Data Analysis Method and System Based on Smart Grid

Abstract

This application relates to the fields of smart grid, artificial intelligence, and big data technology, and in particular provides a data analysis method and system based on the smart grid. This application deeply applies artificial intelligence technology to mine and analyze various types of data in the smart grid, so as to extract the key features of the grid operation and discover potential faults in a timely manner. In addition, this application effectively improves the operation and maintenance efficiency of the smart grid, reduces the risk of faults, and provides a strong guarantee for the stable operation of the power system. Additionally, by accurately predicting the development trend of abnormal state events, big data analysis can be combined to optimize the grid resource allocation, achieving efficient utilization of energy.

Description

Technical Field

[0001] The present application relates to the fields of smart grid, artificial intelligence, and big data technology, and in particular to a data analysis method and system based on smart grid. Background Art

[0002] With the rapid development of smart grid technology, the complexity and refinement requirements of power system operation and maintenance are increasing. In the process of power operation and maintenance, how to accurately identify and predict the development trend of various abnormal state events has become the key to improving the stability and security of the power grid. Traditional power operation and maintenance methods mainly rely on regular inspections, post-maintenance, and alarm systems based on static thresholds. These methods often find it difficult to capture the dynamic changes of the power system in a timely manner, and cannot accurately predict the development trend of abnormal state events.

[0003] In order to overcome the limitations of traditional methods, analysis technology based on smart grid operation data has gradually become a research hotspot in recent years. However, most existing data analysis methods only focus on current operation data, ignoring valuable information in historical operation and maintenance scenarios, such as past abnormal state events and their development trends. This information is of great reference value for understanding and predicting the development of current abnormal state events. Summary of the invention

[0004] In order to improve the above problems, the present application provides a data analysis method and system based on smart grid.

[0005] The present application provides a data analysis method based on a smart grid, which is applied to a smart grid analysis system. The method includes:

[0006] According to the target power operation and maintenance scenario corresponding to the smart grid operation data to be analyzed, determine the past power operation and maintenance scenarios that meet the scenario commonality determination requirements with the target power operation and maintenance scenario, and determine the preceding abnormal state event corresponding to the target abnormal state event to be analyzed; the preceding abnormal state event and the target abnormal state event have at least partially the same fault risk state variables;

[0007] Determine a first abnormal state development trend vector of the target abnormal state event and a second abnormal state development trend vector of the preceding abnormal state event; the first abnormal state development trend vector is used to characterize the trend change of the target abnormal state event in response to the existing fault operation and maintenance label in the target power operation and maintenance scenario; the second abnormal state development trend vector is used to characterize the trend change of the preceding abnormal state event in response to the existing fault operation and maintenance label in the past power operation and maintenance scenario;

[0008] Performing a trend vector integration operation on the second abnormal state development trend vector and the first abnormal state development trend vector to obtain a current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario;

[0009] The current abnormal state development trend vector and the global state development trend vector of the smart grid operation data are jointly analyzed to determine the correlation prediction result between the smart grid operation data and the target abnormal state event.

[0010] Preferably, the number of the preceding abnormal state events is at least two, and the target abnormal state event is one of the preceding abnormal state events;

[0011] The method further includes: obtaining a first abnormal state development trend vector of a candidate abnormal state event other than the target abnormal state event in each of the preceding abnormal state events; the first abnormal state development trend vector of the candidate abnormal state event is used to characterize the trend change of the candidate abnormal state event in response to an existing fault operation and maintenance tag in the target power operation and maintenance scenario;

[0012] The trend vector integration operation is performed on the second abnormal state development trend vector and the first abnormal state development trend vector to obtain the current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario, including: determining the target state development trend commonality value of the target abnormal state event and the candidate abnormal state event in the target power operation and maintenance scenario based on the first abnormal state development trend vector corresponding to each of the preceding abnormal state events; determining the past state development trend commonality value of the target abnormal state event and the candidate abnormal state event in the past power operation and maintenance scenario based on the second abnormal state development trend vector corresponding to each of the preceding abnormal state events; and mining the past state development trend commonality value based on the target state development trend commonality value to obtain the current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario.

[0013] Preferably, based on the target state development trend commonality value, the commonality value of the past state development trend is mined to obtain the current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario, including determining the difference between each of the first abnormal state development trend vectors based on the distribution area projected onto the characteristic relationship network of the target power operation and maintenance scenario; the difference between two first abnormal state development trend vectors is used to characterize the target state development trend commonality value between the two first abnormal state development trend vectors;

[0014] According to the commonality value of the past state development trend, adjusting the difference between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector of the candidate abnormal state event, and obtaining a distribution adjustment area of ​​the first abnormal state development trend vector of the target abnormal state event in the feature relationship network;

[0015] A current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario is determined according to the distribution adjustment area.

[0016] Preferably, the difference between the two first abnormal state development trend vectors has a first quantitative relationship with the target state development trend commonality value between the two first abnormal state development trend vectors;

[0017] The step of adjusting the difference between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector of the candidate abnormal state event according to the commonality value of the past state development trend includes:

[0018] If the commonality value of the past state development trend meets the commonality index, narrowing the difference between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector of the candidate abnormal state event;

[0019] If the commonality value of the past state development trend meets the mutually exclusive index, the difference between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector of the candidate abnormal state event is amplified; the power operation and maintenance scenarios corresponding to the commonality index and the mutually exclusive index are different.

[0020] Preferably, the number of the candidate abnormal state events is at least two; the determining of the target state development trend commonality value of the target abnormal state event and the candidate abnormal state event in the target power operation and maintenance scenario based on the first abnormal state development trend vector corresponding to each of the preceding abnormal state events includes: respectively determining the joint commonality score between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector corresponding to each of the candidate abnormal state events;

[0021] The method of determining the degree of difference between each of the first abnormal state development trend vectors based on the distribution area projected onto the characteristic relationship network of the target power operation and maintenance scenario includes: determining the degree of difference between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector corresponding to each of the candidate abnormal state events based on the distribution area projected onto the characteristic relationship network of the target power operation and maintenance scenario.

[0022] Preferably, the number of the candidate abnormal state events is at least two; the method further comprises: obtaining the abnormal state event knowledge vector corresponding to each preceding abnormal state event, and determining the preceding abnormal state event to be processed that meets the commonality index of the abnormal state event with the target abnormal state event from each of the candidate abnormal state events;

[0023] The method of adjusting the difference between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector of the candidate abnormal state event based on the commonality value of the past state development trends to obtain the distribution adjustment area of ​​the first abnormal state development trend vector of the target abnormal state event in the feature relationship network includes: adjusting the difference between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector of the candidate abnormal state event based on the commonality value of the previous state development trends of the target abnormal state event and the previous abnormal state event to be processed to obtain the distribution adjustment area of ​​the first abnormal state development trend vector of the target abnormal state event in the feature relationship network.

[0024] Preferably, the step of obtaining the current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario includes:

[0025] Based on the target deep learning algorithm, mining the commonality values ​​of the past state development trends between the second abnormal state development trend vectors, to obtain the current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario;

[0026] The steps of debugging and obtaining the target deep learning algorithm include:

[0027] According to the global debugging error determined by the target power operation and maintenance scenario identification error and the consistency analysis error, the basic deep learning algorithm is debugged using the target power operation and maintenance scenario debugging example set to obtain a target deep learning algorithm for determining the current abnormal state development trend vector in the target power operation and maintenance scenario; the target power operation and maintenance scenario debugging example set includes: the abnormal state event knowledge vector of each abnormal state event sample in the target power operation and maintenance scenario, the data knowledge of each smart grid operation data, the target power operation and maintenance of each abnormal state event sample in response to the fault operation and maintenance label in the target power operation and maintenance scenario The scene interaction information, and the second abnormal state development trend vector corresponding to each of the abnormal state event samples in the past power operation and maintenance scenes; the error parameter of the target power operation and maintenance scene identification error is the abnormal state event knowledge vector of each of the abnormal state event samples in the target power operation and maintenance scene, and the data knowledge of each of the smart grid operation data; the error parameter of the consistency analysis error is the target state development trend commonality value of each of the abnormal state event samples in the target power operation and maintenance scene, and the past state development trend commonality value of each of the abnormal state event samples in the past power operation and maintenance scenes.

[0028] Preferably, the step of determining the first abnormal state development trend vector of the target abnormal state event and the second abnormal state development trend vector of the preceding abnormal state event comprises:

[0029] Obtain target power operation and maintenance scenario interaction information in which the target abnormal state event responds to an existing fault operation and maintenance tag in the target power operation and maintenance scenario, and past power operation and maintenance scenario interaction information in which the preceding abnormal state event responds to an existing fault operation and maintenance tag in the past power operation and maintenance scenario; the feature granularity of the target power operation and maintenance scenario interaction information is smaller than the feature granularity of the past power operation and maintenance scenario interaction information;

[0030] Performing trend change mining on the target abnormal state event according to the target power operation and maintenance scenario interaction information to obtain a first abnormal state development trend vector of the target abnormal state event;

[0031] The trend change of the preceding abnormal state event is mined according to the past electric power operation and maintenance scenario interaction information to obtain a second abnormal state development trend vector of the preceding abnormal state event.

[0032] Preferably, the number of the past power operation and maintenance scenarios is several; the step of determining the second abnormal state development trend vector of the preceding abnormal state event comprises:

[0033] Obtaining an abnormal state preceding trend vector of a preceding abnormal state event in each of the past power operation and maintenance scenarios;

[0034] The abnormal state preceding trend vectors corresponding to the preceding abnormal state events in each of the past power operation and maintenance scenarios are integrated to obtain a second abnormal state development trend vector of the preceding abnormal state event.

[0035] An embodiment of the present application provides a smart grid analysis system, comprising at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor performs the above method.

[0036] An embodiment of the present application provides a computer-readable storage medium on which a computer program is stored. The computer program implements the above method when running.

[0037] In the embodiment of the present application, by deeply mining the smart grid operation data, the embodiment of the present application not only focuses on the current target abnormal state event, but also associates it with the previous abnormal state events in the past power operation and maintenance scenarios. These previous abnormal state events and the target abnormal state events have at least some of the same fault risk state variables, so they can provide important clues for predicting the development trend of the target abnormal state event.

[0038] To achieve this goal, the embodiment of the present application first determines the first abnormal state development trend vector of the target abnormal state event and the second abnormal state development trend vector of the preceding abnormal state event. These two vectors respectively characterize the trend changes of the target abnormal state event and the preceding abnormal state event in different operation and maintenance scenarios. Then, by performing a trend vector integration operation on these two vectors, the comprehensive development trend of the target abnormal state event in the current operation and maintenance scenario is obtained.

[0039] Finally, the embodiment of the present application jointly analyzes the current abnormal state development trend vector and the global state development trend vector of the smart grid operation data to determine the correlation prediction result between the operation data and the abnormal state event. This method not only improves the accuracy of abnormal state event prediction, but also provides strong support for preventive maintenance and optimization of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 A flow chart of a data analysis method based on a smart grid provided in an embodiment of the present application.

[0041] Figure 2 A schematic diagram of the structure of a smart grid analysis system 200 provided in an embodiment of the present application. DETAILED DESCRIPTION

[0042] In order to better understand the above technical scheme, the technical scheme of the present application is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present application and the specific features in the embodiments are detailed descriptions of the technical scheme of the present application, rather than limitations on the technical scheme of the present application. In the absence of conflict, the embodiments of the present application and the technical features in the embodiments can be combined with each other.

[0043] Figure 1 A data analysis method based on a smart grid is shown, which is applied to a smart grid analysis system. The method includes the following steps 110 to 140.

[0044] Step 110: Based on the target power operation and maintenance scenario corresponding to the smart grid operation data to be analyzed, determine the past power operation and maintenance scenarios that meet the scenario commonality determination requirements with the target power operation and maintenance scenario, and determine the preceding abnormal state event corresponding to the target abnormal state event to be analyzed; the preceding abnormal state event and the target abnormal state event have at least some of the same fault risk state variables.

[0045] Step 120, determine the first abnormal state development trend vector of the target abnormal state event, and the second abnormal state development trend vector of the previous abnormal state event; the first abnormal state development trend vector is used to characterize the trend change of the target abnormal state event in response to the existing fault operation and maintenance label in the target power operation and maintenance scenario; the second abnormal state development trend vector is used to characterize the trend change of the previous abnormal state event in response to the existing fault operation and maintenance label in the past power operation and maintenance scenario.

[0046] Step 130: Perform a trend vector integration operation on the second abnormal state development trend vector and the first abnormal state development trend vector to obtain a current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario.

[0047] Step 140: jointly analyze the current abnormal state development trend vector and the global state development trend vector of the smart grid operation data to determine a prediction result of the correlation between the smart grid operation data and the target abnormal state event.

[0048] In an exemplary application scenario, the smart grid analysis system receives a batch of smart grid operation data, which reflects the current power operation and maintenance situation, called the target power operation and maintenance scenario. The system first analyzes the data and identifies the historical power operation and maintenance scenarios that match it. The historical scenarios and the target scenarios meet the common determination requirements, such as the topology of the power grid, the type of equipment, and the operating environment. At the same time, the system identifies the preceding abnormal state events related to the target abnormal state events. The preceding events have at least some of the same fault risk state variables as the target events, such as voltage fluctuations, frequency anomalies, etc.

[0049] Next, the smart grid analysis system further analyzes the development trends of the target abnormal state event and the previous abnormal state event. For the target abnormal state event, the system generates a first abnormal state development trend vector, which describes the trend change of the target event in the current power operation and maintenance scenario in response to the fault operation and maintenance tags that have appeared (such as overload, short circuit, etc.). Similarly, for the previous abnormal state event, the system generates a second abnormal state development trend vector, which reflects the change trend of the above event in the past power operation and maintenance scenarios.

[0050] Then, the smart grid analysis system integrates the first abnormal state development trend vector and the second abnormal state development trend vector. This integration process takes into account the weight and correlation of the two trend vectors, and finally forms a comprehensive current abnormal state development trend vector, which more comprehensively reflects the development trend of the target abnormal state event in the target power operation and maintenance scenario.

[0051] Finally, the smart grid analysis system jointly analyzes the current abnormal state development trend vector with the global state development trend vector of the smart grid operation data. The global state development trend vector reflects the overall operation trend of the entire power grid system. By comparing and analyzing these two vectors, the system can determine the correlation prediction results between the smart grid operation data and the target abnormal state event, thereby more accurately judging the health status of the power grid and potential failure risks.

[0052] Through the above steps, the smart grid analysis system can achieve real-time monitoring of the grid operation status and fault warning, effectively improve the operation and maintenance efficiency of the smart grid, reduce the risk of failure, and provide strong guarantee for the stable operation of the power system. At the same time, the system can also optimize the allocation of grid resources through data analysis and achieve efficient use of energy.

[0053] In combination with the above application scenarios, steps 110 to 140 are further introduced below.

[0054] In step 110, the smart grid operation data to be analyzed refers to the grid operation data collected in real time by the smart grid system and transmitted to the analysis system, including but not limited to various parameters such as voltage, current, power factor, and equipment status. The above data reflects the real-time operation status of the power grid and is an important basis for power grid monitoring, fault warning, and optimal configuration. The target power operation and maintenance scenario refers to the power operation and maintenance environment or scenario that the smart grid analysis system is currently concerned about. It can be a specific power grid area, a device cluster, or a specific operating period. The target power operation and maintenance scenario is the focus of data analysis, and all data analysis and prediction are based on this scenario. The scenario commonality judgment requirement refers to a series of judgment criteria or conditions based on when analyzing past power operation and maintenance scenarios and target power operation and maintenance scenarios. The above requirements may include aspects such as grid structure similarity, equipment type consistency, and operating environment equivalence to ensure the comparability of historical data with current data. The past power operation and maintenance scenario refers to the scenario corresponding to the historical power operation and maintenance data stored in the smart grid analysis system. The above data records the past operation of the power grid and the abnormal events that occurred, which has important reference value for analyzing the current power grid status and predicting future trends. The target abnormal state event to be analyzed refers to the abnormal state or event of the power grid that the smart grid analysis system is currently paying attention to, such as voltage sag, frequency anomaly, equipment failure, etc. The above events may pose a threat to the stable operation of the power grid, and the system needs to identify and respond in time. The previous abnormal state event refers to an abnormal state event that occurred in the past and has a certain similarity in nature or impact with the target abnormal state event. Although the above events occur at different time points, they may have the same fault risk state variables as the target abnormal state event, such as voltage instability, overload, etc., so they can be used to assist in analyzing the development trend and possible causes of the target abnormal state event. Fault risk state variables refer to a series of variables used to describe the abnormal state or fault risk of the power grid, such as voltage, current, temperature, pressure, etc. The above variables will change during the operation of the power grid. When they exceed the normal range, they may cause abnormal state or fault of the power grid.

[0055] Further, with respect to step 110, the smart grid analysis system first receives a large amount of real-time smart grid operation data, which is a direct reflection of the current operation state of the power grid. The system performs preliminary processing and analysis on the above data to identify the target power operation and maintenance scenarios to which they correspond. Once the target power operation and maintenance scenario is determined, the system will begin to look for similar past power operation and maintenance scenarios. This process is achieved by comparing the common determination requirements of the two scenarios, such as whether the topological structures of the power grids are similar, whether the types of equipment involved are consistent, whether the operating environment is equivalent, etc. After finding the qualified past power operation and maintenance scenarios, the system further analyzes the abnormal state events recorded in the above historical data, especially those preceding abnormal state events with similar properties to the target abnormal state events. The above preceding events are important because they may reveal the development laws and potential causes of the target abnormal state events. In order to find the above valuable preceding events, the system checks whether they have at least some of the same fault risk state variables as the target events, such as voltage fluctuations, current overloads, etc. Through such analysis, the smart grid analysis system can establish a comprehensive data foundation to provide strong support for subsequent in-depth analysis of the abnormal status and failure risks of the power grid.

[0056] In step 120, the first abnormal state development trend vector refers to a development trend description generated by the smart grid analysis system for the target abnormal state event of current concern. The vector collects multi-dimensional data (such as time, abnormal index value, etc.) of the target abnormal state event in the target power operation and maintenance scenario, and uses data analysis technology to process it, so as to obtain a mathematical model or data representation that can reflect the development trend of the abnormal state event. This vector not only reveals the current evolution direction of the target abnormal state event, but also predicts its possible future development trend, which is crucial for timely discovery of potential risks and formulation of countermeasures. The second abnormal state development trend vector is similar to the first abnormal state development trend vector, but the second abnormal state development trend vector is generated for the previous abnormal state event in the past power operation and maintenance scenario. The smart grid analysis system reviews historical data, finds out the previous events related to the target abnormal state event, and analyzes the development trend of the above events in the operation and maintenance scenario at that time. The second abnormal state development trend vector is such a mathematical description of a historical trend, which helps to understand how similar abnormal state events have developed in the past, and provides a valuable reference for predicting and responding to the current target abnormal state event. Trend changes of existing fault operation and maintenance labels: During the operation and maintenance of the smart grid, once a fault or abnormal state occurs, the system will label it with a corresponding fault operation and maintenance label, such as "voltage abnormality" or "equipment failure". The above labels not only record the type and location of the fault, but also include information such as the time and severity of the fault. The trend changes of existing fault operation and maintenance labels refer to the changing trends in the frequency, duration or severity of the above labels in historical data. The smart grid analysis system monitors and analyzes the above trend changes in order to promptly detect potential fault risks and take corresponding preventive measures.

[0057] Furthermore, after determining the similarity between the target power operation and maintenance scenario and the past power operation and maintenance scenarios, the smart grid analysis system further delves into the development trends of the target abnormal state event and the previous abnormal state events. The core of this step is to generate two key development trend vectors: the first abnormal state development trend vector and the second abnormal state development trend vector. First, the smart grid analysis system focuses on the target abnormal state event and examines how this event responds to the existing fault operation and maintenance labels in the target power operation and maintenance scenario. The system collects the status data of this event under various operation and maintenance labels, such as the changes in abnormal indicators, the duration of the fault, and the affected scope. Through comprehensive analysis and processing of the above data, the system can extract a first abnormal state development trend vector. This vector not only reflects the development trend of the target abnormal state event in the current scenario but also reveals how it is influenced by the existing fault operation and maintenance labels and the possible future evolution direction. Second, the system reviews and analyzes the previous abnormal state events related to the target abnormal state event. How did the above previous events develop in the past power operation and maintenance scenarios and how did they respond to the fault operation and maintenance labels at that time? The smart grid analysis system uses historical data to construct a second abnormal state development trend vector. This vector describes the evolution law of the previous events in a similar scenario and provides an important historical reference for understanding and predicting the target abnormal state event. It can be seen that step 120 is a crucial link in the smart grid analysis system that connects the preceding and the following. It generates two key development trend vectors through in-depth analysis of the target abnormal state event and the previous abnormal state events. These two vectors not only provide a comprehensive perspective on the abnormal state events but also lay a solid foundation for subsequent prediction and formulation of response strategies.

[0058] In step 130, the trend vector integration operation is used to achieve the comprehensive processing of two or more abnormal state development trend vectors (in this scenario, the first abnormal state development trend vector and the second abnormal state development trend vector). This process involves the weight evaluation, data fusion, and algorithm processing of each vector to generate a new, integrated trend vector. This integrated trend vector can more comprehensively and accurately reflect the comprehensive development trend of the target abnormal state event in a specific power operation and maintenance scenario. The current abnormal state development trend vector refers to the new vector obtained after the trend vector integration operation. It integrates the development trend information of the target abnormal state event in the target power operation and maintenance scenario and the previous abnormal state events in the past power operation and maintenance scenarios. This vector provides the latest and most comprehensive development trend analysis of the target abnormal state event in the current scenario, helping to predict future state changes and formulate timely response strategies.

[0059] In detail, the smart grid analysis system performs a key data integration task, namely the trend vector integration operation. The purpose of this operation is to merge two important abnormal state development trend vectors (the first abnormal state development trend vector and the second abnormal state development trend vector) to obtain a more comprehensive and accurate current abnormal state development trend vector.

[0060] (1) Data preparation: The system first ensures that the first abnormal state development trend vector and the second abnormal state development trend vector have been accurately generated. These two vectors represent the development trend of the target abnormal state event in the target power operation and maintenance scenario, and the development trend of the previous abnormal state event in similar scenarios in the past.

[0061] (2) Weight assignment: Before the integration operation, the system assigns different weights to the two vectors based on the importance, timeliness, and relevance of the information they contain. This usually involves complex algorithms and data analysis to ensure the scientificity and accuracy of the weights.

[0062] (3) Vector fusion: Next, the smart grid analysis system fuses the two weighted vectors using corresponding algorithms (such as weighted average, machine learning algorithm, etc.). This process aims to extract common trends and features in the two vectors while weakening or excluding possible noise and outliers.

[0063] (4) Result output: Finally, after the trend vector integration operation, the system generates a new current abnormal state development trend vector. This vector integrates the development trend information of the target abnormal state event and the previous abnormal state event, providing a more accurate and comprehensive abnormal state development trend prediction tool for power grid operation and maintenance personnel.

[0064] In this way, the smart grid analysis system can more effectively monitor the health of the power grid, promptly detect potential failure risks, and develop targeted prevention and response measures.

[0065] In step 140, conjoint analysis is a multivariate statistical method that comprehensively considers the interrelationships between multiple variables to reveal their interactions and influences. In smart grid analysis, conjoint analysis can be used to simultaneously examine the relationships between multiple variables such as grid operation data, abnormal state events, and environmental factors, so as to more accurately identify problems and risks in grid operation. Through conjoint analysis, the correlation, degree of influence, and possible causal relationship between various variables can be obtained, providing a scientific basis for the optimized operation and fault prevention of the grid. In addition, in smart grid analysis, the correlation prediction result refers to the result of predicting and judging the correlation between grid operation data and specific abnormal state events through methods such as conjoint analysis. This prediction result can help understand which factors in the grid operation data are closely related to the occurrence of abnormal state events, and how changes in these factors affect the development trend of abnormal state events. The correlation prediction result is an important basis for smart grid fault warning and prevention, and helps to improve the safety and stability of the grid.

[0066] In detail, the smart grid analysis system will perform key data joint analysis operations in step 140 to determine the correlation prediction results between the smart grid operation data and the target abnormal state event.

[0067] (1) Data preparation and integration: The system will first collect and organize the current abnormal state development trend vector, which reflects the real-time development trend of the target abnormal state event in the power operation and maintenance scenario. At the same time, the system will also obtain the global state development trend vector of the smart grid operation data, which represents the operation status and trend of the entire power grid system.

[0068] (2) Joint analysis process: Using advanced statistical models and algorithms, such as regression analysis and time series analysis, the smart grid analysis system will conduct a joint analysis of the two vectors. This process aims to explore the potential correlation between the global operation status of the power grid and the target abnormal state event. The system pays special attention to those power grid operation data indicators that are highly correlated with the development trend of the target abnormal state event, such as voltage fluctuations and current changes.

[0069] (3) Determination of correlation prediction results: Through joint analysis, the system can determine which grid operation data indicators have significant correlations with the target abnormal state events. These correlations can be positive or negative, reflecting how small changes in the grid operation status affect the development trend of abnormal state events. The system also evaluates the strength and stability of this correlation to determine the reliability of the prediction results.

[0070] (4) Result output and application: The correlation prediction results will be presented to the grid operation and maintenance personnel in the form of a visual report or data analysis report. This helps them better understand the relationship between the grid operation status and abnormal status events, so as to formulate more effective prevention and response measures. These prediction results can also be used to optimize the grid operation strategy and improve the safety and stability of the grid. For example, the probability of abnormal status events can be reduced by adjusting grid parameters or improving equipment configuration.

[0071] By applying the embodiments of the present application, firstly, by determining the past power operation and maintenance scenarios that have commonalities with the target power operation and maintenance scenarios, and identifying the preceding abnormal state events corresponding to the target abnormal state events, it is possible to more accurately capture and analyze the potential fault risk state variables in the power system. In this way, not only the accuracy of fault warning is improved, but also early signs that may cause serious faults are discovered and handled in a timely manner.

[0072] Secondly, by determining the development trend vectors of the target abnormal state event and the previous abnormal state event, we can gain in-depth insights into the evolution of these abnormal states in their respective operation and maintenance scenarios. This accurate grasp of trends enables the operation and maintenance team to make effective predictions before a fault occurs, and formulate targeted prevention and response measures accordingly, thereby significantly improving the stability and safety of the power system.

[0073] Furthermore, by integrating the second abnormal state development trend vector and the first abnormal state development trend vector, the present application can generate a vector that comprehensively reflects the development trend of the target abnormal state event in the current operation and maintenance scenario. This innovative data processing method not only improves the efficiency and accuracy of data analysis, but also provides a more comprehensive and intuitive fault warning tool for operation and maintenance personnel.

[0074] Finally, by jointly analyzing the current abnormal state development trend vector and the global state development trend vector of the smart grid operation data, the correlation between the smart grid operation data and the target abnormal state event can be accurately predicted. This prediction capability is of vital importance for optimizing the grid operation strategy, preventing potential failures, and improving the operation efficiency of the entire power system.

[0075] In summary, through a series of innovative data analysis and processing methods, the fault warning and prevention capabilities of smart grids have been significantly improved, providing strong support for ensuring the safe and stable operation of power systems.

[0076] In some possible examples, the number of the preceding abnormal state events is at least two, and the target abnormal state event is one of the preceding abnormal state events.

[0077] The method also includes: obtaining the first abnormal state development trend vector of the candidate abnormal state event other than the target abnormal state event in each of the preceding abnormal state events; the first abnormal state development trend vector of the candidate abnormal state event is used to characterize the trend change of the candidate abnormal state event in response to the existing fault operation and maintenance label in the target power operation and maintenance scenario.

[0078] Based on this, the trend vector integration operation is performed on the second abnormal state development trend vector and the first abnormal state development trend vector to obtain the current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario, including: determining the target state development trend commonality value of the target abnormal state event and the candidate abnormal state event in the target power operation and maintenance scenario based on the first abnormal state development trend vector corresponding to each of the preceding abnormal state events; determining the past state development trend commonality value of the target abnormal state event and the candidate abnormal state event in the past power operation and maintenance scenario based on the second abnormal state development trend vector corresponding to each of the preceding abnormal state events; based on the target state development trend commonality value, mining the past state development trend commonality value to obtain the current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario.

[0079] In the above example, when faced with multiple preceding abnormal state events, and the target abnormal state event is a specific event among these preceding abnormalities, the smart grid analysis system performs more complex analysis and processing.

[0080] For example, the system will first obtain the first abnormal state development trend vectors of other candidate abnormal state events except the target abnormal state event in each previous abnormal state event. These first abnormal state development trend vectors describe how the candidate abnormal state events respond and change to the existing fault operation and maintenance labels in the target power operation and maintenance scenario. In short, this step helps the system understand the development trend of other related abnormal states except the target abnormal state event in the current operation and maintenance scenario of concern.

[0081] Next, the system performs trend vector integration operations, with the goal of synthesizing the information of multiple abnormal state events to derive the development trend of the target abnormal state event in the current operation and maintenance scenario. This process is divided into several key steps: First, the system determines the target state development trend commonality value of the target abnormal state event and the candidate abnormal state event in the target power operation and maintenance scenario based on the first abnormal state development trend vector corresponding to each previous abnormal state event. This commonality value reflects the similarity or correlation between the target abnormal state event and other candidate events in the development trend in the current operation and maintenance scenario; secondly, the system considers information in past power operation and maintenance scenarios. It determines the past state development trend commonality value of the target abnormal state event and the candidate abnormal state event in the past power operation and maintenance scenario based on the second abnormal state development trend vector corresponding to each previous abnormal state event. This commonality value reveals the commonalities or differences in the development trends of these abnormal state events in history; finally, the system conducts in-depth mining of the past state development trend commonality value based on the target state development trend commonality value calculated previously. The purpose of this step is to comprehensively evaluate the development trend of the target abnormal state event in the target power operation and maintenance scenario by combining current and past information. Through this comprehensive consideration, the system can generate a more accurate and comprehensive development trend vector of the current abnormal state.

[0082] In this way, not only the accuracy of the prediction of the development trend of the target abnormal state event is improved, but also the comprehensive analysis capability of the smart grid analysis system for multiple related abnormal state events is enhanced. In this way, the system can more effectively identify and utilize the correlation between different abnormal state events, thereby providing more reliable early warning and decision support for power operation and maintenance personnel. It can be seen that by comprehensively considering the development trends of multiple previous abnormal state events and combining current and past operation and maintenance scenario information, the smart grid analysis system can generate a more accurate current abnormal state development trend vector. This not only helps to improve the stability and safety of the power system, but also provides more efficient and accurate decision-making basis for operation and maintenance personnel, thereby optimizing the power operation and maintenance process and reducing the risk of failure.

[0083] In the next step, based on the commonality value of the target state development trend, the commonality value of the past state development trend is mined to obtain the current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario, including: determining the difference between each first abnormal state development trend vector based on the distribution area projected onto the feature relationship network of the target power operation and maintenance scenario; the difference between two first abnormal state development trend vectors is used to characterize the target state development trend commonality value between the two first abnormal state development trend vectors; based on the commonality value of the past state development trend, adjusting the difference between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector of the candidate abnormal state event to obtain the distribution adjustment area of ​​the first abnormal state development trend vector of the target abnormal state event in the feature relationship network; determining the current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario based on the distribution adjustment area.

[0084] In the next step, the system conducts more in-depth data analysis and processing to determine the current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario.

[0085] First, the system projects each first abnormal state development trend vector into the feature relationship network of the target power operation and maintenance scenario. This feature relationship network is a multi-dimensional data structure that can capture and represent the complex relationship between different abnormal state events. Each abnormal state development trend vector will have a specific distribution area in this feature relationship network. For example, the feature relationship network is a three-dimensional space, and each dimension represents a different power operation and maintenance feature, such as voltage fluctuation, current intensity, and temperature change. Each first abnormal state development trend vector will occupy a specific position in this three-dimensional space according to the values ​​of these features.

[0086] Next, the system calculates the difference between these first abnormal state development trend vectors. Difference is a quantitative indicator used to measure the similarity or difference between two vectors. In this context, the difference is used to characterize the target state development trend commonality value between the two first abnormal state development trend vectors. If the difference between the two vectors is small, it means that their development trends in the target power operation and maintenance scenario are more similar. For a more intuitive understanding, the positions of the two vectors A and B in the three-dimensional feature relationship network can be set. If vectors A and B are very close, their difference is small, which means that their development trend commonality value is high.

[0087] Then, the system uses the commonality value of the past state development trend to adjust the difference between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector of the candidate abnormal state event. This process can be understood as the system correcting the current development trend forecast based on historical information. For example, if past data shows that the target abnormal state event and a candidate abnormal state event have a high commonality of development trends in history, then the system may reduce their difference in the current forecast.

[0088] Finally, the system determines the distribution adjustment area of ​​the first abnormal state development trend vector of the target abnormal state event in the feature relationship network based on the adjusted difference. This adjustment area reflects the system's latest prediction of the development trend of the target abnormal state event after comprehensively considering current and past information. Based on this distribution adjustment area, the system finally determines the current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario. This vector is the system's accurate prediction of the future development trend of the target abnormal state event, which combines the feature information of the current scenario with past development trend data.

[0089] Through this series of complex calculations and adjustment processes, the system can generate more accurate and reliable forecasts of abnormal state development trends. This not only helps power operation and maintenance personnel to understand potential risks and problems in a timely manner, but also provides them with a scientific basis for decision-making, thereby improving the stability and safety of the power system. It can be seen that by comprehensively considering the characteristic information and historical data of multiple dimensions, the development trend of abnormal state events in the power system can be accurately predicted. This prediction capability is of great significance for preventing potential power failures, optimizing operation and maintenance strategies, and improving the overall performance of the power system.

[0090] In another example, the difference between two first abnormal state development trend vectors has a first quantitative relationship with the target state development trend commonality value between the two first abnormal state development trend vectors; then, based on the past state development trend commonality value, the difference between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector of the candidate abnormal state event is adjusted, including: if the past state development trend commonality value meets the commonality index, limiting the difference between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector of the candidate abnormal state event; if the past state development trend commonality value meets the mutually exclusive index, expanding the difference between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector of the candidate abnormal state event; the commonality index and the mutually exclusive index correspond to different power operation and maintenance scenarios.

[0091] For example, the system can further clarify the relationship between the difference between two first abnormal state development trend vectors and the target state development trend commonality value between them. This relationship is quantified as a first quantitative relationship, which determines how to adjust the current difference based on the past state development trend commonality value.

[0092] In some application scenarios, there are two first abnormal state development trend vectors V1 and V2, which respectively represent the development trends of two different abnormal state events in the target power operation and maintenance scenario. These two vectors can be represented as points in a multidimensional space, and each dimension corresponds to a specific power operation and maintenance feature. For example, V1 = (0.5, 0.3, 0.2) and V2 = (0.4, 0.4, 0.2) can be two three-dimensional feature vectors, where each dimension represents characteristics such as voltage stability, current fluctuation, and temperature change.

[0093] The difference is a value that measures the similarity between two vectors, which can be obtained by calculating the Euclidean distance between the two vectors. In this example, the difference can be a specific value, such as 0.1, which represents the distance between V1 and V2 in the feature space.

[0094] The commonality value of the target state development trend is calculated based on the difference, which reflects the similarity of the development trends of the two abnormal state events. There is a first quantitative relationship between this commonality value and the difference, that is, the commonality value can be an inverse function of the difference. The smaller the difference, the greater the commonality value.

[0095] When the system considers the commonality value of the past state development trend, it will adjust the current difference based on this commonality value. If the commonality value of the past state development trend meets the commonality index (that is, the development trends of the two abnormal state events are similar in history), the system limits the difference between V1 and V2. For example, if the original difference is 0.1, the system may adjust it to 0.08, indicating that in the current power operation and maintenance scenario, the development trends of the two abnormal state events are more similar. Conversely, if the commonality value of the past state development trend meets the mutually exclusive index (that is, the development trends of the two abnormal state events are opposite or unrelated in history), the system amplifies the difference between V1 and V2. For example, the original difference increases from 0.1 to 0.12, indicating that in the current scenario, the development trends of the two events are more different. Importantly, the power operation and maintenance scenarios corresponding to the commonality index and the mutually exclusive index are different. This means that the system dynamically adjusts the difference and commonality value according to different historical development trends and current operation and maintenance scenarios. Through this dynamic adjustment mechanism, the system can more accurately capture and predict the development trend of abnormal state events in specific power operation and maintenance scenarios. It not only improves the sensitivity and accuracy of the early warning system, but also provides more accurate and timely decision-making support for operation and maintenance personnel.

[0096] It can be seen that by dynamically adjusting the difference and commonality values, a fine prediction of the development trend of abnormal state events is achieved. This prediction method combines historical data and current scene characteristics, effectively improving the stability and security of the power system and bringing significant benefits to power operation and maintenance.

[0097] In some other preferred embodiments, the number of candidate abnormal state events is at least two; the target state development trend commonality value of the target abnormal state event and the candidate abnormal state event in the target power operation and maintenance scenario is determined based on the first abnormal state development trend vector corresponding to each of the preceding abnormal state events, including: respectively determining the joint commonality score between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector corresponding to each of the candidate abnormal state events. Then, the difference between each of the first abnormal state development trend vectors is determined based on the distribution area of ​​each of the first abnormal state development trend vectors projected into the characteristic relationship network of the target power operation and maintenance scenario, including: determining the difference between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector corresponding to each of the candidate abnormal state events based on the distribution area of ​​each of the first abnormal state development trend vectors projected into the characteristic relationship network of the target power operation and maintenance scenario.

[0098] In some cases, the system processes more than one candidate abnormal state event, but at least two or more, which increases the complexity and accuracy of the analysis because the system needs to evaluate the correlation between the target abnormal state event and multiple candidate abnormal state events.

[0099] For example, the system will first determine the target abnormal state event and the target state development trend commonality value of each candidate abnormal state event in the target power operation and maintenance scenario based on the first abnormal state development trend vector corresponding to each previous abnormal state event. This process involves calculating an indicator called "joint commonality score". The joint commonality score is a quantitative value used to measure the similarity or correlation between the development trends of two abnormal state events.

[0100] For example, the target abnormal state event A has two candidate abnormal state events B and C. Events A, B, and C have their own first abnormal state development trend vectors V_A, V_B, and V_C, respectively. The system calculates the joint commonality score between V_A and V_B, and the joint commonality score between V_A and V_C. These scores reflect the degree of commonality between event A and events B and C in development trends, respectively.

[0101] Next, the system projects these first abnormal state development trend vectors into the feature relationship network of the target power operation and maintenance scenario. The feature relationship network is a multidimensional space, in which each dimension represents a specific power operation and maintenance feature (such as voltage fluctuation, power factor, etc.). In this space, the development trend vector of each abnormal state event will occupy a specific position, forming a distribution area.

[0102] The system then determines the difference between these vectors based on their distribution areas in the feature relationship network. Difference is a quantitative indicator used to measure the spatial distance or difference between two vectors. In this context, difference reflects the similarity or difference between the development trends of different abnormal state events.

[0103] Continuing with the above example, the system calculates the difference between V_A and V_B, and the difference between V_A and V_C. These difference values ​​provide detailed information about the similarity or difference in the development trend between event A and events B and C.

[0104] Finally, the system comprehensively evaluates the correlation between the target abnormal state event and the candidate abnormal state events based on these difference values ​​and joint commonality scores. This helps the system to more accurately predict and identify potential power operation and maintenance problems, so as to take corresponding preventive and response measures in advance.

[0105] In this way, by performing complex data analysis and comparison among multiple candidate abnormal state events, the system can more comprehensively understand the development trends and interrelationships of various abnormal state events in the target power operation and maintenance scenario. This not only improves the efficiency and safety of power operation and maintenance, but also provides more powerful decision-making support for operation and maintenance personnel. At the same time, the method of this preferred embodiment also enhances the flexibility and adaptability of the system, enabling it to cope with more complex and changeable power operation and maintenance scenarios.

[0106] In some other possible embodiments, the number of the candidate abnormal state events is at least two; then the method further includes: obtaining the abnormal state event knowledge vector corresponding to each preceding abnormal state event, and determining from each of the candidate abnormal state events a to-be-processed preceding abnormal state event that meets the common abnormal state event indicators with the target abnormal state event.

[0107] Based on this, the method adjusts the difference between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector of the candidate abnormal state event according to the commonality value of the past state development trends, and obtains the distribution adjustment area of ​​the first abnormal state development trend vector of the target abnormal state event in the feature relationship network, including: adjusting the difference between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector of the candidate abnormal state event according to the commonality value of the previous state development trends of the target abnormal state event and the previous abnormal state event to be processed, and obtaining the distribution adjustment area of ​​the first abnormal state development trend vector of the target abnormal state event in the feature relationship network.

[0108] When the number of candidate abnormal state events is at least two, the system's processing flow will be further refined. The key to this process is to use the knowledge vector of the abnormal state event to determine the previous abnormal state event to be processed that has commonalities with the target abnormal state event.

[0109] First, the system obtains the abnormal state event knowledge vector corresponding to each previous abnormal state event. These knowledge vectors are constructed based on historical data and expert knowledge, and they describe the key features and attributes of abnormal state events. For example, an abnormal state event knowledge vector about voltage anomaly can include characteristic values ​​such as voltage fluctuation range, duration, and occurrence frequency.

[0110] Next, the system determines the previous abnormal state events to be processed from multiple candidate abnormal state events that meet the commonality index of the target abnormal state event. Here, the "commonality index" is a preset standard used to measure the similarity between two abnormal state events. The system compares the knowledge vectors of the target abnormal state event and the candidate abnormal state events to screen out the previous events that have significant commonality with the target event.

[0111] For example, if the target abnormal state event is an event about voltage sag, the system compares the knowledge vectors of this event with all candidate abnormal state events. If a candidate event is also a voltage-related problem and its occurrence time, location or cause is similar to the target event, then this candidate event may be determined as a previous abnormal state event to be processed.

[0112] After determining the previous abnormal state events to be processed, the system adjusts the difference between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector of the candidate abnormal state event based on the commonality value of the previous state development trend between these events and the target abnormal state event. The "previous state development trend commonality value" here reflects the similarity in development trend between the target event and the previous event to be processed.

[0113] Specifically, the system calculates the difference between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector of each candidate abnormal state event. Then, these differences are adjusted according to the commonality value of the previous state development trend. If the commonality value is high, it means that the development trend of the target event and a candidate event is very similar, then the system may reduce the difference between them; conversely, if the commonality value is low, the system may increase the difference.

[0114] Finally, based on the adjusted difference, the system can obtain the distribution adjustment area of ​​the first abnormal state development trend vector of the target abnormal state event in the feature relationship network. This adjustment area more accurately reflects the actual development trend of the target abnormal state event in the current power operation and maintenance scenario.

[0115] It can be seen that by introducing the knowledge vector of abnormal state events and the commonality value of the development trend of the previous state, the system can more accurately identify and predict the previous events that have commonalities with the target abnormal state events, and adjust the prediction results of the event development trend accordingly. This not only improves the accuracy of the prediction, but also provides more targeted decision-making support for power operation and maintenance personnel.

[0116] In another embodiment, the step of obtaining the current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario includes: based on the target deep learning algorithm, mining the common values ​​of past state development trends between each of the second abnormal state development trend vectors, to obtain the current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario.

[0117] Furthermore, the step of debugging to obtain the target deep learning algorithm includes: according to the global debugging error determined by the target power operation and maintenance scenario identification error and the consistency analysis error, using the target power operation and maintenance scenario debugging example set to debug the basic deep learning algorithm to obtain a target deep learning algorithm used to determine the current abnormal state development trend vector in the target power operation and maintenance scenario; the target power operation and maintenance scenario debugging example set includes: the abnormal state event knowledge vector of each abnormal state event sample in the target power operation and maintenance scenario, the data knowledge of each smart grid operation data, and each of the abnormal state event samples in response to the existing The target power operation and maintenance scenario interaction information of the fault operation and maintenance label, and the second abnormal state development trend vector corresponding to each of the abnormal state event samples in the past power operation and maintenance scenarios; the error parameter of the target power operation and maintenance scenario identification error is the abnormal state event knowledge vector of each of the abnormal state event samples in the target power operation and maintenance scenario, and the data knowledge of each of the smart grid operation data; the error parameter of the consistency analysis error is the target state development trend commonality value of each of the abnormal state event samples in the target power operation and maintenance scenario, and the past state development trend commonality value of each of the abnormal state event samples in the past power operation and maintenance scenarios.

[0118] In another embodiment, in order to obtain the current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario, the system adopts a method based on deep learning. The key step involved in this process is to mine the commonality value of the past state development trend between each second abnormal state development trend vector.

[0119] First of all, it should be clear that the second abnormal state development trend vector represents the development trend of abnormal state events in past power operation and maintenance scenarios. By collecting and analyzing these vectors, the system can mine the common values ​​between them, that is, the similarities in the development trends of these abnormal state events.

[0120] In order to accurately mine these common values ​​and determine the current abnormal state development trend vector, the system uses a targeted deep learning algorithm that is carefully debugged to ensure that it can accurately identify the development trend of abnormal state events in the target power operation and maintenance scenario.

[0121] The process of debugging the target deep learning algorithm is complex and sophisticated. The process relies on a dataset called the "target power operation and maintenance scenario debugging example set", which contains information about multiple abnormal state event samples. This information includes abnormal state event knowledge vectors, data knowledge of smart grid operation data, interaction information related to existing fault operation and maintenance labels in the target power operation and maintenance scenario, and second abnormal state development trend vectors in past power operation and maintenance scenarios.

[0122] During the debugging process, the system calculates two main errors: target power operation and maintenance scenario identification error and consistency analysis error. The former focuses on whether the system can accurately identify abnormal state events and their related data, and its error parameters are abnormal state event knowledge vectors and data knowledge. The latter focuses on whether the system can consistently analyze the common values ​​of the development trend of abnormal state events in different power operation and maintenance scenarios, and its error parameters are the common values ​​of the target state development trend and the common values ​​of the past state development trend.

[0123] By continuously adjusting the parameters and structure of the deep learning algorithm to minimize these two errors, the system can eventually obtain an optimized target deep learning algorithm. This algorithm can accurately mine the common values ​​of past state development trends between the second abnormal state development trend vectors, and determine the current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario.

[0124] For example, the system has collected a series of second abnormal state development trend vectors of abnormal state event samples about voltage anomalies. Through the mining of the target deep learning algorithm, the system found that these vectors have significant commonalities in voltage fluctuation amplitude and duration. Based on these common values, the system can determine the current abnormal state development trend vector of voltage anomaly events in the current power operation and maintenance scenario, so as to more accurately predict and respond to possible voltage anomaly problems.

[0125] In this way, by introducing the target deep learning algorithm and the sophisticated debugging process, the system can more accurately identify and predict the development trend of abnormal state events in the target power operation and maintenance scenario. This not only improves the efficiency and safety of power operation and maintenance, but also provides more powerful decision-making support for operation and maintenance personnel. At the same time, this deep learning-based method also enhances the system's adaptability and scalability, enabling it to cope with more complex and changeable power operation and maintenance scenarios.

[0126] In some alternative embodiments, the step of determining the first abnormal state development trend vector of the target abnormal state event and the second abnormal state development trend vector of the previous abnormal state event includes: obtaining target power operation and maintenance scenario interaction information of the target abnormal state event in response to an existing fault operation and maintenance tag in the target power operation and maintenance scenario, and past power operation and maintenance scenario interaction information of the previous abnormal state event in response to an existing fault operation and maintenance tag in the past power operation and maintenance scenario; the feature granularity of the target power operation and maintenance scenario interaction information is smaller than the feature granularity of the past power operation and maintenance scenario interaction information; performing trend change mining on the target abnormal state event based on the target power operation and maintenance scenario interaction information to obtain the first abnormal state development trend vector of the target abnormal state event; performing trend change mining on the previous abnormal state event based on the past power operation and maintenance scenario interaction information to obtain the second abnormal state development trend vector of the previous abnormal state event.

[0127] The process of determining the first abnormal state development trend vector of the target abnormal state event and the second abnormal state development trend vector of the preceding abnormal state event in the above embodiment involves multiple key steps.

[0128] First, the system obtains the target power operation and maintenance scenario interaction information of the target abnormal state event in response to the target power operation and maintenance scenario with the fault operation and maintenance tag, and the previous abnormal state event in response to the past power operation and maintenance scenario with the fault operation and maintenance tag. This information reflects the interaction between the abnormal state event and the power operation and maintenance scenario.

[0129] Importantly, the feature granularity of the target power operation and maintenance scenario interaction information is smaller than that of the previous power operation and maintenance scenario interaction information. This means that the target power operation and maintenance scenario interaction information provides a more detailed and sophisticated feature description, allowing the system to more accurately capture and understand the development trend of abnormal state events.

[0130] Next, the system conducts trend change mining based on these interactive information. For the target abnormal state event, the system mines the interactive information of the target power operation and maintenance scenario to obtain the first abnormal state development trend vector of the target abnormal state event. This vector describes the development trend of the target abnormal state event in the target power operation and maintenance scenario.

[0131] Similarly, for the previous abnormal state event, the system mines the interactive information of past power operation and maintenance scenarios to obtain the second abnormal state development trend vector of the previous abnormal state event. This vector describes the development trend of the previous abnormal state event in past power operation and maintenance scenarios.

[0132] Taking a specific numerical example, the first abnormal state development trend vector of the target abnormal state event can be expressed as [0.8, 0.5, -0.3], where each value represents the development trend of the event in a specific dimension. Similarly, the second abnormal state development trend vector of the previous abnormal state event may be expressed as [0.6, -0.2, 0.4]. These values ​​are obtained by the system through complex algorithms and data mining techniques, and they reflect the changing trends of abnormal state events in different dimensions.

[0133] By acquiring more fine-grained interactive information and mining trend changes, the system can more accurately predict and understand the development trend of abnormal state events. This not only helps to timely discover and respond to potential power operation and maintenance problems, but also improves the stability and security of the power system.

[0134] In this way, by introducing more fine-grained interactive information and trend change mining technology, the system's ability to predict and understand the development trend of abnormal state events has been significantly improved. This improvement not only helps the stable operation of the power system, but also provides more accurate and timely decision support for operation and maintenance personnel, thereby improving overall operation and maintenance efficiency and safety.

[0135] In some other examples, the number of past power operation and maintenance scenarios is several; the step of determining the second abnormal state development trend vector of the previous abnormal state event includes: obtaining the abnormal state preceding trend vector of the previous abnormal state event in each of the past power operation and maintenance scenarios; integrating the abnormal state preceding trend vectors corresponding to the previous abnormal state event in each of the past power operation and maintenance scenarios to obtain the second abnormal state development trend vector of the previous abnormal state event.

[0136] In some other examples, considering that there may be more than one power operation and maintenance scenario in the past, but there are several different scenarios. In this context, determining the second abnormal state development trend vector of the previous abnormal state event requires a process of integrating multiple scenario information.

[0137] First, the system obtains the abnormal state preceding trend vectors of the preceding abnormal state events for each past power operation and maintenance scenario. These vectors reflect the development trend and characteristics of the preceding abnormal state events in a specific scenario.

[0138] Taking the numerical abnormal state preceding trend vector as an example, in three different past power operation and maintenance scenarios, the abnormal state preceding trend vectors of the preceding abnormal state events are [0.6, -0.4, 0.3], [-0.2, 0.5, -0.1], and [0.3, -0.2, 0.5] respectively. Each value in these vectors represents the trend change in different dimensions, which can be voltage fluctuation, current change or other key indicators related to power operation and maintenance.

[0139] Next, the system integrates these abnormal state preceding trend vectors obtained from different scenarios. The integration method can be simple averaging, weighted averaging, or more complex machine learning algorithms such as principal component analysis (PCA) or cluster analysis to extract common trends and features from these vectors.

[0140] Using a simple averaging method for integration, the average of the three abnormal state preceding trend vectors is the second abnormal state development trend vector, and the calculation result is [0.23, -0.03, 0.23] (the calculation in the embodiment of the present application is to average the values ​​in each dimension separately). This second abnormal state development trend vector represents the comprehensive development trend of the preceding abnormal state events in multiple past power operation and maintenance scenarios.

[0141] In this way, the system can make full use of information from multiple past power operation and maintenance scenarios to more comprehensively understand the development trend of previous abnormal state events. This not only improves the accuracy of abnormal state event predictions, but also provides a more reliable basis for decision-making for power operation and maintenance personnel. Furthermore, by integrating the abnormal state previous trend vectors in multiple past power operation and maintenance scenarios, the system can derive a more comprehensive and accurate second abnormal state development trend vector. This helps to improve the efficiency and safety of power operation and maintenance and reduce the potential risks caused by abnormal state events.

[0142] In addition, the innovative application of artificial intelligence technology in the above technical solutions is mainly reflected in the following aspects:

[0143] First, artificial intelligence technology is used to identify past power operation and maintenance scenarios that have commonalities with the target power operation and maintenance scenario, and to identify previous abnormal state events corresponding to the target abnormal state event. This involves pattern recognition and data analysis technology, which uses intelligent algorithms to mine a large amount of historical data to find past scenarios similar to the target scenario, as well as similar abnormal state events that have occurred in these scenarios;

[0144] Secondly, artificial intelligence technology is used to determine the development trend vector of abnormal state events. This includes using machine learning algorithms to train and learn historical data, so as to predict and describe the development trend of abnormal state events in specific power operation and maintenance scenarios. In this way, the system can generate a first abnormal state development trend vector and a second abnormal state development trend vector, which respectively represent the development trends of the target abnormal state event and the previous abnormal state event in different scenarios;

[0145] Next, artificial intelligence technology is used to integrate the two development trend vectors. This step usually involves complex mathematical calculations and data processing techniques, such as vector operations and data fusion. Through artificial intelligence technology, the system can automatically and accurately complete these calculation and processing tasks to obtain the current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario;

[0146] Finally, AI technology is also used to jointly analyze the current abnormal state development trend vector and the global state development trend vector of the smart grid operation data. The purpose of this step is to determine the correlation prediction results between the smart grid operation data and the target abnormal state event. Through advanced AI technologies such as deep learning and data mining, the system can reveal the potential correlations and patterns hidden in large amounts of data, thereby providing valuable prediction and decision support for power operation and maintenance personnel.

[0147] It can be seen that the above-mentioned technical solution, through the deep application of artificial intelligence technology, has achieved in-depth analysis of smart grid operation data and accurate prediction of abnormal status events, providing strong support for the intelligent and automated power operation and maintenance.

[0148] In combination with the above content, the embodiment of the present application also provides a global debugging error determined by the target power operation and maintenance scenario identification error and the consistency analysis error, and uses the target power operation and maintenance scenario debugging example set to debug the basic deep learning algorithm, and obtains a further implementation method of the target deep learning algorithm for determining the current abnormal state development trend vector in the target power operation and maintenance scenario as follows:

[0149] 1) Determine the global debugging error:

[0150] First, the recognition error of the target power operation and maintenance scenario is calculated, which can be done by comparing the actual label and the predicted label of the target power operation and maintenance scenario;

[0151] Next, the consistency analysis error is calculated, which involves evaluating whether the identification results for the same power operation and maintenance scenario are consistent at different time points or under different conditions;

[0152] The above two errors are combined to determine the global debugging error, which will serve as the basis for subsequent algorithm debugging.

[0153] 2) Build a set of debugging examples for target power operation and maintenance scenarios:

[0154] Collect and organize a series of sample data representing the target power operation and maintenance scenarios. These data should cover various possible abnormal status events and their development trends.

[0155] Preprocess the sample data, including data cleaning, feature extraction, and standardization, to ensure data quality and adapt to the needs of deep learning algorithms.

[0156] 3) Basic deep learning algorithm selection and initialization:

[0157] Choose a deep learning algorithm suitable for processing time series data and performing trend forecasting, such as the Long Short-Term Memory (LSTM) or Transformer model;

[0158] Initialize algorithm parameters, including learning rate, batch size, number of iterations, etc.

[0159] 4) Algorithm debugging process:

[0160] Use the target power operation and maintenance scenario debugging example set to train the basic deep learning algorithm;

[0161] During the training process, adjust algorithm parameters such as learning rate and regularization strength according to the global debugging error to optimize model performance;

[0162] Monitor the performance of the model on the validation set. When overfitting or underfitting occurs, adjust the model structure or introduce regularization technology in a timely manner.

[0163] Repeat the above steps until the performance of the model on the validation set reaches the preset standard or no longer improves significantly.

[0164] 5) Determine the target deep learning algorithm:

[0165] After multiple rounds of debugging and optimization, the deep learning model with the best performance is selected as the target deep learning algorithm;

[0166] The algorithm will be used to determine the current abnormal state development trend vector in the target power operation and maintenance scenario;

[0167] The target deep learning algorithm is encapsulated and saved for subsequent application in actual power operation and maintenance scenarios.

[0168] 6) Algorithm verification and deployment:

[0169] Verify the performance of the target deep learning algorithm on an independent test set;

[0170] If the performance meets the requirements, the algorithm will be deployed in the actual power operation and maintenance system to monitor and predict the development trend of abnormal status events in real time.

[0171] The above content can be summarized as follows: determine the global debugging error, which is determined by the identification error and consistency analysis error of the target power operation and maintenance scenario; construct a debugging example set for the target power operation and maintenance scenario, which contains example data of various abnormal state events and their development trends, and preprocess the data; select and initialize a deep learning algorithm suitable for processing time series data and performing trend prediction; use the debugging example set to train the deep learning algorithm, adjust the algorithm parameters according to the global debugging error during the training process to optimize the model performance, monitor the performance of the model on the validation set, and make timely adjustments to prevent overfitting or underfitting until the model performance reaches the preset standard; select the deep learning model with the best performance as the target deep learning algorithm, and encapsulate and save it; verify the performance of the target deep learning algorithm on the test set, and if the performance meets the requirements, deploy it to the actual power operation and maintenance system to monitor and predict the development trend of abnormal state events in real time.

[0172] In another independent embodiment, the current abnormal state development trend vector and the global state development trend vector of the smart grid operation data are jointly analyzed to determine the correlation prediction result between the smart grid operation data and the target abnormal state event, including:

[0173] Extract the current abnormal state development trend vector in the target power operation and maintenance scenario obtained by the deep learning algorithm; at the same time, obtain the global state development trend vector of the smart grid operation data, which reflects the change in the operation state of the entire power grid system;

[0174] Perform data normalization on the current abnormal state development trend vector and the global state development trend vector to ensure that the two are compared within the same numerical range; perform time synchronization on the two vectors to ensure that they correspond to the same time period or time point;

[0175] The preprocessed current abnormal state development trend vector and the global state development trend vector are compared and analyzed element by element, and the degree of association between the two vectors is quantified by calculating the similarity or correlation index (such as Pearson correlation coefficient, cosine similarity, etc.) between the two vectors; whether the change trends of the two vectors in the time series are consistent or there is an obvious correlation, for example, when a certain indicator in the global state development trend vector rises, observe whether the current abnormal state development trend vector also shows an upward trend; further use statistical models or machine learning algorithms, such as regression analysis, support vector machines, etc., to predict and verify the correlation between the two, and adjust model parameters according to the prediction results to improve prediction accuracy;

[0176] Based on the results of the above joint analysis, the correlation prediction results between the smart grid operation data and the target abnormal state event are determined; if there is a significant correlation between the two, this correlation can be used to predict abnormal state events that may occur in the future, or the development trend and impact range of the current abnormal state event can be evaluated based on the changes in the smart grid operation data; the correlation prediction results are displayed to the power operation and maintenance personnel in a visual form so that they can take timely response measures;

[0177] By comparing with actual abnormal state events, the accuracy of the correlation prediction results is verified; based on the verification results, the joint analysis method is optimized and adjusted, such as improving the similarity calculation method, introducing more advanced prediction models, etc., to improve the accuracy of future predictions.

[0178] In the operation and maintenance of smart grids, it is crucial to predict and respond to abnormal conditions in a timely manner. To achieve this goal, the system conducts in-depth analysis of the current abnormal state development trend and the global state development trend of the entire grid.

[0179] First, the system uses deep learning algorithms to extract the development trend vector of the current abnormal state for a specific power operation and maintenance scenario. This vector can be viewed as a series of values, each of which represents the development trend of the abnormal state in a specific aspect. For example, a four-dimensional current abnormal state development trend vector may be similar to [0.3, 0.5, -0.2, 0.1], where the values ​​represent the development trends of the abnormal state in four aspects: voltage fluctuation, current intensity, frequency stability, and temperature change.

[0180] At the same time, the system will also obtain the global state development trend vector of the smart grid operation data. This vector reflects the operation status of the entire power grid, including but not limited to the comprehensive change trends of multiple key indicators such as voltage, current, power factor, temperature, etc. The global state development trend vector is also a sequence of numerical values, such as [-0.1, 0.2, 0.0, 0.3], which represents the state changes of the global power grid in different aspects.

[0181] Before conducting joint analysis, the system normalizes the data of the two vectors to ensure that their value ranges are consistent, which is convenient for subsequent comparative analysis. At the same time, time synchronization is also an essential step to ensure that the time periods or time points corresponding to the two vectors are exactly the same.

[0182] Next, the system compares the two vectors element by element and calculates a similarity or correlation index between them. For example, the Pearson correlation coefficient can be used to measure the strength and direction of the linear relationship between the two vectors. If the calculated correlation coefficient is close to 1 or -1, it means that there is a strong positive or negative correlation between the two vectors.

[0183] In addition, the system will also observe the changing trends of the two vectors in time series. If an indicator in the global state development trend vector continues to rise, and the corresponding indicator in the current abnormal state development trend vector also shows an upward trend, then there may be some correlation between the two.

[0184] To further verify this association and predict future abnormal state events, the system may use statistical models such as regression analysis and support vector machines or machine learning algorithms. These algorithms can learn from historical data and predict future trends.

[0185] Based on the above analysis, the system can determine the correlation prediction results between the smart grid operation data and the target abnormal state events. For example, the system may find that when the voltage fluctuation of the global power grid increases, the abnormal state of the current intensity in a specific area will also increase. This correlation prediction result is very valuable for power operation and maintenance personnel because it can help them to warn in advance and take necessary measures to prevent potential failures.

[0186] Finally, the system displays these prediction results in an intuitive visual form, such as curves and heat maps, so that power operation and maintenance personnel can quickly understand and respond. At the same time, the system will verify the accuracy of the prediction results based on actual abnormal state events, and continuously optimize and adjust its analysis methods based on the verification results to improve the accuracy of future predictions.

[0187] In this way, the smart grid system can not only monitor the operation status of the power grid in real time, but also predict abnormal events that may occur in the future, thereby greatly improving the stability and security of the power grid. At the same time, this also provides a powerful tool for power operation and maintenance personnel, enabling them to manage and maintain the power grid system more efficiently.

[0188] Further, Figure 2 Schematic diagram of a smart grid analysis system 200 provided in an embodiment of the present application. Figure 2 The smart grid analysis system 200 shown includes a processor 210, which can call and run a computer program from a memory to implement the method in the embodiment of the present application.

[0189] Alternatively, if Figure 2 As shown, the smart grid analysis system 200 may further include a memory 230. The processor 210 may call and run a computer program from the memory 230 to implement the method in the embodiment of the present application.

[0190] The memory 230 may be a separate device independent of the processor 210 , or may be integrated into the processor 210 .

[0191] Alternatively, if Figure 2 As shown, the smart grid analysis system 200 may further include a transceiver 220, and the processor 210 may control the transceiver 220 to interact with other devices, specifically, may send information or data to other devices, or receive information or data sent by other devices.

[0192] Optionally, the smart grid analysis system 200 may implement the corresponding processes corresponding to the storage engine or components in the storage engine (such as a processing module) or a device deployed with a storage engine in each method of the embodiments of the present application, which will not be described here for the sake of brevity.

[0193] It should be understood that the processor of the embodiment of the present application may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method embodiment can be completed by the hardware integrated logic circuit or software instructions in the processor. The above processor can be a general processor, a digital signal processor (Digital Signal Processor, DSP), an application specific integrated circuit (Application Specific Integrated Circuit, ASIC), a field programmable gate array (Field Programmable Gate Array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present application can be implemented or executed. The general processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in the embodiment of the present application can be directly embodied as a hardware decoding processor to perform, or the hardware and software modules in the decoding processor are combined and performed. The software module can be located in a random access memory, a flash memory, a read-only memory, a programmable read-only memory or an electrically erasable programmable memory, a register, and other mature storage media in the art. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.

[0194] It can be understood that the memory in the embodiments of the present application can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory can be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct RAM bus random access memory (DRRAM). It should be noted that the memory of the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0195] It should be understood that the above-mentioned memory is exemplary but not restrictive. For example, the memory in the embodiments of the present application may also be static random access memory (static RAM, SRAM), dynamic random access memory (dynamic RAM, DRAM), synchronous dynamic random access memory (synchronous DRAM, SDRAM), double data rate synchronous dynamic random access memory (double data rate SDRAM, DDR SDRAM), enhanced synchronous dynamic random access memory (enhanced SDRAM, ESDRAM), synchronous link dynamic random access memory (synch link DRAM, SLDRAM) and direct memory bus random access memory (DirectRambus RAM, DR RAM), etc. That is to say, the memory in the embodiments of the present application is intended to include but not limited to these and any other suitable types of memory.

[0196] Based on the above, a computer-readable storage medium is provided, on which a computer program is stored, and the computer program implements the above method when running.

[0197] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations.

Claims

1. A data analysis method based on smart grid, characterized in that: Applied to a smart grid analysis system, the method comprises: According to the target power operation and maintenance scenario corresponding to the smart grid operation data to be analyzed, determine the past power operation and maintenance scenarios that meet the scenario commonality determination requirements with the target power operation and maintenance scenario, and determine the preceding abnormal state event corresponding to the target abnormal state event to be analyzed; the preceding abnormal state event and the target abnormal state event have at least partially the same fault risk state variables; Determine a first abnormal state development trend vector of the target abnormal state event and a second abnormal state development trend vector of the preceding abnormal state event; the first abnormal state development trend vector is used to characterize the trend change of the target abnormal state event in response to the existing fault operation and maintenance label in the target power operation and maintenance scenario; the second abnormal state development trend vector is used to characterize the trend change of the preceding abnormal state event in response to the existing fault operation and maintenance label in the past power operation and maintenance scenario; Performing a trend vector integration operation on the second abnormal state development trend vector and the first abnormal state development trend vector to obtain a current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario; Jointly analyzing the current abnormal state development trend vector and the global state development trend vector of the smart grid operation data to determine a correlation prediction result between the smart grid operation data and the target abnormal state event; The number of the preceding abnormal state events is at least two, and the target abnormal state event is one of the preceding abnormal state events; The method further includes: obtaining a first abnormal state development trend vector of a candidate abnormal state event other than the target abnormal state event in each of the preceding abnormal state events; the first abnormal state development trend vector of the candidate abnormal state event is used to characterize the trend change of the candidate abnormal state event in response to an existing fault operation and maintenance tag in the target power operation and maintenance scenario; The performing trend vector integration operation on the second abnormal state development trend vector and the first abnormal state development trend vector to obtain the current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario includes: determining the target state development trend commonality value of the target abnormal state event and the candidate abnormal state event in the target power operation and maintenance scenario based on the first abnormal state development trend vector corresponding to each of the preceding abnormal state events; determining the past state development trend commonality value of the target abnormal state event and the candidate abnormal state event in the past power operation and maintenance scenario based on the second abnormal state development trend vector corresponding to each of the preceding abnormal state events; mining the past state development trend commonality value based on the target state development trend commonality value to obtain the current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario; The method comprises mining the commonality values ​​of the past state development trends according to the commonality values ​​of the target state development trends to obtain the current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario, including determining the difference between each of the first abnormal state development trend vectors based on the distribution area projected onto the characteristic relationship network of the target power operation and maintenance scenario; the difference between two first abnormal state development trend vectors is used to characterize the commonality value of the target state development trend between the two first abnormal state development trend vectors; According to the commonality value of the past state development trend, adjusting the difference between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector of the candidate abnormal state event, and obtaining a distribution adjustment area of ​​the first abnormal state development trend vector of the target abnormal state event in the feature relationship network; A current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario is determined according to the distribution adjustment area.

2. The method according to claim 1, characterized in that The difference between the two first abnormal state development trend vectors has a first quantitative relationship with the target state development trend commonality value between the two first abnormal state development trend vectors; The step of adjusting the difference between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector of the candidate abnormal state event according to the commonality value of the past state development trend includes: If the commonality value of the past state development trend meets the commonality index, narrowing the difference between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector of the candidate abnormal state event; If the commonality value of the past state development trend meets the mutually exclusive index, the difference between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector of the candidate abnormal state event is amplified; the power operation and maintenance scenarios corresponding to the commonality index and the mutually exclusive index are different.

3. The method according to claim 1, characterized in that The number of the candidate abnormal state events is at least two; the determining of the target state development trend commonality value of the target abnormal state event and the candidate abnormal state event in the target power operation and maintenance scenario based on the first abnormal state development trend vector corresponding to each of the preceding abnormal state events includes: respectively determining the joint commonality score between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector corresponding to each of the candidate abnormal state events; The method of determining the degree of difference between each of the first abnormal state development trend vectors based on the distribution area projected onto the characteristic relationship network of the target power operation and maintenance scenario includes: determining the degree of difference between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector corresponding to each of the candidate abnormal state events based on the distribution area projected onto the characteristic relationship network of the target power operation and maintenance scenario.

4. The method according to claim 1, characterized in that: The number of candidate abnormal state events is at least two; the method further comprises: obtaining abnormal state event knowledge vectors corresponding to each preceding abnormal state event, and determining, from each of the candidate abnormal state events, a to-be-processed preceding abnormal state event that meets the commonality index of abnormal state events with the target abnormal state event; The method of adjusting the difference between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector of the candidate abnormal state event based on the commonality value of the past state development trends to obtain the distribution adjustment area of ​​the first abnormal state development trend vector of the target abnormal state event in the feature relationship network includes: adjusting the difference between the first abnormal state development trend vector of the target abnormal state event and the first abnormal state development trend vector of the candidate abnormal state event based on the commonality value of the previous state development trends of the target abnormal state event and the previous abnormal state event to be processed to obtain the distribution adjustment area of ​​the first abnormal state development trend vector of the target abnormal state event in the feature relationship network.

5. The method according to claim 1, characterized in that The step of obtaining the current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario includes: Based on the target deep learning algorithm, mining the commonality values ​​of the past state development trends between the second abnormal state development trend vectors, to obtain the current abnormal state development trend vector of the target abnormal state event in the target power operation and maintenance scenario; The steps of debugging and obtaining the target deep learning algorithm include: According to the global debugging error determined by the target power operation and maintenance scenario identification error and the consistency analysis error, the basic deep learning algorithm is debugged using the target power operation and maintenance scenario debugging example set to obtain a target deep learning algorithm for determining the current abnormal state development trend vector in the target power operation and maintenance scenario; the target power operation and maintenance scenario debugging example set includes: the abnormal state event knowledge vector of each abnormal state event sample in the target power operation and maintenance scenario, the data knowledge of each smart grid operation data, the target power operation and maintenance of each abnormal state event sample in response to the fault operation and maintenance label in the target power operation and maintenance scenario The scene interaction information, and the second abnormal state development trend vector corresponding to each of the abnormal state event samples in the past power operation and maintenance scenes; the error parameter of the target power operation and maintenance scene identification error is the abnormal state event knowledge vector of each of the abnormal state event samples in the target power operation and maintenance scene, and the data knowledge of each of the smart grid operation data; the error parameter of the consistency analysis error is the target state development trend commonality value of each of the abnormal state event samples in the target power operation and maintenance scene, and the past state development trend commonality value of each of the abnormal state event samples in the past power operation and maintenance scenes.

6. The method according to claim 1, characterized in that The step of determining a first abnormal state development trend vector of the target abnormal state event and a second abnormal state development trend vector of the preceding abnormal state event comprises: Obtain target power operation and maintenance scenario interaction information in which the target abnormal state event responds to an existing fault operation and maintenance tag in the target power operation and maintenance scenario, and past power operation and maintenance scenario interaction information in which the preceding abnormal state event responds to an existing fault operation and maintenance tag in the past power operation and maintenance scenario; the feature granularity of the target power operation and maintenance scenario interaction information is smaller than the feature granularity of the past power operation and maintenance scenario interaction information; Performing trend change mining on the target abnormal state event according to the target power operation and maintenance scenario interaction information to obtain a first abnormal state development trend vector of the target abnormal state event; The trend change of the preceding abnormal state event is mined according to the past electric power operation and maintenance scenario interaction information to obtain a second abnormal state development trend vector of the preceding abnormal state event.

7. The method according to claim 1, characterized in that The number of the past power operation and maintenance scenarios is several; the step of determining the second abnormal state development trend vector of the preceding abnormal state event comprises: Obtaining an abnormal state preceding trend vector of a preceding abnormal state event in each of the past power operation and maintenance scenarios; The abnormal state preceding trend vectors corresponding to the preceding abnormal state events in each of the past power operation and maintenance scenarios are integrated to obtain a second abnormal state development trend vector of the preceding abnormal state event.

8. A smart grid analysis system, characterized in that: The method comprises at least one processor and a memory; the memory stores computer-executable instructions; the at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the method according to any one of claims 1 to 7.

Images