Power transformation equipment digital twin data processing method and system based on big data

Abstract

The embodiment of the invention provides a big data-based power transformation equipment digital twin data processing method and system, and belongs to the technical field of fault prediction. The method comprises the following steps: acquiring a multi-modal data set of power transformation equipment, and constructing an equipment dependency graph according to the multi-modal data set; and inputting the equipment dependency relationship graph into a preset graph convolutional network to obtain an equipment health state representation. And inputting the equipment health state representation into a preset converter model for time sequence modeling to obtain health state prediction data. And inputting the equipment health state representation and the health state prediction data into a preset multi-layer perceptron model to obtain an equipment fault prediction result. And the fault prediction accuracy of the power transformation equipment is improved.

Description

Technical Field

[0001] This application relates to the field of fault prediction technology, and in particular to a method and system for processing digital twin data of power equipment based on big data. Background Technology

[0002] With the increasing scale and complexity of power systems, substation equipment, as a core component, directly impacts the normal operation of the system through its stability and security. Currently, substation fault prediction technologies typically rely on static models or traditional machine learning methods. However, these methods struggle to handle the dynamic changes in time-series data and the complexities of interactions between devices, resulting in significant shortcomings in real-time fault warnings and early anomaly pattern detection. Therefore, improving the accuracy of substation fault prediction has become a pressing technical challenge. Summary of the Invention

[0003] The main objective of this application is to propose a method and system for processing digital twin data of power equipment based on big data, which aims to improve the accuracy of power equipment fault prediction.

[0004] To achieve the above objectives, a first aspect of this application proposes a method for processing digital twin data of substation equipment based on big data, the method comprising: Obtain multimodal datasets of power equipment; Construct a device dependency graph based on the multimodal dataset; The device dependency graph is input into a preset graph convolutional network to obtain a representation of the device health status; The device health status representation is input into a preset converter model for time series modeling to obtain health status prediction data; The device health status representation and the health status prediction data are input into a preset multilayer perceptron model to obtain the device fault prediction result.

[0005] In some embodiments, after inputting the device health status representation and the health status prediction data into a preset multilayer perceptron model to obtain the device fault prediction result, the method further includes: Obtain the importance weight of each power equipment; The global health status assessment result is obtained by weighting the importance weights and the health status prediction data.

[0006] In some embodiments, after obtaining a global health status assessment result by weighted averaging based on the importance weights and the health status prediction data, the method further includes: The equipment failure prediction results, the importance weights, the global health status assessment results, and the preset maintenance strategies are input into a preset multi-objective optimization model to obtain the optimal maintenance strategy for each substation. Each piece of equipment is maintained according to the optimal maintenance strategy described above.

[0007] In some embodiments, the step of inputting the equipment fault prediction results, the importance weights, the global health status assessment results, and the preset maintenance strategies into a preset multi-objective optimization model to obtain a maintenance strategy for each substation device includes: A maintenance cost function is constructed based on the equipment failure prediction results, the global health status assessment results, and the maintenance strategy. Based on the importance weights and the maintenance cost function, optimization calculations are performed to obtain the optimal maintenance strategy for each substation.

[0008] In some embodiments, acquiring the multimodal dataset of the power equipment includes: Obtain the original dataset of the power equipment; wherein the original dataset includes temperature data, current data, and voltage data; The temperature data, current data, and voltage data are preprocessed to obtain the multimodal dataset.

[0009] In some embodiments, the step of inputting the device health status representation into a preset converter model for time-series modeling to obtain health status prediction data includes: The device health status representations are arranged in chronological order to obtain a time input sequence; The time input sequence is input into the converter model for time series modeling to obtain the health status prediction data; wherein, the converter model includes a time difference regularization term to ensure that the converter model maintains the smooth change of the device health status in time series modeling.

[0010] In some embodiments, the step of inputting the device health status representation and the health status prediction data into a preset multilayer perceptron model to obtain device fault prediction results includes: The device health status representation and the health status prediction data are concatenated to obtain a concatenated vector. The spliced ​​vector is input into the multilayer perceptron model to obtain the equipment fault prediction result.

[0011] To achieve the above objectives, a second aspect of this application proposes a digital twin data processing system for substation equipment based on big data, the system comprising: The acquisition module is used to acquire multimodal datasets of power equipment. A building module is used to construct a device dependency graph based on the multimodal dataset; The graph convolution module is used to input the device dependency graph into a preset graph convolution network to obtain a representation of the device health status; The timing module is used to input the device health status representation into a preset converter model for timing modeling to obtain health status prediction data; The multi-layer sensing module is used to input the device health status representation and the health status prediction data into a preset multi-layer sensing machine model to obtain the device fault prediction result.

[0012] To achieve the above objectives, a third aspect of this application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the method described in the first aspect.

[0013] To achieve the above objectives, a fourth aspect of the present application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in the first aspect.

[0014] This application proposes a method and system for digital twin data processing of substation equipment based on big data. It acquires a multimodal dataset of substation equipment and constructs an equipment dependency graph based on the dataset. The equipment dependency graph is then input into a pre-defined graph convolutional network to obtain a representation of the equipment's health status. This health status representation is then input into a pre-defined converter model for time-series modeling to obtain health status prediction data. Finally, the equipment health status representation and the predicted health status data are input into a pre-defined multilayer perceptron model to obtain equipment fault prediction results. This improves the accuracy of substation equipment fault prediction. Attached Figure Description

[0015] Figure 1 This is a flowchart of a digital twin data processing method for power equipment based on big data, provided in an embodiment of this application. Figure 2 yes Figure 1 The flowchart of step S101 in the text; Figure 3 yes Figure 1 The flowchart of step S104 in the process; Figure 4 yes Figure 1 The flowchart of step S105 in the process; Figure 5 This is a flowchart of a digital twin data processing method for power equipment based on big data, provided in another embodiment of this application; Figure 6This is a flowchart of a digital twin data processing method for power equipment based on big data, provided in the third embodiment of this application; Figure 7 yes Figure 6 The flowchart of step S601 in the process; Figure 8 This is a schematic diagram of the structure of the digital twin data processing system for power equipment based on big data provided in the embodiments of this application; Figure 9 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0017] It should be noted that although functional modules are divided in the system diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the system or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0019] As the scale and complexity of power systems continue to increase, the stability and safety of substation equipment, as a core component of the power system, directly affect the normal operation of the system. Among related technologies, fault prediction techniques for substation equipment typically rely on static models or traditional machine learning methods. These methods struggle to handle the dynamic changes in time-series data and the complexity of interactions between equipment, resulting in significant shortcomings in real-time fault warning and early anomaly pattern detection.

[0020] Based on this, embodiments of this application provide a method and system for processing digital twin data of substation equipment based on big data, aiming to acquire multimodal datasets of substation equipment. A device dependency graph is constructed based on the multimodal dataset. The device dependency graph is input into a preset graph convolutional network to obtain a representation of the current equipment health status. The device health status representation is input into a preset converter model for time series modeling to obtain health status prediction data. The device health status representation and health status prediction data are input into a preset multilayer perceptron model to obtain equipment fault prediction results, thereby improving the accuracy of substation equipment fault prediction.

[0021] The method and system for processing digital twin data of power equipment based on big data provided in this application are specifically described through the following embodiments. First, the method for processing digital twin data of power equipment based on big data in this application embodiment is described.

[0022] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0023] Foundational technologies for artificial intelligence generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies mainly encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0024] The big data-based digital twin data processing method for substation equipment provided in this application can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, etc.; the server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the big data-based digital twin data processing method for substation equipment, but is not limited to the above forms.

[0025] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0026] Please refer to Figure 1 , Figure 1 This is a flowchart of a digital twin data processing method for power equipment based on big data, provided in an embodiment of this application. Figure 1 The method may include, but is not limited to, steps S101 to S105.

[0027] Step S101: Obtain the multimodal dataset of the power equipment; Step S102: Construct a device dependency graph based on the multimodal dataset; Step S103: Input the device dependency graph into a preset graph convolutional network to obtain a representation of the device health status; Step S104: Input the device health status representation into the preset converter model for time series modeling to obtain health status prediction data; Step S105: Input the equipment health status representation and health status prediction data into the preset multilayer perceptron model to obtain the equipment fault prediction result.

[0028] Please see Figure 2 In some embodiments, step S101 may include, but is not limited to, steps S201 to S202: Step S201: Obtain the original dataset of the power equipment; wherein, the original dataset includes temperature data, current data and voltage data; Step S202: Preprocess the temperature data, current data, and voltage data to obtain a multimodal dataset.

[0029] In step S201 of some embodiments, the raw dataset comes from the real-time monitoring system of the power equipment and is collected in real time by a sensor acquisition module. The data acquired by the sensors includes temperature data, current data, and voltage data.

[0030] Temperature data: The operating temperature of the equipment is used to detect whether the equipment is overheating and is often an early indication of equipment failure. The temperature sensor collects data once per second.

[0031] Current data: The operating current of the equipment reflects its operation under different loads and can reveal abnormal currents when the load is too high or the equipment malfunctions. The current sensor collects data every 2 seconds.

[0032] Voltage data: The operating voltage of the equipment, used to monitor voltage fluctuations, especially since unstable voltage may lead to equipment failure or damage. The voltage sensor collects data once per second.

[0033] Each data point is accompanied by a precise timestamp. Original dataset Each element in Indicates device Sensor data collected at a certain moment includes multi-dimensional information such as temperature, current, and voltage.

[0034] In step S202 of some embodiments, the core objective of data preprocessing is to ensure that the acquired data can be efficiently utilized in subsequent analysis. Different types of sensors may have different sampling frequencies, and the timestamps of the data may not be completely consistent. The timestamps of different sensors need to be aligned to ensure that their time steps are consistent. For example, assuming that a temperature sensor samples once per second and a current sensor samples once every 2 seconds, the current data will be synchronized to once per second using an interpolation method to ensure time alignment throughout the entire dataset.

[0035] During operation, equipment may be subject to environmental interference, leading to outliers or noise in the collected data. A sliding window method can be used to denoise the raw data. Specifically, a sliding window (e.g., window size 5) is set for the data from each sensor. Data within the window is averaged or processed using the median, thereby removing abrupt changes and noise. For example, when current data fluctuates drastically at a certain moment, and this fluctuation does not conform to the trend at other time points, the sliding window method can effectively smooth out these outliers.

[0036] In addition, equipment may fail to record data at certain times due to malfunctions or operational interruptions, resulting in data gaps. For handling missing data, linear interpolation is used, which involves interpolating and filling in the missing data based on values ​​before and after the missing data. For example, if voltage data is missing at a certain moment, a reasonable filler value can be calculated using data from the previous and next moments. For long-term outages or very short intervals, trend line interpolation of time series data can also be used to fill in the missing data.

[0037] Finally, to avoid dimensional differences between different sensors, all sensor data needs to be normalized. A Min-Max normalization method is used to adjust the values ​​of each sensor to the range [0,1]. Specifically, the minimum and maximum values ​​of each sensor's data are calculated and used to standardize the sensor data. For each sensor data... The result after normalization This normalization is achieved by subtracting the minimum value from the original data and then dividing by the difference between the maximum and minimum values. This process ensures that the numerical ranges of all sensors remain consistent, thus eliminating dimensional differences and preventing excessively large or small sensor values ​​from unnecessarily impacting model training. Normalized data ensures that each feature has a consistent impact on subsequent model training, thereby improving model stability and performance.

[0038] After the above preprocessing, a standardized multimodal dataset is output. Each of them It is a device that has undergone preprocessing (denoising, timing alignment, missing value imputation, and normalization). Multimodal data. This data includes standardized values ​​from sensors such as temperature, current, and voltage, providing high-quality input data for subsequent modeling of inter-device interactions and health status prediction.

[0039] Through the above steps S201 to S202, the equipment data acquisition and preprocessing are realized, which solves the problems of time sequence alignment, noise processing, missing value filling and dimensional difference of sensor data, and provides high-quality input for subsequent models.

[0040] In steps S102 to S103 of some embodiments, each node in the device dependency graph represents a device, and the node features contain the device's multimodal data. Furthermore, the interactions between devices (such as electrical connections and physical dependencies) constitute the edges in the graph, and the weight of each edge reflects the mutual influence between devices. These weights can be dynamically calculated based on factors such as current and voltage fluctuations between devices to construct the device dependency graph.

[0041] A pre-trained graph convolutional network is used to input the device dependency graph. The graph convolutional network aggregates the health status information of the devices through multiple layers of graph convolution. The graph convolution operation of each layer is shown in the following formula (1): (1) in, Indicates device In the Health status representation in layered graph convolutional networks Indication and equipment A set of adjacent connected devices. and Representing the equipment and equipment The degree is the number of devices directly connected to these devices. The degree reflects the connectivity of a device in the system and is used to adjust the weight of interactions between devices. and Indicates the first The weight matrix and bias terms of a layered graphical neural network. This represents the activation function (such as ReLU).

[0042] Used to indicate equipment and equipment The electrical load relationship or mutual influence coefficient between them. The specific calculation method is based on the equipment. and equipment The current and voltage data are used to calculate their correlation, reflecting the synchronicity of their electrical signals within the same time frame. Then, by comparing the similarity of current and voltage fluctuations between devices, a dynamic influence coefficient is derived, i.e. A larger value indicates a stronger interaction between devices, and vice versa. This method ensures that the interaction between devices can be dynamically adjusted according to changes in electrical load, thus accurately reflecting the mutual influence between devices. λ is a hyperparameter used to control... The degree of impact on the model output.

[0043] Specifically, after processing by the graph convolutional network, the output is a representation of the device health status of each device. ,in This indicates the number of layers in the graph neural network. Each It is equipment In graph convolutional networks, the health status representation after multiple convolutions includes the device's current health status and its interactions with other devices. These outputs, fusing the device's own health status with its interactions with other devices, better capture the dynamic dependencies between devices. This provides accurate global health status data for subsequent steps such as time-series data modeling and fault prediction.

[0044] Please see Figure 3 In some embodiments, step S104 may include, but is not limited to, steps S301 to S302: Step S301: Arrange the device health status representations in chronological order to obtain a time input sequence; Step S302: Input the time input sequence into the converter model for time series modeling to obtain health status prediction data; wherein, the converter model includes a time difference regularization term to ensure that the converter model maintains the smooth change of the device health status in time series modeling.

[0045] In step S301 of some embodiments, the device health status representations are arranged in chronological order to form a time input sequence.

[0046] In step S302 of some embodiments, the time input sequence is input into the transformer model for time series modeling. The core of the transformer model is a self-attention mechanism, which calculates a query matrix. Key matrix Sum matrix To learn the dependencies between different time steps.

[0047] Specifically, health status indicates The device health status is transformed into a query, key, and value matrix through linear mapping, and then self-attention is computed to capture the long-term dependencies of device health status over time. The core idea of ​​the self-attention mechanism is to dynamically adjust the weights between time steps by calculating the relationship between the query information of the device health representation and other time steps. The device health representation is converted into three vectors: query, key, and value. The query vector represents the current device health status, the key vector represents the device health status at historical time steps, and the value vector contains specific health status information. To evaluate the correlation between the current state and historical states, the similarity between the query vector and the key vector is calculated. The resulting similarity values ​​are then normalized using softmax to ensure that the sum of the similarity values ​​is 1.

[0048] These similarity weights are then applied to the value vector to obtain weighted health status information. Specifically, higher similarity means a stronger correlation between the current device health status and its historical status at a given point in time, thus assigning a higher weight to the health status at that point. This allows the model to dynamically adjust the dependencies between different time steps based on changes in device health status through a self-attention mechanism, thereby better capturing the evolution trend of device health status over time. A time difference regularization term is also introduced on top of the traditional converter model to ensure that the model maintains a smooth change in device health status during time-series modeling, avoiding overfitting. (Time difference regularization term) As shown in the following formula (2): (2) in, It is equipment In the The health status representation output by the layer graph convolutional network. It is equipment The regularization term prompts the model to pay more attention to the stationarity of device health status during prediction, thereby enhancing the stability and accuracy of device health status prediction. Ultimately, the converter model outputs the health status prediction data for each device at the next time step. This output indicates the device At any moment Health status prediction These represent future time steps. By predicting these health statuses, the model can provide early warnings of potential equipment failures and offer a more accurate global health assessment.

[0049] The interactions and health status of the equipment have been modeled using a graph neural network (GNN), resulting in a global health status representation for each device. However, the health status of the equipment is not static; it changes continuously over time. This is especially true for systems like power substations, which exhibit long-term dependencies. Therefore, by employing steps S301 to S302, a transformer model is used to perform time-series modeling of the equipment health status data. The transformer model, due to its superior self-attention mechanism and ability to capture long-term dependencies, can efficiently handle changes in equipment health status over time, thereby improving the accuracy of fault prediction and health status assessment.

[0050] Please see Figure 4 In some embodiments, step S105 may include, but is not limited to, steps S401 to S402: Step S401: The device health status representation and health status prediction data are concatenated to obtain a concatenated vector; Step S402: Input the spliced ​​vector into the multilayer perceptron model to obtain the equipment fault prediction result.

[0051] In step S401 of some embodiments, the current and future health status data are merged into a vector, that is, the device health status representation. and health status prediction data Concatenate the vectors to obtain the concatenated vector. .

[0052] In step S402 of some embodiments, the concatenated vector is input into a multilayer perceptron (MLP) model for prediction to obtain a device fault prediction result. The model output is a binary classification result indicating whether the device has a fault risk. The MLP loss function uses the cross-entropy loss function, enabling the model to learn a more accurate fault prediction capability. The calculation formula for MLP is shown in formula (3) below: (3) in, Indicates device The equipment failure prediction results indicate that the equipment The probability of failure in the future. It is the Sigmoid activation function, which outputs a failure probability value, ranging from [0,1]. The larger the value, the higher the risk of equipment failure. and It is the weight matrix of the MLP model. and It is a bias term. Indicates the health status of the device. and health status prediction data The concatenated vector.

[0053] It should be noted that during the training of the MLP model, by labeling the historical fault data and health status of the equipment, the MLP model can learn to identify the correlation between the health status of the equipment and faults, thereby achieving accurate fault prediction.

[0054] Through steps S401 to S402 above, the multilayer perceptron model determines the current health status of the device. and prediction of future health status As input, the mapping relationship between equipment health status and failure occurrence is learned through several layers of nonlinear transformation, which can more accurately identify the risk of equipment failure.

[0055] Steps S101 to S105 of this embodiment involve acquiring a multimodal dataset of substation equipment and constructing an equipment dependency graph based on the dataset. The equipment dependency graph is then input into a preset graph convolutional network to obtain a representation of the equipment health status. This representation is then input into a preset converter model for time-series modeling to obtain health status prediction data. Finally, the equipment health status representation and the health status prediction data are input into a preset multilayer perceptron model to obtain equipment fault prediction results. This approach solves the problems of temporal alignment, noise processing, missing value imputation, and dimensional differences in sensor data, providing high-quality input for subsequent models. Next, a device dependency graph is constructed between devices using a graph convolutional network to capture the dependencies and dynamic interactions between devices. The converter model is used for time-series modeling of equipment health status, handling long-term dependencies and accurately predicting changes in equipment health. Combining health status prediction and multilayer perceptron fault prediction improves the accuracy of substation equipment fault prediction.

[0056] Please refer to Figure 5 In some embodiments, after step S105, the method for processing digital twin data of power equipment based on big data may also include, but is not limited to, steps S501 to S502: Step S501: Obtain the importance weight of each power equipment; Step S502: Perform a weighted average based on importance weights and health status prediction data to obtain the global health status assessment result.

[0057] In step S501 of some embodiments, the importance weight of the power equipment is... Calculations can be made based on factors such as the importance of the equipment, its load, and historical fault records.

[0058] In step S502 of some embodiments, a weighted average is performed based on importance weights and health status prediction data to obtain the global health status assessment result, as shown in the following formula (4): (4) in, This represents the overall health status assessment result, indicating the overall health level of the substation system. Its value is typically between [0,1], with higher values ​​indicating better overall system health. Indicates device The importance weight indicates the relative importance of the device in the system. Indicates device In the future Health status prediction data.

[0059] Steps S501 to S502 shown in this embodiment can not only assess the health status of individual devices, but also determine the health status of the entire substation system through a global health status assessment. By combining the weights of each device, the impact of key devices on the overall system health can be assessed more accurately.

[0060] Please refer to Figure 6 In some embodiments, after step S105, the method for processing digital twin data of power equipment based on big data may also include, but is not limited to, steps S501 to S502: Step S601: Input the equipment fault prediction results, importance weights, global health status assessment results and preset maintenance strategies into the preset multi-objective optimization model to obtain the optimal maintenance strategy for each substation equipment. Step S602: Perform maintenance on each substation device according to the optimal maintenance strategy.

[0061] Please see Figure 7 In some embodiments, step S601 may include, but is not limited to, steps S701 to S702: Step S701: Construct a maintenance cost function based on the equipment failure prediction results, the global health status assessment results, and the maintenance strategy; Step S702: Perform optimization calculations based on importance weights and maintenance cost functions to obtain the optimal maintenance strategy for each substation.

[0062] In step S701 of some embodiments, Indicates device Maintenance decisions, It is a maintenance cost function calculated based on the results of global health status assessment and equipment failure prediction, balancing the improvement of equipment health status and the rational allocation of system resources.

[0063] In step S702 of some embodiments, the input to the multi-objective optimization model includes equipment failure prediction results. and overall health status assessment results Equipment failure prediction results Indicates device The probability of future failures, and the results of the global health status assessment. This reflects the overall health level of the system. Using these two key data points, the system can determine the health status and failure risk of equipment and optimize maintenance decisions based on this information. Furthermore, the equipment's historical maintenance records and importance weights will also serve as auxiliary inputs to help the reinforcement learning model formulate more personalized maintenance strategies.

[0064] The multi-objective optimization model considers equipment failure prediction, global health assessment, and the actual importance of the equipment. Through the multi-objective optimization model, the system can balance maintenance costs and the improvement of equipment health status, generating the optimal maintenance strategy. For example, the optimization objective function can be formalized as shown in the following formula (5): (5) in, Indicates device Maintenance decisions, This refers to the importance weight of the equipment. Indicates the equipment failure prediction results and This indicates the overall health status assessment result. This maintenance cost function is calculated based on global health status assessment results and equipment failure prediction results, balancing the improvement of equipment health status with the rational allocation of system resources. This optimization process ensures that the system can adaptively adjust the maintenance strategy for each device based on real-time data, generating the optimal maintenance strategy. This avoids unnecessary repairs and improves equipment operating efficiency.

[0065] Through the above steps S701 to S702, the optimal maintenance strategy can be generated intelligently and automatically, ensuring that the health management of equipment is optimized at the system level.

[0066] In step S602 of some embodiments, when executing the optimized maintenance strategy, the system makes an optimal maintenance decision based on the generated optimal maintenance decision. Automatically perform preventative or corrective maintenance on equipment. This is based on equipment failure prediction results. When the value is high, the system will trigger preventative maintenance operations to reduce the likelihood of equipment failure; while when the equipment failure prediction result is high... When the threshold approaches 1, the system will perform corrective maintenance operations to ensure timely repair of the equipment. In addition, the system will optimize resource scheduling based on the importance of the equipment, its load, and its operating status, ensuring that high-risk and critical equipment receives priority for repair.

[0067] For example, when the fault prediction probability When the voltage is between 0.6 and 0.9, the system determines that the equipment has shown significant degradation but has not yet reached the point of failure. At this time, it will automatically issue a preventive maintenance command. Preventive maintenance typically includes online thermal imaging and partial discharge detection, tightening terminals and cleaning terminal blocks, rapid sampling and analysis of insulating oil or SF6 gas, and short-term shutdowns during low load periods to replace consumable parts (such as surge arrester discharge counters, aging cable heads, or fan bearings). These measures can eliminate potential hazards without affecting long-term shutdown of the main circuit. Pull it back below the safety threshold and extend the device's healthy lifespan.

[0068] when When the value approaches 1, it indicates that a fault is almost certain to occur in the near future, and the system immediately enters corrective maintenance mode: First, the relevant bays or feeders are switched to backup paths according to the dispatching procedures. Then, unplanned power outages or live-line maintenance are arranged (depending on the site conditions). Failed components are replaced or repaired, such as rewinding transformer coils, repairing circuit breaker mechanisms, replenishing fluid or replacing insulation media, and the withstand voltage and breaking tests of the primary circuit are completed. During this stage, the maintenance log and model weights are also updated simultaneously to ensure that subsequent predictions are more consistent with the latest operation and maintenance data.

[0069] Steps S601 to S602 of this embodiment generate an optimal maintenance strategy and perform maintenance on each substation device according to the strategy. This not only reduces equipment downtime but also improves equipment lifespan and overall system stability, providing a scientific basis and technical support for intelligent equipment maintenance.

[0070] Please see Figure 8 This application also provides a big data-based digital twin data processing system for substation equipment, which can implement the above-mentioned big data-based digital twin data processing method for substation equipment. The system includes: The acquisition module 801 is used to acquire the multimodal dataset of the power equipment. Module 802 is used to build a device dependency graph based on a multimodal dataset; Graph convolution module 803 is used to input the device dependency graph into a preset graph convolution network to obtain a representation of the device health status; The timing module 804 is used to input the device health status representation into the preset converter model for timing modeling to obtain health status prediction data; The multi-layer sensing module 805 is used to input the equipment health status representation and health status prediction data into the preset multi-layer sensing machine model to obtain the equipment fault prediction result.

[0071] The specific implementation of this big data-based digital twin data processing system for power equipment is basically the same as the specific implementation of the big data-based digital twin data processing method for power equipment described above, and will not be repeated here.

[0072] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described method for processing digital twin data of power equipment based on big data. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0073] Please see Figure 9 , Figure 9 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes: The processor 901 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application. The memory 902 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 902 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 902 and is called and executed by the processor 901 according to the big data-based digital twin data processing method for substation equipment of this application embodiment. The input / output interface 903 is used to implement information input and output; The communication interface 904 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 905 transmits information between various components of the device (e.g., processor 901, memory 902, input / output interface 903, and communication interface 904); The processor 901, memory 902, input / output interface 903, and communication interface 904 are connected to each other within the device via bus 905.

[0074] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for processing digital twin data of power equipment based on big data.

[0075] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0076] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0077] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0078] The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0079] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0080] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0081] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0082] In the embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection of the system or units may be electrical, mechanical, or other forms.

[0083] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0084] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0085] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0086] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A method for processing digital twin data of substation equipment based on big data, characterized in that, The method includes: Obtain multimodal datasets of power equipment; Construct a device dependency graph based on the multimodal dataset; The device dependency graph is input into a preset graph convolutional network to obtain a representation of the device health status; The device health status representation is input into a preset converter model for time series modeling to obtain health status prediction data; The device health status representation and the health status prediction data are input into a preset multilayer perceptron model to obtain the device fault prediction result.

2. The method according to claim 1, characterized in that, After inputting the device health status representation and the health status prediction data into a preset multilayer perceptron model to obtain the device fault prediction result, the method further includes: Obtain the importance weight of each power equipment; The global health status assessment result is obtained by weighting the importance weights and the health status prediction data.

3. The method according to claim 2, characterized in that, After obtaining the global health status assessment result by weighted averaging based on the importance weights and the health status prediction data, the method further includes: The equipment failure prediction results, the importance weights, the global health status assessment results, and the preset maintenance strategies are input into a preset multi-objective optimization model to obtain the optimal maintenance strategy for each substation. Each piece of equipment is maintained according to the optimal maintenance strategy described above.

4. The method according to claim 3, characterized in that, The step of inputting the equipment fault prediction results, the importance weights, the global health status assessment results, and the preset maintenance strategies into a preset multi-objective optimization model to obtain the maintenance strategy for each substation equipment includes: A maintenance cost function is constructed based on the equipment failure prediction results, the global health status assessment results, and the maintenance strategy. Based on the importance weights and the maintenance cost function, optimization calculations are performed to obtain the optimal maintenance strategy for each substation.

5. The method according to claim 1, characterized in that, The acquisition of the multimodal dataset of the power equipment includes: Obtain the original dataset of the power equipment; wherein the original dataset includes temperature data, current data, and voltage data; The temperature data, current data, and voltage data are preprocessed to obtain the multimodal dataset.

6. The method according to claim 1, characterized in that, The step of inputting the device health status representation into a preset converter model for time series modeling to obtain health status prediction data includes: The device health status representations are arranged in chronological order to obtain a time input sequence; The time input sequence is input into the converter model for time series modeling to obtain the health status prediction data; wherein, the converter model includes a time difference regularization term to ensure that the converter model maintains the smooth change of the device health status in time series modeling.

7. The method according to claim 1, characterized in that, The step of inputting the device health status representation and the health status prediction data into a preset multilayer perceptron model to obtain the device fault prediction result includes: The device health status representation and the health status prediction data are concatenated to obtain a concatenated vector. The spliced ​​vector is input into the multilayer perceptron model to obtain the equipment fault prediction result.

8. A digital twin data processing system for substation equipment based on big data, characterized in that, The system includes: The acquisition module is used to acquire multimodal datasets of power equipment. A building module is used to construct a device dependency graph based on the multimodal dataset; The graph convolution module is used to input the device dependency graph into a preset graph convolution network to obtain a representation of the device health status; The timing module is used to input the device health status representation into a preset converter model for timing modeling to obtain health status prediction data; The multi-layer sensing module is used to input the device health status representation and the health status prediction data into a preset multi-layer sensing machine model to obtain the device fault prediction result.

9. An electronic device, characterized in that, The electronic device includes a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements the digital twin data processing method for power equipment based on big data as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the digital twin data processing method for power equipment based on big data as described in any one of claims 1 to 7.

Images