Historical data reconstructed power grid digital twinning fault simulation equipment

Abstract

The invention provides power grid digital twinning fault simulation equipment for historical data reconstruction, and relates to the technical field of power digital twinning, and the equipment comprises the steps that a fault feature collection unit collects power grid fault feature information when a target power grid has a fault; a historical data retrieval unit performs historical power grid fault indexing according to the power grid fault feature information to obtain a same-family historical power grid fault and a historical power grid fault operation data set when the same-family historical power grid fault occurs; the historical data reconstruction unit obtains device maintenance data of a plurality of power grid devices after a same-family historical power grid fault, carries out historical data reconstruction generation, and obtains a reconstructed power grid fault operation data set; and the digital twinborn simulation unit constructs a power grid fault twinborn simulation result by adopting the reconstructed power grid fault operation data set. The technical problem that the accuracy of fault diagnosis is limited due to the fact that an existing power grid fault diagnosis method lacks effective utilization of historical data is solved, and the accuracy of fault diagnosis and the perspectiveness of fault prevention are remarkably improved.

Description

Technical Field

[0001] This application relates to the technical field of power digital twins, and specifically to a power grid digital twin fault simulation device for historical data reconstruction. Background Art

[0002] Power grid fault diagnosis is an important link in the operation of power systems, which is directly related to the safety, stability, and power supply reliability of the power grid. With the expansion of the power grid scale and the increase in structural complexity, the types and occurrence conditions of power grid faults have become more diverse and complex, and accurate and rapid fault diagnosis has become the focus of power system research.

[0003] Currently, power grid fault diagnosis mainly relies on the analysis of real-time monitoring data and equipment status. Usually, sensors and monitoring devices are used to collect power grid operation data in real time to monitor the working status of power grid equipment, and fault detection algorithms, models, and rules are used for fault location. However, there are some obvious deficiencies in the existing fault diagnosis methods. First, most fault diagnoses rely on real-time collected data, which makes it possible that real-time data may not accurately capture the full picture of a fault when certain fault types (such as transient faults or hidden faults) occur. Second, there are a wide variety of power grid equipment, and the fault characteristics are complex. A single monitoring method cannot fully cover all fault modes. Therefore, the existing methods often have problems such as incomplete fault feature recognition, insufficient fault simulation, and a relatively high misdiagnosis rate. Summary of the Invention

[0004] This application provides a power grid digital twin fault simulation device for historical data reconstruction, which solves the technical problem that the existing power grid fault diagnosis methods lack the effective use of historical data, resulting in limited accuracy of fault diagnosis, and achieves the technical effect of significantly improving the accuracy of fault diagnosis and the forward-looking of fault prevention.

[0005] In view of the above problems, the present application provides a power grid digital twin fault simulation device for historical data reconstruction, and the device includes: a fault feature acquisition unit, which is used to acquire power grid fault feature information when a fault is detected in the target power grid; a historical data retrieval unit, which is used to perform historical power grid fault indexing according to the power grid fault feature information to obtain homologous historical power grid faults, and index the historical power grid fault operation data set when the homologous historical power grid faults occur, wherein the historical power grid fault operation data set includes multiple historical fault operation data of multiple power grid devices in the target power grid; a historical data reconstruction unit, which is used to obtain multiple device maintenance data of the multiple power grid devices after the homologous historical power grid faults in the target power grid, and perform historical data reconstruction on the historical power grid fault operation data set to generate a reconstructed power grid fault operation data set; a digital twin simulation unit, which is used to adopt the reconstructed power grid fault operation data set and construct a power grid fault twin simulation result based on digital twin.

[0006] One or more technical solutions provided in the present application have at least the following technical effects or advantages: When the fault feature acquisition unit detects a fault in the target power grid, it acquires power grid fault feature information. The historical data retrieval unit performs historical power grid fault indexing according to the power grid fault feature information to obtain homologous historical power grid faults, and indexes the historical power grid fault operation data set when the homologous historical power grid faults occur. By combining the real-time feature data when the power grid fault occurs with the historical data through the fault feature acquisition unit and the historical data retrieval unit, and adopting the method of feature-based indexing to retrieve the historical power grid fault data set, the fault features can be understood more comprehensively and accurately by combining similar fault historical data, providing richer data support for subsequent simulations.

[0007] The historical data reconstruction unit obtains multiple device maintenance data of the multiple power grid devices after the homologous historical power grid faults in the target power grid, and performs historical data reconstruction on the historical power grid fault operation data set to generate a reconstructed power grid fault operation data set. By reconstructing the historical fault operation data through the historical data reconstruction unit, a predicted operation data set under the current power grid fault is generated, supplementing the missing data and improving the comprehensiveness and accuracy of the fault simulation.

[0008] The digital twin simulation unit adopts the reconstructed power grid fault operation data set and constructs a power grid fault twin simulation result based on digital twin, simulating a more accurate power grid fault situation, providing real-time simulation results for power grid operation and maintenance personnel, and assisting in decision-making.

[0009] In summary, by integrating the above functional units, this application combines historical fault data reconstruction with digital twin technology to achieve systematic analysis and simulation of power grid faults, enabling prediction of power grid behavior under different fault conditions, significantly improving the accuracy of fault diagnosis and the foresight of fault prevention. The application of digital twin technology allows testing and optimization of faults in a virtual environment, reducing risks in actual operations and providing a scientific basis for maintenance decisions, thereby improving power grid operation efficiency and overall stability.

[0010] The above description is only an overview of the technical solution of this application. In order to understand the technical means of this application more clearly, it can be implemented according to the content of the specification. In order to make the above and other purposes, features and advantages of this application more obvious and understandable, the following specific embodiments of this application are specifically given. Brief Description of the Drawings

[0011] Figure 1 It is a schematic structural diagram of a power grid digital twin fault simulation device for historical data reconstruction provided by an embodiment of this application.

[0012] Figure 2 It is a schematic flow diagram of obtaining a historical power grid fault operation data set in a power grid digital twin fault simulation device for historical data reconstruction provided by an embodiment of this application.

[0013] Figure 3 It is a schematic flow diagram of generating historical data reconstruction for a historical power grid fault operation data set in a power grid digital twin fault simulation device for historical data reconstruction provided by an embodiment of this application.

[0014] Description of the reference numerals: Fault feature acquisition unit 10, historical data retrieval unit 20, historical data reconstruction unit 30, digital twin simulation unit 40. Detailed Description of the Embodiment

[0015] An embodiment of this application provides a power grid digital twin fault simulation device for historical data reconstruction, which uses historical data reconstruction and digital twin technology to simulate power grid faults, solves the technical problem that the existing power grid fault diagnosis method lacks effective utilization of historical data, resulting in limited accuracy of fault diagnosis, and achieves the technical effect of significantly improving the accuracy of fault diagnosis and the foresight of fault prevention.

[0016] As Figure 1 shown, an embodiment of this application provides a power grid digital twin fault simulation device for historical data reconstruction, and the device includes: A fault feature acquisition unit 10, which is used to acquire power grid fault feature information when a fault is detected in the target power grid.

[0017] Specifically, the target power grid refers to the power system or network that is being monitored and has a fault, which can be a specific distribution network, substation, or a power transmission network in a broad sense. Grid fault feature information refers to the key feature data when the grid has a fault, including the fault location and fault parameters.

[0018] During the operation of the power grid, the monitoring devices continuously collect the operation parameters of the power grid. For example, the current and voltage sensors and watt-hour meters installed in the substation monitor the load changes and power quality of the power grid in real time; the temperature and vibration sensors deployed on the transmission line detect the early signs of overheating or mechanical damage of the line. When the target power grid has a fault, the fault feature acquisition unit 10 connects to the acquisition devices such as sensors, smart meters, and protection devices in the power grid to obtain the feature data related to the fault in real time, including the fault location and fault parameters. For example, the faulty device and the operation parameters such as current, voltage, power, and frequency. These feature data can reveal the specific situation of the power grid fault and provide reliable data support for subsequent fault location and simulation.

[0019] A historical data retrieval unit 20, which is used to perform a historical power grid fault index according to the grid fault feature information to obtain homologous historical power grid faults, and index to obtain the historical power grid fault operation data set when the homologous historical power grid faults occur, where the historical power grid fault operation data set includes multiple historical fault operation data of multiple power grid devices in the target power grid.

[0020] Specifically, homologous historical power grid faults refer to historical faults that are similar to or belong to the same category as the current power grid fault features. The historical power grid fault operation data set is a set containing multiple historical power grid fault operation data. The content in the data set includes information such as the working status, performance parameters, and response measures of multiple power grid devices during the fault. Among them, power grid devices refer to various equipment and devices in the power grid system, such as transformers, switches, circuit breakers, distribution lines, protection equipment, etc.

[0021] The historical data retrieval unit 20 receives the power grid fault feature information collected by the fault feature acquisition unit 10, processes these fault features, and extracts key data points such as current and voltage fluctuations. Then, according to the current power grid fault features, it compares with the historical fault data stored in the power grid database to find historical fault cases with higher similarity and marks them as the same group of historical power grid faults. During the query, it will match according to multiple dimensions in the fault feature information, such as fault type, occurrence location, and the then power grid load, etc., to ensure that the found historical power grid faults of the same family are highly similar to the current power grid fault. Then, it extracts and summarizes the operating states and behaviors of each power grid device during the fault when the historical power grid faults of the same family occur, and generates a historical power grid fault operation data set. These data sets can be information such as the temperature change of the transformer, the operation records of the circuit breaker, the operating current, voltage, power, and the fluctuation range of these parameters.

[0022] Exemplarily, in the power grid system of a city, currently a short - circuit fault is detected in the distribution network of a certain area, and the fault feature captured by the fault feature acquisition unit 10 is that the current of a specific line suddenly becomes too large. The historical data retrieval unit 20 will search the database for historical fault records of the same situation where the current of the same line has suddenly become too large before. If there is a historical fault record in the database that the line short - circuited due to a tree branch touching it before, then this belongs to the historical power grid faults of the same family. Then, the historical fault operation data of devices such as the transformer and distribution box on this line at that time, such as the oil temperature of the transformer and the voltage fluctuation of the distribution box at that time, are indexed as the historical power grid fault operation data set.

[0023] The historical data retrieval unit 20 can efficiently retrieve a historical fault data set similar to the current fault from the historical fault database according to the real - time collected fault feature information, providing data support for subsequent fault simulation and diagnosis. Especially in the face of complex fault scenarios, through the comparison and analysis of historical data, it can help maintenance personnel better understand the causes and consequences of the fault, and thus make more accurate countermeasures.

[0024] The historical data reconstruction unit 30, the historical data reconstruction unit 30 is used to obtain multiple device maintenance data of the multiple power grid devices after the historical power grid faults of the same family for the target power grid, and generate the reconstructed power grid fault operation data set by performing historical data reconstruction on the historical power grid fault operation data set.

[0025] Specifically, the power grid device maintenance data refers to the data recorded during the maintenance operations of multiple power grid devices in the target power grid after they have experienced homologous historical power grid faults. For example, the maintenance content recorded during the repair of a faulty transformer, the information of replaced components, the detection data after repair, etc. These data can reflect the repair situation of the device after the fault and its state after repair. The reconstructed power grid fault operation dataset is a power grid operation dataset at the time of the current fault reconstructed by combining the device maintenance data with the original historical power grid fault operation dataset.

[0026] First, obtain multiple device maintenance data of multiple power grid devices in the target power grid after homologous historical power grid faults from relevant data sources. These data sources can be databases of power grid equipment maintenance management systems, etc. By analyzing the multiple device maintenance data of the power grid devices, the true performance of the equipment at the time of the fault can be inferred. For example, if a certain transformer had just been overhauled before the fault occurred and the maintenance record shows that its health status was good, then this equipment may show a faster recovery ability in the reconstructed data.

[0027] The historical data reconstruction unit 30 infers and supplements the data that cannot be directly measured or is lost based on the known historical fault data and device maintenance data. These reconstructed data may include the actual performance of the equipment (such as whether the repaired equipment can fully return to the normal working state), the response time at the time of the fault, the changes in current and voltage during the recovery process, etc., to generate a reconstructed power grid fault operation dataset. This reconstructed power grid fault operation dataset can more comprehensively reflect the state changes of power grid devices during the entire process from the occurrence to the repair of the current power grid fault in the target power grid, providing a high-quality data source for subsequent digital twin simulations, so as to more accurately simulate the performance and recovery process of the current power grid fault and help power grid operation and maintenance personnel formulate more scientific emergency response strategies.

[0028] The digital twin simulation unit 40 is used to adopt the reconstructed power grid fault operation dataset to construct a power grid fault twin simulation result based on digital twin.

[0029] Specifically, the digital twin simulation unit 40 uses digital twin technology to construct a virtual power grid model based on information such as the topological structure and device parameters of the power grid. This model includes not only the physical structure of the power grid (such as power transmission lines, substation locations, transformers, etc.), but also takes into account the dynamic performance, operating state, and fault modes of various devices. For example, factors such as the operating characteristics of transformers and the trigger delay of circuit breakers will be incorporated into the model.

[0030] Obtain the reconstructed power grid fault operation dataset generated by the historical data reconstruction unit 30. These datasets include information such as the behavior data, performance parameters, response time, etc. of each device (such as transformers, circuit breakers, cables, etc.) in the target power grid under the current fault condition. Then, using the characteristics of digital twins, namely two-way data interaction and real-time update, input the reconstructed power grid fault operation dataset into the power grid digital model to simulate the response process of the target power grid after the current fault occurs, such as current fluctuations, voltage drops, equipment switching, etc. Finally, construct the power grid fault twin simulation result. This result is a simulation result of the power grid fault constructed through digital twin technology based on the reconstructed power grid fault operation dataset, which can intuitively display various characteristics of the current power grid fault, including various states when the target power grid fault occurs, such as the location of the fault point, the impact of the fault on the power grid voltage and current distribution, and the path of fault propagation, etc.

[0031] By combining real-time data with historical fault data, the digital twin model can accurately reflect the dynamic behavior and response process of the power grid when a fault occurs. For example, by simulating a certain overload fault, the digital twin model can display the key nodes during the power grid recovery process, such as equipment recovery time, load adjustment, etc. The simulation results can help power grid operation and maintenance personnel understand the timing and consequences of the fault occurrence, and can also predict the power grid recovery process, so as to optimize the operation and maintenance strategies of the power grid.

[0032] Furthermore, the fault feature acquisition unit 10 in the embodiment of the present application is further used to execute the following steps: When a fault is detected in the target power grid, collect the fault location information and fault parameters; integrate the fault location information and fault parameters to obtain the power grid fault feature information.

[0033] Specifically, the fault feature acquisition unit 10 obtains the complete power grid fault feature information through two key data points, namely the fault location information and the fault parameters. When a fault occurs in the target power grid, sensors and monitoring devices in the power grid will detect the fault phenomenon in real time. These monitoring devices include current sensors, voltage sensors, fault indicators, etc., and judge the occurrence of the fault by detecting abnormal changes in parameters such as current and voltage.

[0034] Fault location information is usually obtained through a sensor network or a fault location device. In various links of the power grid (such as transformers, circuit breakers, transmission lines, etc.), precise positioning devices are installed, which can upload the area information where the fault occurs in real time through a communication network. For example, when a short circuit occurs, a certain circuit breaker will immediately send a signal to the system, indicating the specific power grid section or location where the fault occurs. When a fault occurs, the sensor network in the power grid will capture the changes in the power grid operation parameters, monitor the operation parameters such as current, voltage, and power in real time, and record the relevant data. For example, a sudden increase in current and a sudden drop in voltage caused by a short circuit fault; continuous high current caused by equipment overload, but there will be no violent fluctuations like a short circuit.

[0035] The fault feature acquisition unit 10 interacts with sensors and monitoring devices in the target power grid, obtains the fault location information and fault parameters when a fault occurs, and uses a data fusion algorithm to integrate the fault location information and fault parameters to generate power grid fault feature information, providing a basis for subsequent fault analysis and simulation.

[0036] Furthermore, as Figure 2 shown, the historical data retrieval unit 20 of the embodiment of the present application is further used to perform the following steps: Step P21: Obtain the historical fault record data of the target power grid within a preset historical time range.

[0037] Step P22: Based on the power grid fault feature information, traverse and index in the historical fault record data to obtain the historical power grid fault feature information with the greatest similarity, as the homologous historical power grid fault.

[0038] Step P23: Retrieve the multiple historical fault operation data of multiple power grid devices in the target power grid when the homologous historical power grid fault occurs to obtain the historical power grid fault operation data set.

[0039] Specifically, the historical data retrieval unit 20 obtains the historical fault record data of the target power grid from the database of the power grid according to the preset historical time range. These data may include information such as the time, location, equipment, fault type, duration, and affected power grid components of the fault occurrence. These historical records provide an important reference for analyzing the current fault.

[0040] Then, using the currently collected power grid fault feature information, traverse and index in the historical fault record data. By using a similarity matching algorithm such as Euclidean distance and cosine similarity during the traversal process, calculate the similarity between the historical fault feature and the current fault feature to quantitatively evaluate the similarity degree between different historical faults, and find the historical fault most similar to the current fault feature as the homologous historical power grid fault.

[0041] After determining the homologous historical grid faults, retrieve the historical fault operation data of relevant grid devices in the target grid when these faults occurred, including data such as current, voltage, temperature, switch status, and repair duration of each grid device, to generate a historical grid fault operation dataset, providing rich reference data for the simulation and analysis of the current fault and improving the accuracy of fault simulation.

[0042] Further, the historical data retrieval unit 20 of the embodiment of the present application is further configured to perform the following steps: Step P221: Based on the grid fault feature information, traverse and index in the historical fault record data, calculate the fault location similarity between the fault location information in the grid fault feature information and multiple historical fault location information, and calculate the fault parameter similarity between the fault parameters in the grid fault feature information and multiple historical fault parameters.

[0043] Step P222: Perform weighted calculation on the fault location similarity and fault parameter similarity of each historical grid fault feature information to obtain multiple historical grid fault similarities.

[0044] Step P223: Select the historical grid fault feature information with the largest historical grid fault similarity and output to obtain the homologous historical grid faults.

[0045] Specifically, the historical data retrieval unit 20 determines the homologous historical grid faults by calculating the fault location similarity and fault parameter similarity between the current grid fault feature information and the historical fault record data.

[0046] The historical data retrieval unit 20 first finds the location information of all faults in the historical fault record data according to the fault location information of the current grid fault, including the grid device, location, and area where the fault occurred. Using methods such as Euclidean distance or cosine similarity, compare the location information of the current fault with multiple locations in the historical fault record to calculate the similarity of the fault location. Exemplarily, the current fault location coordinates are (x1, y1), and the historical fault location coordinates are (x2, y2). Use the Euclidean distance formula , and calculate the similarity between the two according to their relative positions.

[0047] In addition to the location information, this unit also calculates the similarity between the electrical parameters (such as current and voltage fluctuations) of the current fault and the electrical parameters in the historical record. Usually, parameters such as current and voltage are described using statistical methods such as standard deviation, peak value, or mean value, and then calculate between the current fault parameters and the historical fault parameters. Taking voltage as an example, the current fault voltage is V1, and the historical fault voltage is V2. A simple proportional difference method can be used to calculate the similarity, , assuming For multiple fault parameters, their similarities can be calculated separately and then comprehensively considered.

[0048] After obtaining the location similarity and parameter similarity of each historical fault, according to the different importance of fault characteristics, these two similarities are weighted and calculated to obtain the comprehensive similarity of each historical power grid fault. For example, the weight of the fault location similarity is 0.4, the weight of the fault parameter similarity is 0.6, the fault location similarity of a certain historical fault is 0.8, and the fault parameter similarity is 0.7, then its historical power grid fault similarity = 0.4 × 0.8 + 0.6 × 0.7 = 0.74.

[0049] After calculating the historical power grid fault similarities corresponding to all historical power grid fault feature information, select the historical fault corresponding to the historical power grid fault feature information with the largest similarity value as the homologous historical power grid fault. Because the homologous historical power grid faults are the most similar to the current fault in terms of characteristics, their related processing experiences and data have high reference value for the current power grid fault.

[0050] Further, as Figure 3 shown, the historical data reconstruction unit 30 of the embodiment of the present application is further configured to perform the following steps: Step P31: Obtain the maintenance data records of the multiple power grid devices in the target power grid.

[0051] Step P32: In the maintenance data records, retrieve the multiple device maintenance data of the multiple power grid devices within the time after the homologous historical power grid fault.

[0052] Step P33: Generate historical data reconstruction for the historical power grid fault operation dataset according to the multiple device maintenance data.

[0053] Specifically, the maintenance data record refers to the relevant data such as the maintenance, overhaul, and maintenance of power grid equipment. These records include the overhaul time, fault history, maintenance content, equipment status, etc. of the equipment, and can provide detailed information for the health status of the equipment and the repair process after a fault occurs. The maintenance data records are usually stored in the equipment management database of the power grid management system, and the historical data reconstruction unit 30 calls the maintenance data records of each power grid device in the target power grid through the interface, including relevant information such as the regular inspection, repair, and fault repair of each device.

[0054] After obtaining the complete maintenance data record, the historical data reconstruction unit 30 retrieves the grid device maintenance data within a certain period of time after the fault according to the time range of the homologous historical grid faults. These data are closely related to the previously occurred homologous historical grid faults and may contain important information such as the maintenance measures taken for such faults and the state changes of the devices after repair. For example, if the homologous historical grid fault is a short - circuit fault in a certain substation, then the device maintenance data after the fault includes the repair and adjustment data of devices such as transformers and circuit breakers in the substation.

[0055] Based on the selected multiple device maintenance data, the historical data reconstruction of the historical grid fault operation data set is generated. The reconstructed data set can reflect the real performance of the equipment after the current grid fault occurs, providing a more accurate reference for fault analysis and simulation.

[0056] Furthermore, the historical data reconstruction unit 30 in the embodiment of the present application is further used to perform the following steps: Step P331: Train the grid operation data reconstruction channel based on the generative adversarial network.

[0057] Step P332: Input the multiple device maintenance data and the historical grid fault operation data set into the grid operation data reconstruction channel to generate a reconstructed grid fault operation data set.

[0058] Specifically, the historical data reconstruction generation process first needs to use the generative adversarial network to train the grid operation data reconstruction channel. The generative adversarial network consists of a generator and a discriminator. The generator is responsible for generating reconstructed data that is as close as possible to the real fault data based on the input historical fault operation data and device maintenance data. The discriminator distinguishes whether the input data comes from the real data set or the fake data generated by the generator. Through continuous optimization, the discriminator helps the generator improve the authenticity of the data. Through repeated training, the generator and the discriminator continuously perform adversarial optimization. The generator gradually learns how to generate grid fault data more realistically, while the discriminator becomes more sensitive and can accurately judge which data are real and which are generated. Finally, the trained generative adversarial network is output as the grid operation data reconstruction channel. This grid operation data reconstruction channel can learn by inputting historical grid fault operation data and device maintenance data, so as to generate more accurate grid fault operation data.

[0059] Input the device maintenance data of multiple power grid devices and the historical power grid fault data into the trained power grid operation data reconstruction channel. This channel corrects the behavior of power grid devices during faults, supplements missing data, and adjusts the recovery of devices after faults according to maintenance records, making the dataset more in line with the actual operation status of the power grid after faults, and outputs a reconstructed power grid fault operation dataset to provide more reliable data support for subsequent power grid fault simulations.

[0060] Furthermore, the historical data reconstruction unit 30 in the embodiment of the present application is further configured to perform the following steps: Step P331-1: According to the fault operation record data of the same power grid, collect multiple sample historical power grid fault operation datasets and multiple sample device maintenance data sets, and obtain the power grid fault operation datasets after the same power grid fault occurs under different sample device maintenance data sets as multiple sample reconstructed power grid fault operation datasets.

[0061] Step P331-2: Based on the generative adversarial network, construct a generator and a discriminator in the power grid operation data reconstruction channel.

[0062] Step P331-3: Use the multiple sample historical power grid fault operation datasets and multiple sample device maintenance data sets to perform generative supervised training on the generator, and use the multiple sample reconstructed power grid fault operation datasets, combined with the generation results of the generator, to perform discriminative supervised training on the discriminator.

[0063] Step P331-4: Perform supervised training until the convergence requirement is met to obtain the power grid operation data reconstruction channel.

[0064] Specifically, the historical data reconstruction unit 30 trains the power grid operation data reconstruction channel based on the generative adversarial network, and the training process is as follows: Obtain the fault operation record data of the power grid with the same specifications and configurations as the target power grid, extract the power grid state information at the time of fault occurrence and the device maintenance information after fault occurrence under different power grid faults to generate multiple sample historical power grid fault operation datasets and multiple sample device maintenance data sets. At the same time, extract the power grid fault operation datasets when the same power grid fault occurs again after device maintenance according to different sample device maintenance data sets to generate multiple sample reconstructed power grid fault operation datasets. Through these sample data, it is possible to provide diverse inputs for subsequent generative adversarial network training, covering various scenarios and device maintenance situations during power grid fault occurrence.

[0065] Build a generative adversarial network model through a deep learning framework. This model includes a generator and a discriminator. The generator is responsible for generating fake data based on the input historical power grid fault operation data and device maintenance data, that is, predicting the operating state of the power grid when a fault occurs. The discriminator is responsible for evaluating whether the generated fake data is real or consistent with the actual power grid fault operation data set. Exemplarily, the generator can be a model composed of a multi-layer neural network, receiving the sample historical power grid fault operation data set and the sample device maintenance data set as inputs, and after a series of neuron calculations, outputting a reconstructed power grid fault operation data set. The discriminator can also be a neural network, receiving the sample reconstructed power grid fault operation data set and the result generated by the generator as inputs, extracting features and making judgments on the input data, and outputting a probability value representing the authenticity of the data.

[0066] During the training process, use multiple sample historical power grid fault operation data sets and sample device maintenance data sets to perform generative supervised training on the generator. The generator will generate prediction results (i.e., fake data) based on the input sample data. After the generator goes through a preset number of training steps, it outputs the predicted reconstructed power grid fault operation data set. Then, use multiple sample reconstructed power grid fault operation data sets and combine the generation results of the generator to train the discriminator. The discriminator learns how to distinguish between the real sample reconstructed power grid fault operation data set and the results generated by the generator, and adjusts its own parameters according to the accuracy of its judgment to improve the discrimination ability. After going through a preset number of training steps, it outputs the discrimination result. The generator adjusts its own parameters according to the discrimination result and performs the next round of training. Alternately train the generator and the discriminator for multiple rounds, continuously adjusting the parameters of the generator and the discriminator until the convergence requirement is met, that is, the loss function of the model reaches a minimum value or there is no obvious decrease. At this time, it means that the data generated by the generator is realistic enough to reconstruct the power grid fault operation data well, and the discriminator is difficult to distinguish between true and false data. Output the trained generator and discriminator to obtain the power grid operation data reconstruction channel.

[0067] Furthermore, the digital twin simulation unit 40 in the embodiment of the present application is further used to perform the following steps: Step P41: Use the multiple reconstructed fault operation data of multiple power grid devices in the reconstructed power grid fault operation data set to identify the multiple power grid devices.

[0068] Step P42: Based on digital twin, according to the identification result, construct a power grid fault twin model as the power grid fault twin simulation result.

[0069] Specifically, the reconstructed fault operation data of multiple grid devices in the reconstructed power grid fault operation dataset is used to identify the multiple grid devices, and the specific attributes or states of each grid device in the fault operation state are determined. For example, for a transformer in a substation, the reconstructed fault operation data includes information such as its oil temperature and voltage change rate during the fault. Based on this data, the transformer can be identified to determine what special state the transformer is in during the fault, such as whether there is an overheating risk or whether the voltage is unstable.

[0070] The identification process first classifies different grid devices according to the functions and types of the devices in the power grid, such as transformers, circuit breakers, switches, distribution lines, etc. The operating characteristics of each device are extracted from the reconstructed power grid fault operation dataset. These characteristics usually include: electrical parameters: such as voltage, current, power, frequency, etc.; equipment state changes: such as actions like equipment startup, stop, switching, tripping, recovery, etc.; time series data: such as equipment response time, recovery time, etc.

[0071] Based on the behavior changes of the equipment, specific fault types are identified, such as short - circuit faults, overload faults, grounding faults, equipment faults, etc. The state changes of each grid device at the time of the fault are associated with key information such as device type, fault occurrence time, and fault type, and the associated data and corresponding identification tags of each grid device in a specific fault scenario are established. For example, transformer A has a short - circuit fault at a certain time point, and transformer B shows an overload protection action. These fault information will be corresponding to the equipment one by one. The identification tag of transformer A is "short - circuit fault", the state record of device A is "in fault", and there are fault times (start time, end time) and related electrical parameters. Finally, the identification results are output in a standardized format (such as database, JSON, XML, etc.) for subsequent digital twin simulation. These data include the fault characteristics, equipment states, operation timings, etc. of each device, making the fault simulation of the power grid more real and reliable, and being able to effectively reflect the actual performance of the power grid in the case of faults.

[0072] After identifying the fault states of each grid device, a corresponding virtual model is constructed in the digital space, that is, the power grid fault twin model, which can accurately reflect the states of each device in the power grid during the fault. Taking a power grid containing multiple substations, transmission lines, and distribution network equipment as an example, after identifying the fault states of each grid device, a corresponding virtual model is constructed in the digital space. If a fault hidden danger is found in a certain area of a transmission line during the identification process, in the power grid fault twin model, the corresponding digital model part of the transmission line will show characteristics related to the fault hidden danger, such as abnormal resistance, signal transmission interruption, etc.

[0073] Taking this power grid fault twin model as the power grid fault twin simulation result, it provides an intuitive virtual scenario highly similar to the actual fault situation for power grid maintenance personnel, so that they can conduct fault analysis, formulate repair strategies, and carry out emergency drills, etc. For example, different repair measures can be simulated on this digital twin model, and the state changes of power grid devices in the model can be observed to select the optimal repair plan. At the same time, this model can also be used to predict the chain reactions that the fault may have on other parts of the power grid and take preventive measures in advance.

[0074] In summary, the power grid digital twin fault simulation device for historical data reconstruction provided by the embodiments of the present application has the following technical effects: When the fault feature acquisition unit 10 detects a fault in the target power grid, it acquires the power grid fault feature information. The historical data retrieval unit 20 traverses and indexes in the historical fault record data according to the power grid fault feature information to obtain the historical power grid fault feature information with the greatest similarity as the homologous historical power grid fault; retrieves the multiple historical fault operation data of multiple power grid devices in the target power grid when the homologous historical power grid fault occurs to obtain the historical power grid fault operation data set. By combining the real-time feature data when the power grid fault occurs with the historical data through the fault feature acquisition unit 10 and the historical data retrieval unit 20, and using the feature index-based method to retrieve the historical power grid fault data set, by combining similar fault historical data, the fault features can be understood more comprehensively and accurately, providing richer data support for subsequent simulations.

[0075] The historical data reconstruction unit 30 acquires the multiple device maintenance data of the multiple power grid devices after the homologous historical power grid fault in the target power grid, and trains the power grid operation data reconstruction channel based on the generative adversarial network; inputs the multiple device maintenance data and the historical power grid fault operation data set into the power grid operation data reconstruction channel to generate a reconstructed power grid fault operation data set. By reconstructing the historical fault operation data through the historical data reconstruction unit 30, a predicted operation data set under the current power grid fault is generated, supplementing the missing data and improving the comprehensiveness and accuracy of the fault simulation.

[0076] The digital twin simulation unit 40 uses the reconstructed power grid fault operation data set, based on digital twin, constructs the power grid fault twin simulation result, simulates a more accurate power grid fault situation, provides real-time simulation results for power grid operation and maintenance personnel, and assists in decision-making.

[0077] Overall, by integrating the above functional units, the embodiments of the present application combine the reconstruction of historical fault data with digital twin technology, achieving systematic analysis and simulation of power grid faults, being able to predict the power grid behavior under different fault conditions, significantly improving the accuracy of fault diagnosis and the forward-looking nature of fault prevention. Through accurate fault simulation, the changes in the power grid state after a fault can be better predicted, countermeasures can be formulated in advance, the impact of the fault on the power grid operation can be reduced, and the stable operation of the power grid can be ensured. At the same time, the application of digital twin technology allows for testing and optimization of faults in a virtual environment, providing more targeted strategies for power grid maintenance, reducing risks in actual operations, and thus improving the power grid operation efficiency and overall stability.

[0078] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A power grid digital twin fault simulation device for historical data reconstruction, characterized in that The device includes: A fault feature acquisition unit, which is used to acquire grid fault feature information when a fault is detected in the target power grid; A historical data retrieval unit, which is used to perform a historical power grid fault index according to the grid fault feature information to obtain homologous historical power grid faults, and index to obtain a historical power grid fault operation data set when the homologous historical power grid faults occur, where the historical power grid fault operation data set includes multiple historical fault operation data of multiple power grid devices in the target power grid; A historical data reconstruction unit, which is used to obtain multiple device maintenance data of the multiple power grid devices after the homologous historical power grid faults in the target power grid, and perform historical data reconstruction on the historical power grid fault operation data set to generate a reconstructed power grid fault operation data set; A digital twin simulation unit, which is used to adopt the reconstructed power grid fault operation data set and construct a power grid fault twin simulation result based on digital twin.

2. The power grid digital twin fault simulation device for historical data reconstruction according to claim 1, characterized in that The fault feature acquisition unit is further used to perform the following steps: When a fault is detected in the target power grid, acquire fault location information and fault parameters; Integrate the fault location information and fault parameters to obtain grid fault feature information.

3. The power grid digital twin fault simulation device for historical data reconstruction according to claim 1, characterized in that, The historical data retrieval unit is further used to perform the following steps: Obtain historical fault record data of the target power grid within a preset historical time range; Based on the grid fault feature information, perform a traversal index in the historical fault record data to obtain the historical power grid fault feature information with the greatest similarity as the homologous historical power grid faults; Retrieve multiple historical fault operation data of multiple power grid devices in the target power grid when the homologous historical power grid faults occur to obtain the historical power grid fault operation data set.

4. The power grid digital twin fault simulation device for historical data reconstruction according to claim 3, characterized in that, The historical data retrieval unit is further used to perform the following steps: Based on the grid fault feature information, perform a traversal index in the historical fault record data, calculate the fault location similarity between the fault location information in the grid fault feature information and multiple historical fault location information, and calculate the fault parameter similarity between the fault parameters in the grid fault feature information and multiple historical fault parameters; Perform a weighted calculation on the fault location similarity and fault parameter similarity of each historical power grid fault feature information to obtain multiple historical power grid fault similarities; Select the historical power grid fault feature information with the greatest historical power grid fault similarity and output to obtain homologous historical power grid faults.

5. The power grid digital twin fault simulation device for historical data reconstruction according to claim 1, wherein The historical data reconstruction unit is further used to perform the following steps: Obtain the maintenance data records of the multiple power grid devices in the target power grid; In the maintenance data records, retrieve multiple device maintenance data of the multiple power grid devices within the time after the homologous historical power grid faults; Perform historical data reconstruction on the historical power grid fault operation data set according to the multiple device maintenance data.

6. The power grid digital twin fault simulation device for historical data reconstruction according to claim 5, characterized in that The historical data reconstruction unit is further used to perform the following steps: Train a power grid operation data reconstruction channel based on a generative adversarial network; Input the multiple device maintenance data and the historical power grid fault operation data set into the power grid operation data reconstruction channel to generate a reconstructed power grid fault operation data set.

7. The power grid digital twin fault simulation device for historical data reconstruction according to claim 6, characterized in that, The historical data reconstruction unit is further configured to perform the following steps: According to the fault operation record data of the same power grid, collect multiple sample historical power grid fault operation data sets and multiple sample device maintenance data sets, and obtain the power grid fault operation data sets after the same power grid fault occurs under different sample device maintenance data sets as multiple sample reconstructed power grid fault operation data sets; Based on the generative adversarial network, construct a generator and a discriminator in the power grid operation data reconstruction channel; Use the multiple sample historical power grid fault operation data sets and multiple sample device maintenance data sets to perform generative supervised training on the generator, and use the multiple sample reconstructed power grid fault operation data sets and the generation results of the generator to perform discriminative supervised training on the discriminator; Perform supervised training until the convergence requirement is met to obtain the power grid operation data reconstruction channel.

8. The power grid digital twin fault simulation device for historical data reconstruction according to claim 1, characterized in that The digital twin simulation unit is further configured to perform the following steps: Use the multiple reconstructed fault operation data of multiple power grid devices in the reconstructed power grid fault operation data set to identify the multiple power grid devices; Based on digital twin, construct a power grid fault twin model according to the identification result as the power grid fault twin simulation result.

Images