive consumption of the client's hardware resources.

[0028] Unless otherwise specified, all data involved in this application may be stored remotely on a server or on a local terminal device, as long as it is suitable for use by the technical solution of this application.

[0029] Those skilled in the art will understand that although the various methods in this application are described based on the same concept and thus present commonality among them, they can be performed independently unless otherwise specified. Similarly, the various embodiments disclosed in this application are all based on the same inventive concept; therefore, concepts expressed in the same way, as well as concepts that are appropriately changed for convenience but are expressed differently, should be understood equivalently.

[0030] Unless otherwise expressly stated, the various embodiments disclosed in this application can be combined in a cross-cutting manner to flexibly construct new embodiments, as long as such combination does not depart from the inventive spirit of this application and can meet the needs of the prior art or solve a certain deficiency in the prior art. Those skilled in the art should be aware of such modifications.

[0031] Please see Figure 1 In one embodiment of the power transmission and transformation construction site power risk prediction method of this application, the method includes: Step S10: Obtain the historical sample dataset of the target transmission and transformation line. The historical sample dataset includes multiple historical sample data and their corresponding sample labels. The historical sample data includes time series data constructed from electrical operating parameters, construction condition parameters and environmental parameters corresponding to multiple consecutive time steps. The sample labels represent whether there is a risk of electric shock, overload risk and short circuit risk. The power risk prediction system for power transmission and transformation construction sites in the terminal equipment can acquire historical sample datasets of the target power transmission and transformation lines. These historical sample datasets include multiple historical sample data points and their corresponding sample labels. The historical sample data comprises time-series data constructed from electrical operating parameters, construction condition parameters, and environmental parameters corresponding to multiple consecutive time steps. The sample labels indicate whether there is a risk of electric shock, overload, or short circuit. The electrical operating parameters include incoming line voltage, loop voltage, equipment terminal voltage, incoming line current, loop current, equipment terminal current, residual current, short-circuit current, and grounding resistance. The construction condition parameters include high-altitude operation status and equipment start-up and shutdown status. The environmental parameters include line temperature, line humidity, and wind speed.

[0032] Specifically, the historical sample data can be obtained through mature monitoring equipment at power transmission and transformation construction sites, including intelligent distribution boxes with IP65 protection rating, multi-circuit electric shock protection devices, environmental sensors, etc., which can stably acquire historical operating data including electrical operating parameters, construction condition parameters, and environmental parameters under multiple scenarios. The sample labels can be marked in several ways: first, equipment protection action records, such as when the equivalent shunt protection device triggers electric shock protection, the corresponding sample data is marked "risk of electric shock exists"; second, manual operation and maintenance records, such as the parameter sequence corresponding to historical overload faults is marked "risk of overload exists", to ensure the accuracy of the labels.

[0033] Step S20: Input multiple test sample data from the test sample dataset into the first electricity risk prediction model trained using the historical sample dataset to extract the hidden correlation features between electrical operating parameters, construction condition parameters, and environmental parameters at each time step. After obtaining the historical sample dataset of the target transmission and transformation line, multiple test sample data from the test sample dataset are input into the first power consumption risk prediction model trained using the historical sample dataset to extract the hidden correlation features between electrical operating parameters, construction condition parameters, and environmental parameters at each time step; wherein, the basic network architecture of the first power consumption risk prediction model is a long short-term memory neural network. Specifically, leveraging the time-series data processing advantages of Long Short-Term Memory Neural Networks (LSTM), dynamic correlation patterns that are difficult to directly identify manually but are crucial for risk prediction are extracted from the multi-dimensional parameters of the samples to be tested. The gating mechanism of LSTM (input gate, forget gate, output gate) can effectively capture the long-term and short-term dependencies of parameters over time, such as the lag correlation of continuously increasing humidity within 10 minutes leading to a gradual exceedance of residual current, thus avoiding the shortcomings of traditional neural networks in handling time-series data. In power transmission and transformation construction sites, there are interferences such as rainy season humidity and strong winds and dust. LSTM can filter out short-term fluctuation noise (such as instantaneous current surges) through the forget gate, focusing on changes in key parameters that have a lasting impact on risk (such as the trend of current continuously exceeding the rated value during overload).

[0034] Furthermore, the hidden correlation features extracted from the sample data to be tested by the Long Short-Term Memory Neural Network (LSTM) are not isolated features of a single parameter, but rather coupled correlations of multiple types of parameters in the time dimension, such as the correlation between electrical operating parameters and operating condition parameters, the correlation between electrical operating parameters and environmental parameters, and the synergistic correlation of multiple types of parameters including electrical operating parameters, operating condition parameters, and environmental parameters.

[0035] In the correlation between electrical operating parameters and working condition parameters, such as the hidden correlation features of workers being in a high-altitude working state and residual current being greater than or equal to 30mA, the Long Short-Term Memory Neural Network (LSTM) can capture the time lag pattern of residual current exceeding the standard after the start of high-altitude work. In the correlation between electrical operating parameters and environmental parameters, such as the hidden correlation features of "line temperature greater than or equal to 60℃ and circuit current greater than or equal to 130% of the rated value", the Long Short-Term Memory Neural Network (LSTM) can extract the dynamic law of "the probability of current exceeding the rated value increases by 5℃ for every 5℃ increase in temperature". In the collaborative correlation of multiple parameters, including electrical operating parameters, working condition parameters, and environmental parameters, such as the hidden correlation features of "high humidity greater than or equal to 85%, frequent equipment start-ups and shutdowns, and sudden changes in short-circuit current," Long Short-Term Memory (LSTM) neural networks can uncover the temporal synchronization patterns when multiple parameters are simultaneously abnormal, providing high-value graph node attribute information for subsequent graph structure construction. Therefore, LSTM is used as the first power consumption risk prediction model in this application. Once the first power consumption risk prediction model is trained to convergence, it can be put into production use to extract the hidden correlation features between electrical operating parameters, construction working condition parameters, and environmental parameters at each time step.

[0036] In some embodiments, please refer to Figure 2 The steps for training the first electricity consumption risk prediction model include: Step S201: Divide the historical sample dataset into a training set, a test set, and a validation set according to a preset ratio; The historical sample dataset is split into training, validation, and test sets according to a preset ratio to avoid model overfitting caused by a single dataset. This ensures that the model can still accurately discover hidden correlation features on unseen test data and adapt to the parameter variation patterns of different construction scenarios (such as rainy season and high-altitude operations) of power transmission and transformation lines.

[0037] Step S202: Initialize the network parameters of the first electricity risk prediction model, setting the number of network layers to 3, the number of neurons in the hidden layer to 128, the dropout rate to 0.2, and the learning rate to 0.001; The network has clearly defined parameters such as the number of layers, neurons, dropout rate, and learning rate, which are consistent with the feature dimensions of power transmission and transformation time series data. The dropout rate of 0.2 effectively suppresses overfitting, and the learning rate of 0.001 balances training speed and convergence accuracy, reduces invalid iterations, and adapts to the dynamic coupling characteristics of construction site data.

[0038] Step S203: Input the multiple historical sample data into the initialized first electricity risk prediction model, use the sample label as the supervision signal, use the cross-entropy loss function to calculate the prediction error, and use the Adam optimization algorithm to backpropagate and adjust the model network parameters. Using sample labels as supervision signals, and combining the cross-entropy loss function with the Adam optimization algorithm, the model can accurately learn the correlation between "parameter-time-risk", such as the correlation between high humidity and excessive residual current and electric shock risk, thus avoiding the shortcomings of traditional models in capturing dynamic correlations.

[0039] Step S204: Use the validation set to evaluate the model performance after a certain number of iterations. If the loss value on the validation set does not show a decreasing trend for several consecutive iterations, then stop training. By monitoring model performance in real time using the validation set, training is stopped when the loss does not decrease for several consecutive rounds. This ensures that the model learns effective features fully, prevents overfitting that leads to a decrease in generalization ability, and adapts to noise interference in construction site data (such as parameter fluctuations caused by strong winds).

[0040] Step S205: Input the test set into the trained first electricity risk prediction model, calculate the accuracy of the model in extracting hidden related features, and if the accuracy is greater than or equal to a preset threshold, determine that the first electricity risk prediction model has been trained to a convergent state.

[0041] Step S30: Construct each sample data to be detected in the sample dataset to be detected as a graph node of the first graph structure, wherein each parameter in the sample data to be detected is used as the feature vector of the graph node, and the hidden correlation features between the electrical operation parameters, construction condition parameters and environmental parameters at each time step in each sample data are used as the parameter attribute information of the graph node. Multiple test sample data from the test sample dataset are input into the first electricity risk prediction model trained using the historical sample dataset. After extracting the hidden correlation features between electrical operating parameters, construction condition parameters, and environmental parameters at each time step, each test sample data in the test sample dataset is constructed as a graph node of the first graph structure. The parameters in the test sample data are used as the feature vectors of the graph nodes, and the hidden correlation features between electrical operating parameters, construction condition parameters, and environmental parameters at each time step in each sample data are used as the parameter attribute information of the graph node. Specifically, the parameters in the sample data to be tested (incoming voltage, residual current, high-altitude operation status, equipment start-up and shutdown status, line temperature and line humidity, etc.) are directly used as feature vectors of graph nodes, providing quantitative basic information support for the nodes and ensuring that the data integrity is consistent with the dimensions collected by the on-site monitoring equipment.

[0042] The hidden correlation features of each sample data, such as "high humidity associated with excessive residual current" and "high-altitude operation associated with abnormal grounding resistance", are used as the parameter attribute information of the nodes. These features are dynamic correlation patterns that are difficult to be directly identified by humans and are mined by the first power consumption risk prediction model (LSTM). They provide key basis for subsequent judgment of risk correlation between nodes.

[0043] The hidden correlation features output from the first power consumption risk prediction model are transformed into attribute information of graph nodes. This allows the graph structure to focus on the core correlation patterns related to risks, avoiding the problem of inaccurate correlation judgment caused by traditional graph structures relying solely on surface parameters. The feature vectors of graph nodes ensure the integrity of node information, while the parameter attribute information (hidden correlation features) provides a basis for judgment on preset risk correlation rules. For example, it can determine whether two nodes are coupled by "overload correlation feature and high temperature correlation feature", forming a hierarchical protection logic of "one-loop fault, multi-loop early warning", laying the foundation for building the second graph structure.

[0044] Step S40: Based on the parameter attribute information and the preset risk association rule, determine whether the parameter attribute information of any two graph nodes in the first graph structure satisfies the preset risk association rule. If it does, establish an edge connection between the two graph nodes to construct the second graph structure. After constructing each sample data to be detected in the dataset as a graph node of the first graph structure, based on the parameter attribute information and the preset risk association rule, it is determined whether the parameter attribute information between any two graph nodes in the first graph structure satisfies the preset risk association rule. If it does, an edge connection is established between the two graph nodes to construct the second graph structure. In some embodiments, please refer to Figure 3 The step of determining whether the parameter attribute information of any two graph nodes in the first graph structure satisfies the preset risk association rule based on the parameter attribute information and the preset risk association rule, and establishing an edge connection between the two graph nodes to construct the second graph structure, includes: Step S401: If the parameter attribute information of the first graph node includes a residual current greater than or equal to 30mA and a hidden associated feature of decreased insulation performance, and the parameter attribute information of the second graph node includes a worker being in a high-altitude working state and a hidden associated feature of increased worker contact risk, then it is determined that the electric shock risk association rule is satisfied. Step S402: If the parameter attribute information of the first diagram node includes a loop current greater than or equal to 130% of the rated value, and the hidden associated feature is that the load is continuously exceeding the standard, and the parameter attribute information of the second diagram node includes a line temperature greater than or equal to 60℃, and the hidden associated feature is that the line heat dissipation is poor, then it is determined that the overload risk association rule is met. Step S403: If the parameter attribute information of the first graph node includes a sudden change in short-circuit current and the hidden associated feature is line insulation damage, and the parameter attribute information of the second graph node includes line humidity greater than or equal to 85% and the hidden associated feature is enhanced environmental conductivity, then it is determined that the short-circuit risk association rule is satisfied.

[0045] As can be seen from steps S401 to S403 above, the system focuses on multiple core risks such as electric shock risk, overload risk, and short circuit risk. Each risk association rule corresponds to a typical risk scenario at the power transmission and transformation construction site. For example, high-altitude operations are associated with leakage current, high loads are associated with high line temperatures, and insulation damage is associated with high humidity. Moreover, the parameter thresholds (residual current ≥30mA, temperature ≥60℃, humidity ≥85%) are consistent with the equipment protection standards to ensure the scientific nature of the association judgment.

[0046] The risk association rule combines explicit parameter thresholds (such as residual current, loop current, etc.) and implicit association features (such as insulation performance degradation, continuous overload, etc.), which avoids the one-sidedness of single parameter judgment and utilizes the deep patterns mined by the first model, making the node association more in line with the risk evolution logic and solving the shortcomings of traditional rules that rely only on surface data.

[0047] Multiple risk association rules correspond to specific risk types, making the node associations of different risks in the second diagram structure clearly distinguishable. The subsequent second power consumption risk prediction model can focus on the node combinations of various risks, improve the accuracy of electric shock, overload, and short circuit risk probability prediction, and carry out multi-circuit linkage protection and accurate risk early warning.

[0048] Step S50: Input the second graph structure into the second power consumption risk prediction model trained using the historical sample dataset, and output the corresponding electric shock risk probability, overload risk probability, and short circuit risk probability of the transmission and transformation line to be detected; based on the electric shock risk probability, overload risk probability, and short circuit risk probability, determine whether the transmission and transformation line to be detected has electric shock risk, overload risk, and short circuit risk respectively.

[0049] Based on the parameter attribute information and preset risk association rules, it is determined whether the parameter attribute information of any two graph nodes in the first graph structure satisfies the preset risk association rules. If they do, an edge connection is established between the two graph nodes to construct a second graph structure. The second graph structure is then input into a second electricity risk prediction model trained using the historical sample dataset. This model outputs the corresponding electric shock risk probability, overload risk probability, and short circuit risk probability of the transmission and transformation line to be detected. Based on the electric shock risk probability, overload risk probability, and short circuit risk probability, it is determined whether the transmission and transformation line to be detected has electric shock risk, overload risk, and short circuit risk, respectively. The basic network architecture of the second electricity risk prediction model is a graph attention neural network (GAT) model.

[0050] Specifically, power supply risks in transmission and transformation lines (such as electric shock and overload) are induced by a combination of electrical operating parameters, working condition parameters, and environmental parameters (e.g., excessive residual current associated with high-altitude operations → risk of electric shock). The Graph Attention Neural Network (GAT) model, by dynamically allocating attention weights, can automatically focus on high-risk associated graph node combinations (e.g., "short-circuit current mutation node + high humidity node"), accurately uncovering hidden coupling patterns between parameters and solving the problem of traditional models' inadequacy in identifying complex associations.

[0051] Furthermore, the preceding steps have transformed the sample data to be detected into a second graph structure containing feature vectors and risk-related attributes, as well as graph nodes and edge connections. The Graph Attention Neural Network (GAT) model can make full use of the correlation information between node features and edge connections, avoiding data loss.

[0052] Furthermore, the Graph Attention Neural Network (GAT) model does not require manual setting of node importance. It can learn the weights of the impact of different node combinations on risk through training, prioritize the core risk associations, reduce interference from irrelevant parameters, and effectively improve the accuracy of risk prediction.

[0053] Furthermore, the Graph Attention Neural Network (GAT) model possesses multi-dimensional feature output capabilities, enabling it to simultaneously uncover the correlation patterns among three types of risks: electric shock, overload, and short circuit, outputting the corresponding risk probabilities in a single step without requiring multiple modeling steps. Power transmission and transformation construction sites are subject to interference from factors such as rainy season humidity and strong winds and dust, resulting in noise in some monitoring data. The GAT model, through attention weight allocation, can mitigate the impact of noisy nodes, focusing on effective risk correlation features and ensuring stable prediction results even in complex environments. Therefore, the GAT model is adopted as the second power consumption risk prediction model in this application.

[0054] In some embodiments, please refer to Figure 4 The steps of inputting the second graph structure into the second electricity risk prediction model trained using the historical sample dataset, and outputting the corresponding electric shock risk probability, overload risk probability, and short circuit risk probability of the transmission and transformation line to be detected, include: Step S501: Input the second graph structure into the second electricity risk prediction model that has been trained to convergence using the historical sample dataset. Perform feature aggregation operation on the feature vectors and parameter attribute information of each graph node in the second graph structure. Dynamically assign differentiated attention weights to the neighboring nodes of each node by calculating the correlation between nodes, focusing on the node combination that is closely related to electric shock risk, overload risk, and short circuit risk. Step S502: Iteratively update the aggregated features through a multi-layer attention mechanism to gradually extract high-level feature representations that can reflect the core features of the three types of electricity consumption risks; Step S503: Input the high-level feature representation into the classifier in the second electricity risk prediction model, and perform probability mapping on the high-level feature representation through the softmax function to output the electric shock risk probability, overload risk probability and short circuit risk probability corresponding to the transmission and transformation line to be detected.

[0055] As can be seen from steps S501 to S503 above, by calculating the correlation degree of graph nodes and allocating differentiated attention weights, the system automatically focuses on node combinations that are strongly related to risk, such as "excessive residual current associated with high-altitude operations" and "overload characteristics associated with high temperature," thus avoiding the defect of traditional models that treat all nodes equally.

[0056] The multi-layer attention mechanism iteratively updates features, which can extract the core distinguishing features of three types of risks from complex graph structures (such as the "personnel contact + leakage" feature of electric shock risk, and the "insulation damage + high humidity" feature of short circuit risk). This can solve the problem of confusion between different risk features and provide highly recognizable feature support for subsequent probability output.

[0057] By outputting the risk probability in the [0,1] interval using the softmax function, parallel quantitative prediction of three types of risks is achieved. This also aligns with the "tiered early warning" logic of the intelligent distribution box (e.g., triggering an emergency alarm if the short-circuit risk probability is ≥90%). The probability results can be directly linked to preset thresholds to complete risk assessment, providing clear and operable decision-making basis for on-site management. The entire process fully utilizes the node features and edge association information of the second graph structure, inheriting the risk association results from the preceding graph construction stage and transforming structured data into intuitive risk probabilities through model computation.

[0058] For further embodiments, please refer to Figure 5 The steps of determining whether the transmission and transformation line under test has the risk of electric shock, overload, and short circuit based on the probability of electric shock risk, overload risk, and short circuit risk, respectively, include: Step S5001: Compare the electric shock risk probability, overload risk probability, and short circuit risk probability of the transmission and transformation line to be tested with the corresponding judgment thresholds respectively; Step S5002: If the probability of electric shock exceeds the electric shock risk determination threshold, the transmission and transformation line under test is determined to have an electric shock risk; if the probability of overload risk exceeds the overload risk determination threshold, the transmission and transformation line under test is determined to have an overload risk; if the probability of short circuit risk exceeds the short circuit risk determination threshold, the transmission and transformation line under test is determined to have a short circuit risk.

[0059] The electric shock risk assessment threshold can be arbitrarily selected between 85% and 90%, the overload risk assessment threshold can be arbitrarily selected between 80% and 85%, and the short circuit risk assessment threshold can be arbitrarily selected between 90% and 95%. Those skilled in the art can determine the electric shock risk assessment threshold, the overload risk assessment threshold, and the short circuit risk assessment threshold as needed according to the actual application scenario.

[0060] As can be seen from the above embodiments, compared with the prior art, the present application addresses the problem that traditional power supply protection at power transmission and transformation construction sites can only achieve passive protection against simple faults such as overcurrent and short circuits, which has problems such as delayed response, single monitoring dimensions, and high false trip rate. The present application has, but is not limited to, the following beneficial effects: Firstly, this application utilizes a long short-term memory neural network to mine the hidden correlation features of electrical operating parameters, construction condition parameters, and environmental parameters in the time dimension, completely breaking through the limitations of traditional devices that only focus on electrical parameters, and filling the monitoring gap under harsh environments and complex working conditions.

[0061] Secondly, traditional devices rely on passive triggering after a fault occurs. This application trains a long short-term memory neural network and a graph attention neural network using historical sample datasets to discover the evolution patterns of risks in advance and output the probabilities of electric shock, overload, and short circuit risks, thus realizing an active early warning mode of "risk prediction to early prevention".

[0062] Thirdly, by constructing a second graph structure, the scattered sample data is transformed into a structured risk association network. Then, through a graph attention neural network, attention weights are dynamically allocated to focus on high-risk node combinations, accurately distinguishing between real risks and parameter fluctuations, effectively reducing the false alarm rate in harsh environments, and balancing construction continuity with the reliability of safety protection.

[0063] Please see Figure 6A power supply risk prediction device for power transmission and transformation construction sites, provided to meet one of the purposes of this application, includes a dataset acquisition module 1100, a hidden feature extraction module 1200, a graph node construction module 1300, a graph structure construction module 1400, and a risk probability prediction module 1500. The dataset acquisition module 1100 is configured to acquire a historical sample dataset of the target transmission and transformation line. The historical sample dataset includes multiple historical sample data and their corresponding sample labels. The historical sample data includes time-series data constructed from electrical operating parameters, construction condition parameters, and environmental parameters corresponding to multiple consecutive time steps. The sample labels represent whether there is a risk of electric shock, overload, or short circuit. The hidden feature extraction module 1200 is configured to input multiple sample data to be detected from the sample dataset into a first power consumption risk prediction model trained using the historical sample dataset, to extract hidden correlation features between electrical operating parameters, construction condition parameters, and environmental parameters at each time step. The graph node construction module 1300 is configured to construct graph nodes of a first graph structure from each sample data to be detected in the sample dataset, wherein each parameter in the sample data to be detected serves as a graph node. The feature vectors of the graph nodes, and the hidden correlation features between electrical operating parameters, construction condition parameters, and environmental parameters at each time step in each sample data are used as the parameter attribute information of the graph nodes; the graph structure construction module 1400 is configured to determine whether the parameter attribute information of any two graph nodes in the first graph structure satisfies the preset risk association rule based on the parameter attribute information and the preset risk association rule. If it does, an edge connection is established between the two graph nodes to construct the second graph structure; the risk probability prediction module 1500 is configured to input the second graph structure into the second power consumption risk prediction model trained using the historical sample dataset, and output the corresponding electric shock risk probability, overload risk probability, and short circuit risk probability of the transmission and transformation line to be detected; based on the electric shock risk probability, overload risk probability, and short circuit risk probability, it is determined whether the transmission and transformation line to be detected has electric shock risk, overload risk, and short circuit risk, respectively.

[0064] Based on any embodiment of this application, please refer to Figure 7 Another embodiment of this application also provides an electronic device, which can be implemented by a computer device, such as... Figure 7The diagram shows the internal structure of a computer device. This computer device includes a processor, a computer-readable storage medium, a memory, and a network interface connected via a system bus. The computer-readable storage medium stores an operating system, a database, and computer-readable instructions. The database may store control information sequences. When the computer-readable instructions are executed by the processor, the processor can implement a method for predicting electricity consumption risks at power transmission and transformation construction sites. The processor of this computer device provides computing and control capabilities to support the operation of the entire computer device. The memory of this computer device may store computer-readable instructions. When these computer-readable instructions are executed by the processor, the processor can execute the method for predicting electricity consumption risks at power transmission and transformation construction sites as described in this application. The network interface of this computer device is used for communication with terminals. Those skilled in the art will understand that… Figure 7 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0065] In this embodiment, the processor is used to execute... Figure 6 The specific functions of each module are defined within the device, and the memory stores the program code and various data required to execute these modules. The network interface is used for data transmission between the user terminal and the server. In this embodiment, the memory stores the program code and data required to execute all modules in the power transmission and transformation construction site power risk prediction device of this application. The server can call the server's program code and data to execute the functions of all modules.

[0066] This application also provides a storage medium storing computer-readable instructions, which, when executed by one or more processors, cause the one or more processors to perform the steps of the power supply risk prediction method for power transmission and transformation construction sites as described in any embodiment of this application.

[0067] This application also provides a computer program product, including a computer program / instruction, which, when executed by one or more processors, implements the steps of the power supply risk prediction method for power transmission and transformation construction sites described in any embodiment of this application.

[0068] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. This computer program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the methods described above. The aforementioned storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0069] The above description is only a partial embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for predicting electricity consumption risks at power transmission and transformation construction sites, characterized in that, include: Obtain a historical sample dataset of the target transmission and transformation line. The historical sample dataset includes multiple historical sample data and their corresponding sample labels. The historical sample data includes time-series data constructed from electrical operating parameters, construction condition parameters and environmental parameters corresponding to multiple consecutive time steps. The sample labels represent whether there is a risk of electric shock, overload risk and short circuit risk. Multiple test sample data from the test sample dataset are input into the first electricity risk prediction model trained using the historical sample dataset to extract the hidden correlation features between electrical operating parameters, construction condition parameters, and environmental parameters at each time step. Each sample data in the dataset to be tested is constructed as a graph node of the first graph structure. The parameters in the sample data to be tested are used as the feature vectors of the graph node. The hidden correlation features between the electrical operation parameters, construction condition parameters and environmental parameters at each time step in each sample data are used as the parameter attribute information of the graph node. Based on the parameter attribute information and the preset risk association rule, determine whether the parameter attribute information of any two graph nodes in the first graph structure satisfies the preset risk association rule. If it does, establish an edge connection between the two graph nodes to construct the second graph structure. The second graph structure is input into the second electricity risk prediction model trained using the historical sample dataset, and the corresponding electric shock risk probability, overload risk probability, and short circuit risk probability of the transmission and transformation line to be detected are output. Based on the electric shock risk probability, overload risk probability, and short circuit risk probability, it is determined whether the transmission and transformation line to be detected has electric shock risk, overload risk, and short circuit risk, respectively.

2. The method for predicting electricity consumption risks at power transmission and transformation construction sites according to claim 1, characterized in that, The steps for training the primary electricity consumption risk prediction model include: The historical sample dataset is divided into a training set, a test set, and a validation set according to a preset ratio; Initialize the network parameters of the first electricity risk prediction model, setting the number of network layers to 3, the number of neurons in the hidden layer to 128, the dropout rate to 0.2, and the learning rate to 0.001; The multiple historical sample data are input into the initialized first electricity risk prediction model. The sample labels are used as supervision signals. The prediction error is calculated using the cross-entropy loss function. The model network parameters are adjusted by backpropagation through the Adam optimization algorithm. The model performance is evaluated using a validation set after a certain number of iterations. If the loss value on the validation set does not show a decreasing trend over multiple consecutive iterations, training is stopped. The test set is input into the first electricity risk prediction model after training, and the accuracy of the model in extracting hidden related features is calculated. If the accuracy is greater than or equal to a preset threshold, it is determined that the first electricity risk prediction model has been trained to a convergent state.

3. The method for predicting electricity consumption risks at power transmission and transformation construction sites according to claim 1, characterized in that, The step of determining whether the parameter attribute information of any two graph nodes in the first graph structure satisfies the preset risk association rule based on the parameter attribute information and the preset risk association rule, and establishing an edge connection between the two graph nodes to construct the second graph structure, includes: If the parameter attribute information of the first node includes a residual current greater than or equal to 30mA and a hidden associated feature of decreased insulation performance, and the parameter attribute information of the second node includes a worker being in a high-altitude working state and a hidden associated feature of increased worker contact risk, then the electric shock risk association rule is satisfied. If the parameter attribute information of the first node includes a loop current greater than or equal to 130% of the rated value and a hidden associated feature of continuous overload, and the parameter attribute information of the second node includes a line temperature greater than or equal to 60℃ and a hidden associated feature of poor line heat dissipation, then the overload risk association rule is satisfied. If the parameter attribute information of the node in the first figure includes a sudden change in short-circuit current and the hidden associated feature is line insulation damage, and the parameter attribute information of the node in the second figure includes line humidity greater than or equal to 85% and the hidden associated feature is enhanced environmental conductivity, then the short-circuit risk association rule is satisfied.

4. The method for predicting power consumption risks at power transmission and transformation construction sites according to claim 1, characterized in that, The steps of inputting the second graph structure into the second electricity risk prediction model trained using the historical sample dataset, and outputting the corresponding electric shock risk probability, overload risk probability, and short circuit risk probability of the transmission and transformation line to be detected, include: The second graph structure is input into the second electricity risk prediction model, which has been trained to convergence using the historical sample dataset. Feature aggregation is performed on the feature vectors and parameter attribute information of each graph node in the second graph structure. Differentiated attention weights are dynamically assigned to the neighboring nodes of each node by calculating the correlation between nodes, focusing on the combination of nodes closely related to electric shock risk, overload risk, and short circuit risk. By iteratively updating the aggregated features through a multi-layer attention mechanism, a high-level feature representation that can reflect the core characteristics of the three types of electricity consumption risks is gradually extracted. The high-level feature representation is input into the classifier in the second electricity risk prediction model. The high-level feature representation is then mapped using the softmax function to output the electric shock risk probability, overload risk probability, and short circuit risk probability of the transmission and transformation line to be detected.

5. The method for predicting electricity consumption risks at power transmission and transformation construction sites according to claim 1, characterized in that, The steps for determining whether the transmission and transformation line under test has the risk of electric shock, overload, and short circuit based on the probability of electric shock risk, overload risk, and short circuit risk, respectively, include: The probability of electric shock risk, the probability of overload risk, and the probability of short circuit risk of the transmission and transformation line to be tested are compared with the corresponding judgment thresholds respectively. If the probability of electric shock exceeds the electric shock risk determination threshold, the transmission and transformation line under test is determined to have an electric shock risk; if the probability of overload risk exceeds the overload risk determination threshold, the transmission and transformation line under test is determined to have an overload risk; if the probability of short circuit risk exceeds the short circuit risk determination threshold, the transmission and transformation line under test is determined to have a short circuit risk.

6. The method for predicting power consumption risks at power transmission and transformation construction sites according to any one of claims 1 to 5, characterized in that, The electrical operating parameters include incoming line voltage, loop voltage, equipment terminal voltage, incoming line current, loop current, equipment terminal current, residual current, short-circuit current, and grounding resistance; the construction condition parameters include high-altitude operation status and equipment start-up and shutdown status; the environmental parameters include line temperature, line humidity, and wind speed.

7. The method for predicting power consumption risks at power transmission and transformation construction sites according to any one of claims 1 to 5, characterized in that, The basic network architecture of the first electricity risk prediction model is a long short-term memory neural network; the basic network architecture of the second electricity risk prediction model is a graph attention neural network model.

8. A power supply risk prediction device for power transmission and transformation construction sites, characterized in that, include: The dataset acquisition module is configured to acquire the historical sample dataset of the target transmission and transformation line. The historical sample dataset includes multiple historical sample data and their corresponding sample labels. The historical sample data includes time-series data constructed from electrical operating parameters, construction condition parameters and environmental parameters corresponding to multiple consecutive time steps. The sample labels represent whether there is a risk of electric shock, overload risk and short circuit risk. The hidden feature extraction module is configured to input multiple test sample data from the test sample dataset into the first electricity risk prediction model trained using the historical sample dataset, so as to extract the hidden correlation features between electrical operating parameters, construction condition parameters, and environmental parameters at each time step; The graph node construction module is configured to construct graph nodes of the first graph structure from each sample data in the sample dataset to be detected. The parameters in the sample data to be detected are used as the feature vectors of the graph nodes, and the hidden correlation features between the electrical operation parameters, construction condition parameters and environmental parameters at each time step in each sample data are used as the parameter attribute information of the graph nodes. The graph structure construction module is configured to determine whether the parameter attribute information between any two graph nodes in the first graph structure satisfies the preset risk association rule based on the parameter attribute information and the preset risk association rule. If it does, an edge connection is established between the two graph nodes to construct the second graph structure. The risk probability prediction module is configured to input the second graph structure into the second power consumption risk prediction model trained using the historical sample dataset, and output the corresponding electric shock risk probability, overload risk probability, and short circuit risk probability of the transmission and transformation line to be detected; based on the electric shock risk probability, overload risk probability, and short circuit risk probability, it determines whether the transmission and transformation line to be detected has electric shock risk, overload risk, and short circuit risk, respectively.

9. An electronic device comprising a central processing unit and a memory, characterized in that, The central processing unit is used to invoke and run a computer program stored in the memory to perform the steps of the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores, in the form of computer-readable instructions, a computer program implemented according to any one of claims 1 to 7, which, when invoked by a computer, executes the steps included in the corresponding method.

Images