Power grid safety production testing methods, devices, electronic equipment, and storage media

Abstract

This invention discloses a method, apparatus, electronic device, and storage medium for power grid safety production detection. The method includes: acquiring safety production-related data of a target power grid, wherein the safety production-related data includes power grid operation characteristic data and power grid safety Q&A data of the target object; performing risk detection on the safety production-related data using a pre-established safety risk assessment model to obtain risk detection characteristic data; determining a target early warning scheme based on the risk detection characteristic data; predicting the target early warning scheme by executing a risk prediction model; and generating a target safety detection report related to the safety production of the target power grid based on the prediction results. This invention solves the problems of insufficient data integration, inaccurate risk identification, and imprecise countermeasures in related technologies, enabling comprehensive detection of the power grid safety production status, providing accurate early warnings, and significantly improving the efficiency and safety of power grid risk management.

Description

Technical Field

[0001] This invention relates to the field of power grid safety technology, and in particular to a method, apparatus, electronic device, and storage medium for power grid safety production testing. Background Technology

[0002] During the operation of the power grid, its safe and stable operation is of paramount importance.

[0003] However, due to the complexity and dynamic nature of power grid systems, safety risk assessment methods in related technologies often struggle to comprehensively and accurately identify and predict potential safety hazards. Furthermore, models used to monitor power grid safety primarily rely on power grid operational data, neglecting to assess the safety compliance of personnel involved in safety operations. This singular data source makes it difficult to comprehensively evaluate both power grid operational data and the safety awareness of personnel, resulting in inaccurate assessment results and an inability to provide sufficient and targeted solutions for power grid safety. Summary of the Invention

[0004] This invention provides a method, device, electronic equipment, and storage medium for power grid safety production detection, in order to solve the problems of insufficient data integration, inaccurate risk identification, and imprecise countermeasures in related technologies.

[0005] According to one aspect of the present invention, a method for power grid safety production detection is provided, comprising:

[0006] Acquire safety production-related data of the target power grid, wherein the safety production-related data includes power grid operation characteristic data and power grid safety Q&A data of the target object;

[0007] The safety production-related data are subjected to risk detection using a pre-established safety risk assessment model to obtain risk detection feature data.

[0008] A target early warning scheme is determined based on the aforementioned risk detection feature data;

[0009] The target early warning scheme is predicted by executing a risk prediction model, and a target safety inspection report related to the safe production of the target power grid is generated based on the prediction results.

[0010] According to another aspect of the present invention, a power grid safety production detection device is provided, comprising:

[0011] The safety production-related data acquisition module is used to acquire safety production-related data of the target power grid, wherein the safety production-related data includes power grid operation characteristic data and power grid safety Q&A data of the target object;

[0012] The risk detection module is used to perform risk detection on the safety production-related data using a pre-established safety risk assessment model, and obtain risk detection feature data.

[0013] The early warning scheme determination module is used to determine the target early warning scheme based on the risk detection feature data.

[0014] The safety inspection report generation module is used to predict the target early warning scheme by executing a risk prediction model, and generate a target safety inspection report related to the safe production of the target power grid based on the prediction results.

[0015] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0016] At least one processor; and

[0017] A memory communicatively connected to the at least one processor; wherein,

[0018] The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the power grid safety production detection method according to any embodiment of the present invention.

[0019] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the power grid safety production detection method according to any embodiment of the present invention.

[0020] The technical solution of this invention acquires safety production-related data of the target power grid. Since this data includes power grid operation characteristic data and power grid safety Q&A data of the target object, it can integrate related information in the power grid safety production process, improving the accuracy of detection. Next, a pre-established safety risk assessment model is used to perform risk detection on the safety production-related data to obtain risk detection characteristic data, which can accurately identify potential risks in power grid operation. Then, a target early warning scheme is determined based on the risk detection characteristic data, enabling early detection and rapid response to power grid safety hazards. Finally, a risk prediction model is executed to predict the target early warning scheme, and a target safety detection report related to the safety production of the target power grid is generated based on the prediction results. This improves the predictability and accuracy of power grid safety management, solves the problems of insufficient data integration, inaccurate risk identification, and imprecise response measures in related technologies, and can comprehensively detect the power grid safety production status, provide accurate early warnings, and significantly improve the efficiency and safety of power grid risk management.

[0021] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in 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 This is a flowchart of a power grid safety production testing method provided in Embodiment 1 of the present invention;

[0024] Figure 2 This is a flowchart of a power grid safety production testing method provided in Embodiment 2 of the present invention;

[0025] Figure 3 This is a schematic diagram of the structure of a power grid safety production detection device according to Embodiment 3 of the present invention;

[0026] Figure 4 This is a schematic diagram of the structure of an electronic device that implements the power grid safety production detection method according to an embodiment of the present invention. Detailed Implementation

[0027] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. 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 scope of protection of the present invention.

[0028] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0029] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0030] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0031] It is understood that before using the technical solutions disclosed in the various embodiments of this disclosure, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in this disclosure in an appropriate manner in accordance with relevant laws and regulations, and user authorization should be obtained.

[0032] For example, upon receiving a user's active request, a prompt message is sent to the user to explicitly inform them that the requested operation will require the acquisition and use of the user's personal information. This allows the user to independently choose whether to provide personal information to the software or hardware, such as the electronic device, application, server, or storage medium performing the operations of this disclosed technical solution, based on the prompt message.

[0033] As an optional but non-limiting implementation, in response to a user's active request, sending a prompt message to the user can be done via a pop-up window, where the prompt message can be presented in text format. Furthermore, the pop-up window can also include a selection control allowing the user to choose "agree" or "disagree" to provide personal information to the electronic device.

[0034] It is understood that the above notification and user authorization process are merely illustrative and do not constitute a limitation on the implementation of this disclosure. Other methods that comply with relevant laws and regulations may also be applied to the implementation of this disclosure.

[0035] It is understood that the data involved in this technical solution (including but not limited to the data itself, the acquisition or use of the data) shall comply with the requirements of relevant laws, regulations and related provisions.

[0036] Example 1

[0037] Figure 1 The flowchart of a power grid safety production detection method is provided in Embodiment 1 of the present invention. This embodiment is applicable to the comprehensive detection and evaluation of power grid safety production. The method can be executed by a power grid safety production detection device, which can be implemented in hardware and / or software. Optionally, it can be implemented through electronic devices, such as mobile terminals, PCs, or servers.

[0038] like Figure 1 As shown, the method may specifically include:

[0039] S110. Obtain safety production-related data of the target power grid, wherein the safety production-related data includes power grid operation characteristic data and power grid safety Q&A data of the target object.

[0040] The target power grid can be understood as a specific region or type of power network system, which is the object of safety assessment and testing. The safety production-related data can be understood as a collection of various information related to the safe operation of the power grid. This safety production-related data may include, but is not limited to, at least one of the following: power grid operation characteristic data and power grid safety Q&A data of the target object. The power grid operation characteristic data can be understood as various parameters and indicators related to the power grid's operating status, including but not limited to voltage, current, frequency, and power, and may also include at least one of the following: operating time of power equipment, number of failures, and maintenance records. This power grid operation characteristic data can be used to analyze the health status, efficiency, and stability of the power grid, and is an important basis for assessing the safety and reliability of the power grid. The power grid safety Q&A data can be understood as a collection of questions and answers related to power grid safety, used to understand the target object's understanding and implementation of power grid safety regulations, and to identify potential human-caused risk factors. This power grid safety Q&A data may include, but is not limited to, at least one of the following: employee training, safety inspections, and accident case analysis. The target object can be understood as personnel related to power grid safety production, including but not limited to dispatchers, maintenance engineers, and technical support teams.

[0041] Based on the above scheme, optionally, the power grid operation characteristic data of the target power grid can be obtained, including: collecting power grid operation characteristic data by acquiring electricity meters or power sensors installed on target power equipment in the target power grid; or, deploying network proxy services at target nodes in the power equipment and collecting power grid operation characteristic data at preset time intervals through the network proxy services; or, pulling power grid operation characteristic data from network equipment and servers through network management protocols; or, obtaining power grid operation characteristic data by calling preset data acquisition interfaces, etc.

[0042] Based on the above scheme, optionally, the power grid safety Q&A data of the target object can be obtained, including: collecting the power grid safety Q&A data of the target object through an online questionnaire system or intelligent dialogue system; or, interacting with the target object and determining the power grid safety Q&A data based on the interaction content, etc.

[0043] S120. Use a pre-established safety risk assessment model to perform risk detection on the safety production related data to obtain risk detection feature data.

[0044] The safety risk assessment model can be understood as an assessment model used to quantify the degree of uncertainty and potential hazards in the safe production process of the power grid. Specifically, through comprehensive analysis of power grid operation characteristic data and safety Q&A data, potential risk points that may lead to safety accidents can be detected and quantitatively assessed. The risk detection characteristic data can be understood as data output by the safety risk assessment model, reflecting the main risks and characteristics currently faced by the target power grid.

[0045] Based on the above scheme, optionally, the step of using a pre-established safety risk assessment model to perform risk detection on the safety production-related data to obtain risk detection feature data includes: acquiring the operational feature data and acquiring the target semantic features in the power grid safety Q&A data, mapping the operational feature data to a state feature space to determine a preliminary risk point set, and using a power grid safety knowledge graph to perform contextual understanding on the target semantic features to determine the safety score of the target semantic features; using the safety risk assessment model to perform risk detection on each risk point and the safety score in the preliminary risk point set to obtain a preliminary risk detection result, and determining the risk clustering area and high-risk period of the target power grid based on the preliminary risk detection result and the time and regional data of the corresponding risk points in the preliminary risk point set to obtain a risk distribution map; using the power grid safety knowledge graph to perform risk cause detection on the risk clustering area in the high-risk period of the risk distribution map to obtain a preliminary risk explanation list, and based on the corresponding associated risk data and historical risk cases in the power grid safety knowledge graph, performing confidence assessment and structured classification on each explanation in the preliminary risk explanation list to generate risk detection feature data.

[0046] The target semantic features can be understood as information extracted from power grid safety Q&A data, reflecting the target object's understanding and implementation of power grid safety regulations. The state feature space can be understood as an abstract space where each dimension represents a feature or attribute of power grid operation. The state feature space includes at least one of the following: information on the change process of power grid state patterns and annotation information of the power grid state patterns. The state feature space is used to map operational feature data to the state feature space, visually representing the evolution of the power grid state and identifying potential risk points. The information on the change process of power grid state patterns can be understood as the trend and pattern of power grid state changes over time, helping to determine the dynamic behavior of the power grid and predict possible future risk trends. For example, the change process information of the power grid state patterns includes, but is not limited to, at least two processes such as normal operation, mild overload, severe overload, or voltage collapse. The annotation information can be understood as marking the power grid state at a specific point in time, time period, or for power equipment. For example, the annotation information can be a semantic annotation of "increased risk of voltage instability." The preliminary point set can be understood as a set of potentially risky locations or conditions derived from the state feature space analysis. The power grid safety knowledge graph can be understood as a structured knowledge representation that uses nodes (entities) and edges (relationships) to indicate the connections between different power devices in the power grid and how these devices work together to ensure the stability and security of the power grid. Contextual understanding can be understood as accurately capturing the true meaning of words, phrases, or sentences by analyzing the context of target semantic features when processing them. In this embodiment, contextual understanding can help to more accurately parse the target object's answers and questions, thereby better assessing the target object's understanding and implementation of safe production. The safety score can be understood as scoring the target semantic features extracted from the question-and-answer data based on the power grid safety knowledge graph, reflecting the risk level associated with the target semantic features, used to quantify the impact of human factors on power grid safety production, and providing a reference for comprehensive risk assessment. The preliminary risk detection result can be understood as the result obtained by analyzing each risk point and safety score in the preliminary risk point set, indicating the current risk status of power grid safety production and laying the foundation for further analysis. The risk clustering area and high-risk period can be understood as the area and time period with a high probability of risk events occurring in the power grid within a specific time and location. The risk distribution map can be understood as a graphical representation of the risk distribution at different locations and time periods within the power grid. Risk cause detection can be understood as identifying the specific causes of risks by analyzing high-risk periods and clustered areas, providing a basis for developing effective solutions and ensuring that problems are fundamentally resolved. The preliminary risk explanation list can be understood as listing various factors that may lead to risks and their corresponding explanations.The associated risk data can be understood as data related to power grid safety production during the power grid safety production assessment process. This associated risk data can be used to connect data from different sources (such as operational characteristic data or Q&A data) with known risk factors and historical risk cases to identify potential risk points and assess their impact. Historical risk cases can be understood as records of past risk events related to power grid safety or other related fields. These historical risk cases may include, but are not limited to, at least one of the event's background, development process, final result, and post-event assessment. The confidence assessment can be understood as evaluating the credibility of each explanation in the preliminary risk explanation list, judging its accuracy and reliability. The structured classification can be understood as organizing and classifying risk explanations according to certain rules and standards, thereby making the safety production associated data more organized, easier to view, and easier to apply.

[0047] An alternative implementation involves extracting target semantic features from operational feature data using a sliding window technique and from power grid safety question-and-answer data processed using word embedding techniques such as Word2Vec. The extracted operational feature data is then mapped to a constructed state feature space using a nearest neighbor algorithm to identify a preliminary set of risk points. A power grid safety knowledge graph is then used, employing graph embedding techniques such as TransE to perform contextual understanding on the extracted target semantic features, thereby assessing the safety score of the target object. A safety risk assessment model is then used to perform risk detection (safety risk assessment and production anomaly detection) on each risk point and safety score in the preliminary risk point set. Specifically, the safety risk assessment model uses the Analytic Hierarchy Process (AHP) and fuzzy comprehensive evaluation methods, comprehensively considering equipment status, system stability, and personnel factors, to obtain preliminary risk assessment results. Finally, based on the spatiotemporal data of the risk points corresponding to the preliminary risk point set, spatial statistical methods such as kernel density estimation (KDE) are used to identify risk clustering areas. The formula for the two-dimensional KDE is:

[0048]

[0049] Where K is the kernel function, h is the bandwidth parameter, (x i y iHere, ) represents the sample points, and n represents the number of samples. KDE (Knowledge-Defined Analysis) is used to identify power grid risk clusters and high-risk periods in the preliminary risk assessment results, generating a risk distribution map. Then, a Bayesian network is constructed on the power grid safety knowledge graph to detect the causes of risks in high-risk areas, generating a preliminary list of anomaly explanations. Based on the corresponding associated risk data and historical risk cases in the power grid safety knowledge graph, a random forest algorithm is used to assess the confidence level of each explanation in the preliminary anomaly explanation list. By comprehensively considering the similarity of historical cases and the specificity of the current situation, hierarchical clustering or topic models such as LDA are used to perform structured classification of the assessed explanations, ultimately generating the anomaly assessment results. For example, assuming there is real-time voltage data from a substation, wavelet transform is used to extract features such as voltage trends, periodicity, and sudden changes. For target safety question-and-answer data, such as "precautions for construction near high-voltage lines," target semantic features such as "high voltage," "construction," and "safety distance" are identified. Then, the extracted voltage feature vector is mapped to the state feature space to find that it falls near an area representing "voltage instability," thus identifying a potential risk point. Meanwhile, by analyzing the semantic features of the question-and-answer data, the employee's safety awareness score was assessed at 85 points (out of 100). Based on historical data, it was predicted that if the voltage continued to fluctuate, equipment overload might occur within 15 minutes. KDE analysis revealed that the substation area was a high-risk zone, especially during peak electricity consumption periods. Bayesian networks were then used to infer that the possible causes of voltage instability were: a sudden increase in load (confidence 0.7), line fault (confidence 0.5), and equipment aging (confidence 0.3). A random forest model was then used to evaluate these explanations. Considering historical cases and relevant risk data, the final conclusion was that a sudden increase in load was the most likely cause (confidence 0.8), and close monitoring of load changes and preparation for activating backup power were recommended. Through this series of complex analysis processes, real-time data and safety Q&A data are transformed into specific risk assessment results and anomaly explanations. This method not only considers the real-time status of equipment but also combines the level of safety awareness of personnel. At the same time, it utilizes the rich semantic information in the power grid safety knowledge graph. This provides comprehensive and accurate decision support for power grid safety production supervision, helps to discover and prevent potential safety hazards in a timely manner, and thus improves the overall safety and reliability of the power grid.Furthermore, when extracting operational characteristic data, it is necessary to consider the data sampling frequency and potential delays. When assessing personnel's safety awareness, the timeliness of question-and-answer data and employees' work experience need to be considered. When conducting risk detection, it is necessary to balance the weights of risk factors at different levels to ensure that the assessment results reflect both local risks and the safety status at the system level. Further, it is necessary to integrate safety standard learning into the work scenarios of safety production personnel by adding more nodes and relationships related to specific work scenarios to the power grid safety knowledge graph. Special attention should also be paid to medium- and high-risk operations, such as giving these operations higher weights in the risk assessment model.

[0050] This technical solution, by leveraging the target semantic features in power grid operation characteristic data and power grid safety Q&A data, and employing a pre-established safety risk assessment model and power grid safety knowledge graph, achieves in-depth mining and precise positioning of potential risks in the power system. It can map operational characteristics to a state feature space to determine the initial set of risk points and quantify the safety scores of power grid safety Q&A data through contextual understanding. Furthermore, by analyzing the temporal and regional data of the preliminary risk detection results, the solution successfully generates a risk distribution map, visually displaying risk clustering areas and high-incidence periods. Finally, by combining historical risk cases and related risk data, it detects and explains the causes of risks and generates detailed risk detection characteristic data through confidence assessment and structured classification. This significantly improves the predictability, accuracy, and response speed of power grid safety production management, facilitating timely preventative measures and ensuring the stable and safe operation of the power system.

[0051] S130. Determine the target early warning scheme based on the risk detection feature data.

[0052] The target early warning scheme can be understood as a preventive measure and emergency response plan designed for the security risks of the target power grid, which aims to take action in advance to avoid or mitigate the impact of possible security incidents.

[0053] Optionally, based on the above scheme, determining the target early warning scheme according to the risk detection feature data includes: determining a preliminary risk score for a power grid safety production risk event based on the risk detection feature data; determining the risk influencing factors of the target power grid based on the preliminary risk score and a power grid safety knowledge graph; determining the risk influencing path based on the risk influencing factors; determining the risk influencing objects in the risk influencing path; determining the influencing hierarchy structure based on the risk influencing objects and the risk influencing factors; matching the detection and early warning response library corresponding to the influencing hierarchy structure based on a preset multi-level early warning threshold system; and determining the target early warning scheme corresponding to the influencing hierarchy structure based on the detection and early warning response library.

[0054] The preliminary risk score can be understood as the result of a quantitative assessment of power grid safety production risk events based on risk detection feature data. This preliminary risk score can be used to characterize the severity and urgency of the risk. The risk influencing factors can be understood as factors that may lead to risks in power grid operation, including but not limited to at least one of equipment failure, operational errors, and environmental changes. The risk impact path can be understood as a process chain from risk occurrence to the spread of risk impact, used to determine how the risk propagates within the power grid. The risk impact object can be understood as a specific component or area affected by the risk, including but not limited to at least one of specific equipment, substations, or transmission lines. The impact hierarchy structure can be understood as a hierarchical structure constructed based on risk influencing factors and risk impact objects, reflecting the risk impact relationships at different levels, and can be used to achieve targeted hierarchical management of risk impact objects. The multi-level early warning threshold system can be understood as a preset risk level classification standard, with each level corresponding to different response requirements, used to ensure that appropriate levels of early warning and response measures are taken for risks of different degrees, avoiding overreaction or underreaction. The detection, early warning, and response library can be understood as a database containing predefined early warning and response schemes, suitable for determining different types of risk handling schemes based on the impact hierarchy structure. The target early warning scheme can be understood as the preventive measures and emergency response plan determined for power grid safety production risk events of a specific power grid. It can be used to take action in advance to avoid or mitigate the impact of possible power grid safety production risk events and ensure the safe and stable operation of the power grid.

[0055] An optional implementation method employs fuzzy comprehensive evaluation. This involves establishing an evaluation index system, including the scope of risk impact, duration, and potential losses. Then, fuzzy membership functions are used to transform these indicators into fuzzy sets. The fuzzy synthesis operation of the evaluation matrix R and the weight vector A can be expressed as B = A·R, where "·" represents the fuzzy synthesis operator. This allows for the quantitative analysis of the severity and urgency of risk detection feature data, yielding a preliminary risk score for power grid safety production risk events. Based on the preliminary risk score and the power grid safety knowledge graph, a graph neural network (GNN) is used to determine the risk influencing factors of the target power grid and target objects. This captures the impact of power grid safety production risk events on different components and personnel of the power grid, thus obtaining the hierarchical structure of the risk influencing factors. Finally, based on a preset multi-level early warning threshold system, a decision tree algorithm, such as the C4.5 algorithm, is used to select the optimal partitioning feature using the information gain ratio to match the corresponding detection and early warning response library. The calculation formula for the C4.5 algorithm is as follows:

[0056] Gain_ratio(A)=Gain(A) / SplitInfo(A);

[0057] Where Gain(A) is the information gain of feature A, and SplitInfo(A) is the splitting information of feature A. By constructing a decision tree, appropriate early warning response strategies (i.e., detection of the early warning response library) are quickly matched according to different features affecting the hierarchical structure. Then, a collaborative filtering algorithm is used to perform early warning role matching analysis on the detection of the early warning response library. Assuming R is the early warning role-earning strategy scoring matrix, the goal is to decompose it into the product of two matrices U and V: R≈UV^T. The corresponding optimization objective function is expressed as:

[0058]

[0059] Where, r ij It is a known score, u i and v j These are the potential factor vectors for early warning roles and early warning strategies, respectively. λ is the regularization parameter, m is the total number of early warning roles, n is the total number of early warning strategies, i is the early warning role with a value ranging from 1 to m, and j is the early warning strategy with a value ranging from 1 to n. In this way, the most suitable early warning strategy is recommended for each early warning role, generating the final safety production detection early warning strategy. For example, suppose an abnormal temperature rise is detected in a transformer at a substation. Fuzzy comprehensive evaluation might classify this as a moderately severe (0.6) and highly urgent (0.8) risk event, with an initial risk score of 0.7. Using GAT analysis of the power grid knowledge graph, it might be discovered that this anomaly affects not only the substation but also downstream distribution networks and electricity users, forming a three-tiered impact hierarchy. Based on this hierarchy, a decision tree algorithm could be used to match a warning response library containing strategies such as "starting the backup transformer," "limiting non-critical loads," and "notifying the maintenance team." Finally, a collaborative filtering algorithm can assign different tasks to different roles: dispatchers are responsible for starting the backup transformer, maintenance personnel for on-site inspections and maintenance preparation, and customer service personnel for notifying potentially affected key customers. Role matching ensures that each relevant person receives appropriate instructions, thereby improving the efficiency and accuracy of power grid safety production supervision.

[0060] This technical solution analyzes risk detection feature data to determine preliminary risk scores for power grid safety production risk events. It then utilizes a power grid safety knowledge graph to identify key risk influencing factors and their pathways, clarifying the risk impact objects and hierarchical structure. Based on this, a pre-set multi-level early warning threshold system is matched, and the most suitable target early warning scheme is selected from the detection and early warning response library. This not only improves the accuracy and timeliness of early warnings but also enhances the pertinence and effectiveness of response measures, ensuring that appropriate preventative and control measures can be taken before risks occur, effectively guaranteeing the safe and stable operation of the power grid and reducing potential losses.

[0061] S140. The target early warning scheme is predicted by executing the risk prediction model, and a target safety inspection report related to the safe production of the target power grid is generated based on the prediction results.

[0062] The risk prediction model can be understood as a model used to predict the likelihood and impact of risks occurring after the implementation of the target early warning scheme. The target safety inspection report can be understood as the result of all safety inspection activities conducted on the target power grid, including but not limited to at least one of the problems discovered, measures taken, and improvement suggestions. It can be provided to management and relevant departments for reference in order to continuously optimize the safety management and operation of the target power grid.

[0063] The technical solution of this invention acquires safety production-related data of the target power grid. Since this data includes power grid operation characteristic data and power grid safety Q&A data of the target object, it can integrate related information in the power grid safety production process, improving the accuracy of detection. Next, a pre-established safety risk assessment model is used to perform risk detection on the safety production-related data to obtain risk detection characteristic data, which can accurately identify potential risks in power grid operation. Then, a target early warning scheme is determined based on the risk detection characteristic data, enabling early detection and rapid response to power grid safety hazards. Finally, a risk prediction model is executed to predict the target early warning scheme, and a target safety detection report related to the safety production of the target power grid is generated based on the prediction results. This improves the predictability and accuracy of power grid safety management, solves the problems of insufficient data integration, inaccurate risk identification, and imprecise response measures in related technologies, and can comprehensively detect the power grid safety production status, provide accurate early warnings, and significantly improve the efficiency and safety of power grid risk management.

[0064] Example 2

[0065] Figure 2 This is a flowchart of a power grid safety production detection method provided in Embodiment 2 of the present invention. Based on the embodiments described above, this embodiment further refines how to determine the safety risk assessment model. Optionally, determining the safety risk assessment model includes: acquiring multiple safety production reference data of the target power grid; determining the state feature space of the target power grid based on the safety production reference data; wherein the state feature space includes information on the change process of the power grid state mode and the annotation information of the power grid state mode; constructing a knowledge graph based on the state feature space and the safety production reference data to obtain a power grid safety knowledge graph; and using the power grid safety knowledge graph to perform a hierarchical risk assessment on a preset risk assessment model to obtain a safety risk assessment model. For detailed implementation, please refer to the description of this embodiment. Technical features that are the same as or similar to those in the foregoing embodiments will not be repeated here.

[0066] like Figure 2 As shown, the method may specifically include:

[0067] S210. Obtain multiple safety production reference data of the target power grid, and determine the state feature space of the target power grid based on the safety production reference data; wherein, the state feature space includes information on the change process of the power grid state mode and the annotation information of the power grid state mode;

[0068] The safety production reference data can be understood as various historical data related to the safe operation of the power grid, including but not limited to at least one of the following: historical operating parameters, historical maintenance records, historical fault reports, and historical power grid safety Q&A data.

[0069] Based on the above scheme, after obtaining multiple safety production reference data of the target power grid, the method further includes: preprocessing the safety production related data, the preprocessing including at least format unification processing, timestamp alignment and synchronization, missing value detection and imputation, outlier detection and cleaning, data noise reduction and smoothing, and data standardization and normalization processing.

[0070] Based on the above scheme, optionally, determining the state feature space of the target power grid according to the safety production reference data includes: performing time-series decomposition on the safety production reference data to obtain time-domain power grid features, and performing time-frequency transformation on the time-domain power grid features using short-time Fourier transform to obtain frequency-domain power grid features; performing nonlinear transformation on the time-domain power grid features and the frequency-domain power grid features to obtain data distribution features; performing cross-domain feature correlation analysis on the safety production reference data based on the data distribution features to obtain interactive power grid features; performing feature fusion on the time-domain power grid features, the frequency-domain power grid features, the data distribution features, and the interactive power grid features, performing feature dimensionality reduction on the fused features to obtain low-dimensional power grid features; clustering the low-dimensional power grid features to obtain multiple clusters, determining the power grid state pattern according to the cluster to which the low-dimensional power grid features belong, performing semantic annotation on the power grid state pattern, and determining the state feature space of the target power grid based on the semantic annotation results.

[0071] The time-series decomposition can be understood as the process of decomposing time-series data in safety production reference data into multiple components, such as trends, seasonal components, and noise. The time-domain power grid characteristics can be understood as characteristic parameters of the power grid in the time dimension obtained through time-series decomposition, which can be used to capture the behavioral patterns of the target power grid over time, such as voltage fluctuations or current changes. The short-time Fourier transform can be understood as converting time-domain signals into frequency-domain signals while retaining local time information. The frequency-domain power grid characteristics can be understood as characteristic parameters of the target power grid in the frequency dimension obtained through short-time Fourier transform, used to capture information about the frequency components of the target power grid, helping to identify anomalies or patterns at specific frequencies. The nonlinear transformation can be understood as performing nonlinear mapping on data to reveal hidden data structures or features. For example, the nonlinear transformation can utilize methods such as kernel principal component analysis to capture complex nonlinear relationships in the time-domain and frequency-domain power grid characteristics. The data distribution characteristics can be understood as data distribution attributes obtained after nonlinear transformation, used to reflect information such as skewness and / or kurtosis of the data distribution, helping to determine the operating status of the target power grid. The cross-domain feature association analysis can be understood as a method for determining the relationship between different types of safety production reference data, discovering the interactions and dependencies between different types of safety production reference data. The interactive power grid features can be understood as features derived from the cross-domain feature association analysis, used to determine the interactive influence between different data types. The feature fusion can be understood as integrating multiple types of features (such as time-domain power grid features, frequency-domain power grid features, operating state features, and interactive power grid features) into a unified feature representation, ensuring that all relevant features can be utilized. The feature dimensionality reduction can be understood as a method for reducing the number of features while retaining as much important information as possible, simplifying the data structure and reducing computational complexity. The low-dimensional power grid features can be understood as simplified feature representations after feature dimensionality reduction, still retaining the key information of the safety production reference data. The clustering can be understood as grouping low-dimensional power grid features according to their similarity, with each group called a cluster, used to classify low-dimensional power grid features and help identify different power grid state patterns. The cluster can be understood as each group formed after clustering low-dimensional power grid features represents a group of power grid states with similar characteristics, facilitating classification management and targeted processing. The power grid state patterns can be understood as different categories of power grid operating states defined by clustering results, reflecting the power grid's behavior at different times. The semantic annotations can be understood as assigning readable labels to the power grid state patterns, describing their specific meanings, and enhancing the interpretability of the power grid state patterns.

[0072] One optional implementation involves first using empirical mode decomposition (EMD) to perform time-series decomposition on the preprocessed safety production-related data. This decomposes the complex time series into a series of intrinsic mode functions (IMFs), revealing the inherent time-scale characteristics of the preprocessed safety production-related data. Simultaneously, short-time Fourier transform (SFT) is used to perform time-frequency transformation on the time-series decomposition results, acquiring frequency characteristics while preserving time information. This helps identify periodic patterns and transient events in the power grid, ultimately yielding time-domain and frequency-domain power grid characteristics. Then, kernel principal component analysis (KPC) is used to perform nonlinear transformation on the time-domain and frequency-domain power grid characteristics to capture complex nonlinear relationships in the data and extract more representative features, resulting in... Data distribution characteristics (which reflect information such as skewness and kurtosis of the data distribution, helping to more comprehensively describe the operating status of the power grid); then, based on the data distribution characteristics, Tucker decomposition is used to perform cross-domain feature correlation analysis on the preprocessed safety production related data to reveal potential correlations between different types of data, such as the relationship between voltage fluctuations and load changes, or the relationship between weather conditions and equipment failure rates, to obtain interactive power grid characteristics; then, feature cascading or feature fusion methods are used to combine time-domain power grid characteristics, frequency-domain power grid characteristics, data distribution characteristics, and interactive power grid characteristics, while the t-SNE algorithm is used for feature dimensionality reduction, where the objective function of the t-SNE algorithm is:

[0073]

[0074] p ij and q ij The similarity between points i and j in the high-dimensional and low-dimensional spaces is respectively used to preserve the local structure of the data while reducing dimensionality, discovering potential patterns in the data, and obtaining low-dimensional power grid features. Then, a Gaussian Mixture Model (GMM) is used to perform cluster analysis on the low-dimensional power grid features, classifying similar power grid states and identifying typical power grid state patterns, such as normal operation, high load, or fault states. Finally, a Dynamic Bayesian Network (DBN) is used to capture the evolution of power grid states over time (i.e., time-series relationships). The identified state patterns are combined with professional knowledge to give them clear semantic interpretations. The joint probability distribution of the DBN is expressed as:

[0075]

[0076] X t Let P(X1:T) represent the set of state variables at time t, and let P(X1:T) represent the state from X1 to X2. T The joint probability, P(X1) represents the probability of the first variable X1, P(X t |X t-1 ) indicates that given a previous variable Xt-1 Under the condition that variable X t Conditional probabilities. For example, suppose there is a set of voltage data from substations. First, EMD is used to decompose the voltage signal into multiple IMFs, which may correspond to changes at different scales such as daily load variations and periodic fluctuations. Then, WPT is applied to analyze the characteristics of each IMF in the time-frequency domain to discover that certain specific frequency components are related to equipment faults. Next, bispectral analysis may reveal nonlinear couplings between certain frequency components, which may indicate certain abnormal patterns in the power grid. Furthermore, when using GCN for cross-domain feature correlation analysis, substations are treated as nodes in a graph, and power lines as edges, thereby analyzing the mutual influence between different substations. Dimensionality reduction using t-SNE may reveal that the data forms several distinct clusters in the low-dimensional space, which correspond to different power grid operating states. GMM is used to model these clusters to obtain typical power grid state patterns, such as normal operation, mild overload, or severe overload. Finally, DBN is used to model the temporal evolution of these state patterns, which may reveal certain state transition sequences that predict potential fault risks. For example, a state sequence of "normal operation → mild overload → severe overload → voltage collapse" might be discovered, and a semantic annotation of "increased risk of voltage instability" could be added to it. Through this series of complex analysis processes, the original multi-source heterogeneous data is transformed into a state feature space rich in semantic information. This state feature space not only contains the static features of the power grid state, but also captures the dynamic evolutionary relationship between states.

[0077] This technical solution extracts and analyzes multi-dimensional features from safety production reference data, including time-series decomposition, time-frequency transformation, nonlinear transformation, and cross-domain feature correlation. It integrates time-domain, frequency-domain, data distribution, and interaction features, and determines the power grid state pattern through feature dimensionality reduction and clustering analysis. Combined with semantic annotation, it defines the state feature space of the target power grid, which can comprehensively capture the dynamic characteristics and complex patterns of power grid operation, provide more accurate state assessment, provide strong data support and technical guarantee for the safe and stable operation of the power grid, and improve the ability to predict and prevent faults.

[0078] S220. Based on the state feature space, a knowledge graph is constructed from the safety production reference data to obtain a power grid safety knowledge graph;

[0079] Based on the above scheme, optionally, the step of constructing a knowledge graph of the safety production reference data based on the state feature space to obtain a power grid safety knowledge graph includes: determining the classification information of the state feature space; constructing an initial graph body model based on the classification information; and performing entity recognition and relation extraction on the safety production reference data based on the initial graph body model to construct an initial power grid knowledge graph; aligning the initial graph body model and the initial power grid knowledge graph to obtain power grid safety knowledge; adjusting the power grid safety knowledge to obtain an initial dynamic knowledge graph; and constructing a power grid safety knowledge graph based on the entities and relations in the initial dynamic knowledge graph.

[0080] The classification information can be understood as the result of classifying data in the state feature space. The initial graph body model can be understood as the basic framework for constructing a power grid safety knowledge graph, defining the basic relationships between data such as power equipment and power grid state patterns in the graph. Entity recognition can be understood as identifying entities related to the power grid from safety production reference data, including but not limited to at least one of power grid equipment, operators, and safety events, used to convert unstructured safety production reference data into structured information. Relationship extraction can be understood as extracting relationships between entities after entity recognition, including but not limited to at least one of connection, responsibility, and cause. The initial power grid knowledge graph can be understood as a power grid safety knowledge graph built on the initial graph body model and the identified entities and relationships. The initial power grid knowledge graph initially demonstrates the knowledge structure in the field of power grid safety, providing a foundation for further improvement and application. The alignment process can be understood as the process of matching and integrating the initial graph body model from different sources or versions with the initial power grid knowledge graph, ensuring the semantic consistency of concepts and relationships in the graph, improving the completeness and accuracy of the knowledge graph, and avoiding duplicate or contradictory information. The power grid security knowledge can be understood as a set of specific knowledge about power grid security obtained after alignment processing, including but not limited to at least one of entities, relationships, and their attributes. The initial dynamic knowledge graph can be understood as an adjusted power grid security knowledge graph with dynamic update capabilities, which can respond promptly to new data and changes in circumstances, and provide more accurate decision support.

[0081] An alternative implementation, based on a state feature space, involves a systematic review of knowledge in the power grid security domain, including but not limited to classification information across multiple dimensions such as power equipment, operating status, safety regulations, and accident types. For example, a formal ontology language, OWL (Web Ontology Language), can be used to define the concept hierarchy, with a key construct represented as: Class(A partial C1...Cn), where A represents a new class (concept), and C1 to Cn represent the parent classes of A. In this way, a hierarchical structure of concepts such as power grid equipment, safety status, and operating procedures can be defined, thus constructing an initial knowledge graph model. Based on this initial knowledge graph model, for unstructured text data, deep learning models such as BERT are used for entity recognition; for structured data, entities are extracted through predefined rules and mapping relationships. Simultaneously, methods such as remote supervised learning or graph neural networks are used to identify entities such as power grid equipment, operators, and safety events from the raw data, as well as various relationships between them, such as "connection," "responsibility," or "cause," thus constructing an initial power grid knowledge graph. Then, ontology matching techniques, such as semantic similarity calculation based on word vectors or graph structure matching algorithms, are used to identify and merge semantically similar concepts and relationships, thereby improving the initial knowledge graph. The main model and the initial power grid knowledge graph are integrated and aligned with the power grid knowledge and safety ontology to ensure semantic consistency of concepts and relationships in the initial power grid knowledge graph, resulting in semantically unified power grid safety knowledge. Furthermore, the power grid safety knowledge is optimized through graph structure optimization (by adjusting and improving nodes (entities) and edges (relationships) in the power grid safety knowledge graph to improve its quality, efficiency, and expressiveness) and temporal expansion of the graph structure. Graph structure optimization includes, but is not limited to, adding new entities, deleting redundant or irrelevant entities, merging similar entities, or making implicit relationships explicit, to improve the graph's conciseness and expressiveness. Simultaneously, to capture the temporal evolution characteristics of power grid states and events, timestamp attributes are introduced, or temporal relation edges are constructed, and a temporal graph neural network (TGNN) is used to capture dynamic changes to represent changes in entity states and the order of events. The update rule of the TGNN is expressed as follows: Let AGG be the hidden state of node v at time t, AGG be the aggregation function, and N(v) be the set of v's neighbors. In this way, an initial dynamic knowledge graph of the power grid's temporal relationships is obtained (which not only contains static knowledge structures but also reflects the dynamic characteristics of the power grid system). Then, by pre-defining a series of safety-related attributes and relationships, such as the safe operating parameter range of equipment and the safe preconditions for operation, and by constructing safety logic rules—rules that express complex safety constraints and reasoning logic based on entities and relationships in the graph—it is possible to map power grid safety domain knowledge and construct and update safety logic rules for entities and relationships in the dynamic knowledge graph. For example, a rule such as "If the load of transformer A exceeds threshold X and the duration exceeds Y, then an overload warning is triggered" can be defined. These rules are automatically learned from historical data through rule mining algorithms or manually defined by domain experts, and should be able to be dynamically updated based on new data and knowledge, ultimately generating a power grid safety knowledge graph with safety reasoning capabilities. The power grid safety knowledge graph not only contains rich power grid domain knowledge but also embeds safety-related logical rules and reasoning capabilities. It can support complex knowledge queries, correlation analysis, and security reasoning, providing strong knowledge support for power grid safety production supervision. For example, when the system detects abnormal data from a substation, it can quickly locate relevant equipment through graphs, search for similar historical cases, infer possible causes and the scope of impact, and provide corresponding handling suggestions. It can also generate test questions and other content for safety regulation assessments of production personnel.

[0082] This technical solution constructs a power grid security knowledge graph based on a state feature space, determines classification information, builds an initial graph body model, and performs entity recognition and relationship extraction to form an initial power grid knowledge graph. Then, through alignment processing and optimization, an initial dynamic knowledge graph with dynamic update capabilities is generated. Finally, based on the entities and relationships within it, a comprehensive and accurate power grid security knowledge graph is constructed. This allows for the systematic integration of power grid operation data, improves data utilization efficiency, and enhances the understanding and prediction capabilities of the power grid security situation, thereby effectively ensuring the safe and stable operation of the power grid.

[0083] S230. Using the power grid security knowledge graph, a hierarchical risk assessment is performed on the preset risk assessment model to obtain a security risk assessment model.

[0084] Based on the above scheme, optionally, the step of using the power grid security knowledge graph to perform hierarchical risk assessment on the preset risk assessment model to obtain a security risk assessment model includes: determining a set of potential risk factors based on potential risk factors in the power grid security knowledge graph; hierarchically classifying the set of potential risk factors to obtain a risk indicator system; and determining a security risk propagation network corresponding to the set of potential risk factors based on the power grid security knowledge graph; hierarchically decomposing the security risk propagation network based on the risk indicator system to obtain a hierarchical risk assessment scheme; and extracting security rules from the security semantic information in the power grid security knowledge graph based on the hierarchical risk assessment scheme to obtain risk assessment rules; fuzzifying the risk assessment rules to obtain a risk reasoning mechanism; and performing multi-dimensional risk assessment on the set of potential risk factors based on the risk reasoning mechanism and the analytic hierarchy process to obtain a multi-dimensional risk score; and determining a security risk assessment model based on the set of potential risk factors and the multi-dimensional risk score.

[0085] The potential risk factors can be understood as factors that may affect power grid security, including but not limited to at least one of equipment failure, operational errors, and environmental changes. The hierarchical classification can be understood as organizing potential risk factors hierarchically according to a pre-defined logical relationship. The risk indicator system can be understood as a systematic set of indicators used to quantify and evaluate different types of potential risk factors. The security risk propagation network can be understood as a network model of how risks propagate in the target power grid, including but not limited to at least one of risk sources, propagation paths, and affected objects. The hierarchical decomposition can be understood as breaking down a complex risk propagation network into multiple sub-networks, corresponding to hierarchical risk assessment schemes at different levels. The hierarchical risk assessment scheme can be understood as a risk assessment plan developed based on the results of hierarchical decomposition, including but not limited to specific assessment methods and standards at different levels. The security semantic information can be understood as semantic content such as safety procedures and event descriptions contained in the power grid security knowledge graph. The security rule extraction can be understood as extracting specific rules from the security semantic information to guide how to assess and respond to specific types of risks. The risk assessment rules can be understood as a set of specific rules extracted and organized to guide risk assessment. The fuzzification can be understood as transforming quantitative risk assessment rules into a fuzzy logic form, allowing for a certain degree of uncertainty and flexibility. Fuzzification adapts to the uncertainties in real-world application scenarios, making the assessment results closer to reality and avoiding overly rigid judgments. The risk reasoning mechanism can be understood as a reasoning system built upon the fuzzified risk assessment rules, used to derive risk assessment conclusions. The analytic hierarchy process (AHP) can be understood as a multi-criteria decision analysis method that determines the weights of different factors by comparing their importance, helping to quantify the relative importance of different potential risk factors. The multi-dimensional risk scoring can be understood as combining multiple dimensions (including but not limited to at least one such dimension such as severity, probability of occurrence, or scope of impact) to comprehensively score potential risk factors.

[0086] One alternative implementation involves first traversing the potential risk factors in the power grid safety knowledge graph using depth-first search (DFS) to obtain a set of potential risk factors. Then, a hierarchical clustering algorithm is used to progressively merge the most similar clusters from a single data point within this set, ultimately forming a hierarchical structure to achieve hierarchical classification of power grid safety. The distance update formula for agglomerative hierarchical clustering is: d(u,v) = min{d(s,t):s∈u,t∈v}, where u and v are two clusters, and d(u,v) is the distance between them. By iteratively merging the nearest clusters, a multi-level risk indicator system is obtained, which may include risk indicators at the equipment, system, and management levels. Based on the graph structure characteristics of the power grid safety knowledge graph, the PageRank algorithm is used to perform graph structure association analysis on the set of potential risk factors to identify nodes and edges that play a key role in risk propagation, thereby constructing a safety risk propagation network. Then, based on the risk indicator system, spectral clustering is used to hierarchically decompose the safety risk propagation network. The eigenvectors of the graph Laplacian matrix are calculated using the spectral clustering algorithm: Lv = λv (where L is the graph Laplacian matrix, v is the eigenvector, and λ is the corresponding eigenvalue). By analyzing these eigenvectors, the safety risk propagation network is decomposed into sub-networks at different levels, corresponding to different hierarchical risk assessment schemes. Finally, based on this hierarchical risk assessment scheme, [the following steps are taken]... Graph attention networks consider the local structure and global semantic information of nodes to extract safety rules from the power grid safety knowledge graph, resulting in context-aware risk assessment rules. Then, by introducing fuzzy set theory, deterministic rules in the risk assessment are transformed into fuzzy rules, thus obtaining a risk reasoning mechanism. Next, the analytic hierarchy process (AHP) is used to conduct multi-dimensional risk assessments of the potential risk factor set, comprehensively considering multiple risk dimensions such as severity, probability of occurrence, and scope of impact to obtain multi-dimensional risk assessment results. Finally, ensemble learning methods, such as random forests, are used to weight and fuse the multi-dimensional risk assessment results to automatically learn the weights of different risk dimensions, and multiple weak learners are combined into an initial... An initial risk assessment model was developed, and using pre-set risk assessment sample data, the initial risk assessment model was validated and optimized through cross-validation and parameter optimization techniques such as Bayesian optimization to find the optimal model parameters. Finally, a safety risk assessment model was obtained. The safety risk assessment model not only considers the complex structure and dynamic characteristics of the power grid system, but also integrates professional knowledge and historical experience in the field of safety. As a result, it can conduct multi-dimensional and multi-level assessments of potential safety risks and has the ability of context awareness and fuzzy reasoning. When the model is used in conjunction with real-time operating data and employee safety Q&A data, it can provide timely and accurate risk assessment results for power grid safety production supervision, effectively improving the safety and reliability of the power grid.This allows for the acquisition of real-time operational data of the power grid under supervision and safety Q&A data of the employees under supervision. Real-time operational data is collected in real-time through SCADA (Supervisory Control and Data Acquisition) systems or Internet of Things (IoT) devices, including key parameters such as voltage, current, and power. Safety Q&A data can be collected through online questionnaire systems or intelligent dialogue systems; this data reflects the employees' level of safety knowledge and awareness.

[0087] This technical solution utilizes a power grid safety knowledge graph to hierarchically assess a pre-defined risk assessment model. First, it identifies and categorizes a set of potential risk factors, constructing a risk indicator system and a safety risk propagation network. Next, based on the risk indicator system, it hierarchically decomposes the propagation network, formulating a layered risk assessment scheme and extracting safety rules to form risk assessment rules. By fuzzifying these rules, a risk reasoning mechanism is established, and combined with the analytic hierarchy process (AHP) to achieve multi-dimensional risk assessment. Ultimately, a safety risk assessment model containing multi-dimensional risk scores is obtained, capable of accurately quantifying and predicting potential risks in the power grid, improving the scientific rigor and predictability of risk management, and enhancing the safety and stability of the power grid.

[0088] S240. Obtain safety production-related data of the target power grid, wherein the safety production-related data includes power grid operation characteristic data and power grid safety Q&A data of the target object.

[0089] S250. Use a pre-established safety risk assessment model to perform risk detection on the safety production related data to obtain risk detection feature data.

[0090] S260. Determine the target early warning scheme based on the risk detection feature data.

[0091] S270. The target early warning scheme is predicted by executing the risk prediction model, and a target safety inspection report related to the safe production of the target power grid is generated based on the prediction results.

[0092] The technical solution of this invention acquires safety production reference data of the target power grid, constructs its state feature space to capture the power grid state patterns and their changes, and performs semantic annotation. Based on this state feature space, a power grid safety knowledge graph is further constructed, integrating entities and relationships in power grid operation. The power grid safety knowledge graph is used to perform hierarchical evaluation of the preset risk assessment model. Through the analysis of risk indicator system and propagation network, multi-dimensional risk scoring is achieved, ultimately forming a safety risk assessment model. This not only improves the accuracy and comprehensiveness of risk assessment but also enhances the predictability and response speed of power grid safety management, providing a solid guarantee for the stable operation of the power grid and effectively reducing potential safety risks.

[0093] Example 3

[0094] Figure 3 This is a schematic diagram of the structure of a power grid safety production detection device provided in Embodiment 3 of the present invention. Figure 3 As shown, the device includes: a safety production-related data acquisition module 310, a risk detection module 320, an early warning scheme determination module 330, and a safety detection report generation module 340. Among them,

[0095] The safety production-related data acquisition module 310 is used to acquire safety production-related data of the target power grid, wherein the safety production-related data includes power grid operation characteristic data and power grid safety Q&A data of the target object; the risk detection module 320 is used to perform risk detection on the safety production-related data using a pre-established safety risk assessment model to obtain risk detection characteristic data; the early warning scheme determination module 330 is used to determine the target early warning scheme based on the risk detection characteristic data; and the safety inspection report generation module 340 is used to predict the target early warning scheme by executing a risk prediction model and generate a target safety inspection report related to the safety production of the target power grid based on the prediction results.

[0096] The technical solution of this invention involves acquiring safety production-related data of the target power grid through a safety production-related data acquisition module 310. Since this data includes power grid operation characteristic data and power grid safety Q&A data of the target object, it can integrate related information in the power grid safety production process, improving detection accuracy. Next, a pre-established safety risk assessment model is used to perform risk detection on the safety production-related data, obtaining risk detection characteristic data, which can accurately identify potential risks in power grid operation. Then, a target early warning scheme is determined based on the risk detection characteristic data, enabling early detection and rapid response to power grid safety hazards. Finally, a risk prediction model is executed to predict the target early warning scheme, and a target safety detection report related to the safety production of the target power grid is generated based on the prediction results. This improves the predictability and accuracy of power grid safety management, solves the problems of insufficient data integration, inaccurate risk identification, and imprecise response measures in related technologies, and can comprehensively detect the power grid safety production status, provide accurate early warnings, and significantly improve the efficiency and safety of power grid risk management.

[0097] Based on the above scheme, optionally, the power grid safety production detection device includes: a state feature space determination module, a power grid knowledge graph construction module, and a risk assessment module. The state feature space determination module is used to acquire multiple safety production reference data of the target power grid and determine the state feature space of the target power grid based on the safety production reference data; wherein, the state feature space includes information on the change process of power grid state patterns and annotation information of the power grid state patterns; the knowledge graph construction module is used to construct a knowledge graph based on the state feature space and the safety production reference data to obtain a power grid safety knowledge graph; the risk assessment module is used to perform hierarchical risk assessment on a preset risk assessment model using the power grid safety knowledge graph to obtain a safety risk assessment model.

[0098] Based on the above scheme, optionally, the state feature space determination module includes: a time-frequency transformation submodule, a nonlinear transformation submodule, a cross-domain feature correlation analysis submodule, a feature dimensionality reduction submodule, and a state feature space determination submodule. The system comprises the following components: a time-frequency transformation submodule, used to perform time-series decomposition on the safety production reference data to obtain time-domain power grid features, and using short-time Fourier transform to perform time-frequency transformation on the time-domain power grid features to obtain frequency-domain power grid features; a nonlinear transformation submodule, used to perform nonlinear transformation on the time-domain power grid features and the frequency-domain power grid features to obtain data distribution features; a cross-domain feature association analysis submodule, used to perform cross-domain feature association analysis on the safety production reference data based on the data distribution features to obtain interactive power grid features, wherein the cross-domain feature association analysis is used to determine the relationship between different types of safety production reference data; a feature dimensionality reduction submodule, used to perform feature fusion on the time-domain power grid features, the frequency-domain power grid features, the data distribution features, and the interactive power grid features, and to perform feature dimensionality reduction on the fused features to obtain low-dimensional power grid features; and a state feature space determination submodule, used to cluster the low-dimensional power grid features to obtain multiple clusters, determine the power grid state pattern according to the cluster to which the low-dimensional power grid features belong, perform semantic annotation on the power grid state pattern, and determine the state feature space of the target power grid based on the semantic annotation results.

[0099] Based on the above scheme, optionally, the power grid knowledge graph construction module includes: an initial power grid knowledge graph construction submodule, an adjustment submodule, and a power grid safety knowledge graph construction submodule. The initial power grid knowledge graph construction submodule is used to determine the classification information of the state feature space, construct an initial graph body model based on the classification information, and perform entity recognition and relationship extraction on the safety production reference data based on the initial graph body model to construct an initial power grid knowledge graph. The adjustment submodule is used to align the initial graph body model and the initial power grid knowledge graph to obtain power grid safety knowledge, and adjust the power grid safety knowledge to obtain an initial dynamic knowledge graph. The power grid safety knowledge graph construction submodule is used to construct a power grid safety knowledge graph based on the entities and relationships in the initial dynamic knowledge graph.

[0100] Based on the above scheme, optionally, the risk assessment module includes: a security risk propagation network determination submodule, a security rule extraction submodule, and a security risk assessment model determination submodule. Specifically, the security risk propagation network determination submodule is used to determine a set of potential risk factors based on potential risk factors in the power grid security knowledge graph, perform hierarchical classification of the set of potential risk factors to obtain a risk indicator system, and determine the security risk propagation network corresponding to the set of potential risk factors based on the power grid security knowledge graph; the security rule extraction submodule is used to perform hierarchical decomposition of the security risk propagation network based on the risk indicator system to obtain a hierarchical risk assessment scheme, and extract security rules from the security semantic information in the power grid security knowledge graph based on the hierarchical risk assessment scheme to obtain risk assessment rules; the security risk assessment model determination submodule is used to fuzzify the risk assessment rules to obtain a risk reasoning mechanism, perform multi-dimensional risk assessment on the set of potential risk factors based on the risk reasoning mechanism and the analytic hierarchy process (AHP) to obtain a multi-dimensional risk score, and determine the security risk assessment model based on the set of potential risk factors and the multi-dimensional risk score.

[0101] Based on the above scheme, optionally, the risk detection module includes: a safety score determination submodule, a risk distribution map determination submodule, and a risk detection feature data generation submodule. The safety score determination submodule is used to acquire the operational feature data and the target semantic features in the power grid safety Q&A data, map the operational feature data to the state feature space, determine a preliminary risk point set, and use the power grid safety knowledge graph to perform contextual understanding on the target semantic features to determine the safety score of the target semantic features. The risk distribution map determination submodule is used to use the safety risk assessment model to perform risk detection on each risk point in the preliminary risk point set and the safety score, obtain preliminary risk detection results, and determine the risk clustering area and high-risk period of the target power grid based on the preliminary risk detection results and the time and regional data of the corresponding risk points in the preliminary risk point set, thereby obtaining a risk distribution map. The risk detection feature data generation submodule is used to use the power grid safety knowledge graph to detect the risk causes of the risk clustering area in the high-risk period of the risk distribution map, obtain a preliminary risk explanation list, and perform confidence assessment and structured classification on each explanation in the preliminary risk explanation list based on the corresponding associated risk data and historical risk cases in the power grid safety knowledge graph, thereby generating risk detection feature data.

[0102] Based on the above scheme, optionally, the early warning scheme determination module includes: a risk influencing factor determination submodule, an influence hierarchy structure determination submodule, and a target early warning scheme determination submodule. The risk influencing factor determination submodule is used to determine a preliminary risk score for a power grid safety production risk event based on the risk detection feature data, and to determine the risk influencing factors of the target power grid based on the preliminary risk score and a power grid safety knowledge graph. The preliminary risk score is used to characterize the severity and urgency of the risk. The influence hierarchy structure determination submodule is used to determine the risk impact path based on the risk influencing factors, identify the risk impact objects in the risk impact path, and determine the influence hierarchy structure based on the risk impact objects and the risk influencing factors. The target early warning scheme determination submodule is used to match a detection and early warning response library corresponding to the influence hierarchy structure based on a preset multi-level early warning threshold system, and to determine the target early warning scheme corresponding to the influence hierarchy structure based on the detection and early warning response library.

[0103] The power grid safety production testing device provided in this embodiment of the invention can execute the power grid safety production testing method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of executing the method.

[0104] Example 4

[0105] Figure 4A schematic diagram of an electronic device 10 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0106] like Figure 4 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 may also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0107] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0108] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, central processing unit (CPU), graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as a power grid safety production detection method.

[0109] In some embodiments, a power grid safety production detection method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the power grid safety production detection method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform a power grid safety production detection method by any other suitable means (e.g., by means of firmware).

[0110] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0111] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0112] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0113] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0114] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0115] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0116] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0117] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for detecting power grid safety production, characterized in that, include: Acquire safety production-related data of the target power grid, wherein the safety production-related data includes power grid operation characteristic data and power grid safety Q&A data of the target object; The safety production-related data are subjected to risk detection using a pre-established safety risk assessment model to obtain risk detection feature data. A target early warning scheme is determined based on the aforementioned risk detection feature data; The target early warning scheme is predicted by executing a risk prediction model, and a target safety inspection report related to the safe production of the target power grid is generated based on the prediction results; Determining the security risk assessment model includes: Multiple safety production reference data of the target power grid are acquired, and the state feature space of the target power grid is determined based on the safety production reference data; wherein, the state feature space includes information on the change process of the power grid state mode and the annotation information of the power grid state mode, and the safety production reference data are various historical data related to the safe production and operation of the power grid; Based on the state feature space, a knowledge graph is constructed from the safety production reference data to obtain a power grid safety knowledge graph; The power grid security knowledge graph is used to perform hierarchical risk assessment on the preset risk assessment model to obtain a security risk assessment model; The step of determining the state characteristic space of the target power grid based on the safety production reference data includes: The safety production reference data is decomposed into time series to obtain time-domain power grid characteristics, and then the time-domain power grid characteristics are transformed into frequency-domain power grid characteristics using short-time Fourier transform. The time-domain power grid characteristics and the frequency-domain power grid characteristics are subjected to nonlinear transformation to obtain the data distribution characteristics; Based on the data distribution characteristics, cross-domain feature correlation analysis is performed on the safety production reference data to obtain interactive power grid characteristics, wherein the cross-domain feature correlation analysis is used to determine the relationship between different types of safety production reference data; The time-domain power grid features, the frequency-domain power grid features, the data distribution features, and the interactive power grid features are fused, and the dimensionality of the fused features is reduced to obtain low-dimensional power grid features. The low-dimensional power grid features are clustered to obtain multiple clusters. The power grid state pattern is determined according to the cluster to which the low-dimensional power grid features belong. The power grid state pattern is semantically labeled. The state feature space of the target power grid is determined based on the semantic labeling results.

2. The power grid safety production detection method according to claim 1, characterized in that, The process of constructing a knowledge graph based on the state feature space of the safety production reference data to obtain a power grid safety knowledge graph includes: The classification information of the state feature space is determined, an initial graph subject model is constructed based on the classification information, and entity recognition and relationship extraction are performed on the safety production reference data based on the initial graph subject model to construct an initial power grid knowledge graph. Alignment processing is performed on the initial graph main model and the initial power grid knowledge graph to obtain power grid security knowledge, and the power grid security knowledge is adjusted to obtain an initial dynamic knowledge graph; A power grid security knowledge graph is constructed based on the entities and relationships in the initial dynamic knowledge graph.

3. The power grid safety production detection method according to claim 1, characterized in that, The process of using the power grid security knowledge graph to perform hierarchical risk assessment on a preset risk assessment model to obtain a security risk assessment model includes: Based on the potential risk factors in the power grid security knowledge graph, a set of potential risk factors is determined, the set of potential risk factors is hierarchically classified to obtain a risk indicator system, and a security risk propagation network corresponding to the set of potential risk factors is determined based on the power grid security knowledge graph. Based on the risk indicator system, the security risk propagation network is hierarchically decomposed to obtain a hierarchical risk assessment scheme. Based on the hierarchical risk assessment scheme, security rules are extracted from the security semantic information in the power grid security knowledge graph to obtain risk assessment rules. The risk assessment rules are fuzzified to obtain a risk reasoning mechanism. Based on the risk reasoning mechanism and the analytic hierarchy process, a multi-dimensional risk assessment is performed on the set of potential risk factors to obtain a multi-dimensional risk score. A safety risk assessment model is determined based on the set of potential risk factors and the multi-dimensional risk score.

4. The power grid safety production detection method according to claim 1, characterized in that, The method of using a pre-established safety risk assessment model to perform risk detection on the safety production-related data yields risk detection feature data, including: The operation feature data and the target semantic features in the power grid safety Q&A data are obtained, and the operation feature data is mapped to the state feature space to determine a preliminary set of risk points. The power grid safety knowledge graph is used to perform contextual understanding on the target semantic features to determine the safety score of the target semantic features. The safety risk assessment model is used to detect risks in each risk point and the safety score in the preliminary risk point set to obtain preliminary risk detection results. Based on the preliminary risk detection results and the time and regional data of the corresponding risk points in the preliminary risk point set, the risk clustering area and high-risk period of the target power grid are determined to obtain a risk distribution map. The power grid safety knowledge graph is used to detect the risk causes of risk clusters during high-risk periods in the risk distribution map, and a preliminary risk explanation list is obtained. Based on the corresponding associated risk data and historical risk cases in the power grid safety knowledge graph, the confidence level of each explanation in the preliminary risk explanation list is evaluated and structured, and risk detection feature data is generated.

5. The power grid safety production testing method according to claim 1, characterized in that, The step of determining the target early warning scheme based on the risk detection feature data includes: A preliminary risk score for a power grid safety production risk event is determined based on the risk detection feature data. Based on the preliminary risk score and the power grid safety knowledge graph, the risk influencing factors of the target power grid are determined. The preliminary risk score is used to characterize the severity and urgency of the risk. Based on the risk influencing factors, determine the risk influencing path, identify the risk influencing objects in the risk influencing path, and determine the influencing hierarchy structure based on the risk influencing objects and the risk influencing factors; Based on a preset multi-level early warning threshold system, a detection and early warning response library corresponding to the influence hierarchy structure is matched, and a target early warning scheme corresponding to the influence hierarchy structure is determined according to the detection and early warning response library.

6. A power grid safety production detection device, characterized in that, include: The safety production-related data acquisition module is used to acquire safety production-related data of the target power grid, wherein the safety production-related data includes power grid operation characteristic data and power grid safety Q&A data of the target object; The risk detection module is used to perform risk detection on the safety production-related data using a pre-established safety risk assessment model, and obtain risk detection feature data. The early warning scheme determination module is used to determine the target early warning scheme based on the risk detection feature data. The safety inspection report generation module is used to predict the target early warning scheme by executing a risk prediction model, and generate a target safety inspection report related to the safe production of the target power grid based on the prediction results; The power grid safety production detection device includes: The state feature space determination module is used to acquire multiple safety production reference data of the target power grid and determine the state feature space of the target power grid based on the safety production reference data; wherein, the state feature space includes information on the change process of the power grid state mode and the annotation information of the power grid state mode, and the safety production reference data are various historical data related to the safe production and operation of the power grid; The knowledge graph construction module is used to construct a knowledge graph based on the state feature space and the safety production reference data to obtain a power grid safety knowledge graph. The risk assessment module is used to perform hierarchical risk assessment on the preset risk assessment model using the power grid security knowledge graph to obtain a security risk assessment model. The state feature space determination module includes: The time-frequency transformation submodule is used to perform time-series decomposition on the safety production reference data to obtain time-domain power grid characteristics, and to perform time-frequency transformation on the time-domain power grid characteristics using short-time Fourier transform to obtain frequency-domain power grid characteristics. The nonlinear transformation submodule is used to perform nonlinear transformation on the time-domain power grid characteristics and the frequency-domain power grid characteristics to obtain data distribution characteristics; The cross-domain feature association analysis submodule is used to perform cross-domain feature association analysis on the safety production reference data based on the data distribution characteristics to obtain interactive power grid characteristics. The cross-domain feature association analysis is used to determine the relationship between different types of safety production reference data. The feature dimensionality reduction submodule is used to perform feature fusion on the time-domain power grid features, the frequency-domain power grid features, the data distribution features, and the interactive power grid features, and to perform feature dimensionality reduction on the fused features to obtain low-dimensional power grid features; The state feature space determination submodule is used to cluster the low-dimensional power grid features to obtain multiple clusters, determine the power grid state pattern according to the cluster to which the low-dimensional power grid features belong, perform semantic annotation on the power grid state pattern, and determine the state feature space of the target power grid based on the semantic annotation results.

7. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the power grid safety production detection method according to any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that are used to cause a processor to execute the power grid safety production detection method according to any one of claims 1-5.

Images