A new energy station high-voltage voltage transformer error evaluation method and system

Abstract

The application provides a new energy station high-voltage voltage transformer error evaluation method and system, the method comprises the following steps: collecting voltage phasors of all measurement nodes in a predetermined time period in the substation and the new energy station simultaneously, obtaining voltage transformer equipment parameters and line parameters between measurement nodes, and constructing a measurement data set; based on the power system line topology structure and the data in the measurement data set, the corresponding adjacency matrix and node feature matrix are constructed by extracting the line impedance and node working condition characteristics, and the time series graph structure data set is obtained according to the adjacency matrix and the node feature matrix; a graph neural network is constructed, and the graph neural network is trained based on the time series graph structure data set; input the voltage phasor of each measurement node, and judge the voltage transformer state through the preset threshold based on the output result of the trained graph neural network. Through the scheme, the voltage transformer error state online evaluation can be realized, and the accuracy and interpretability of the error evaluation are improved.

Description

Technical Field

[0001] This invention belongs to the field of power equipment monitoring technology, and in particular relates to a method and system for error assessment of high-voltage voltage transformers in new energy power plants. Background Technology

[0002] High-voltage transformers (VTs) are critical primary / secondary conversion devices in the power metering, relay protection, and monitoring control chain. Their ratio and phase errors directly affect the accuracy of power calculation and the reliability of protection settings. With the large-scale grid connection of new energy power plants such as wind and solar power, the power grid operation is characterized by rapid power fluctuations, frequent switching of operating modes, and a more complex electromagnetic environment. Traditional error assessment methods based on static operating condition assumptions are facing challenges in terms of accuracy, timeliness, and robustness.

[0003] From the perspective of error mechanism, the measurement error of VT is affected by the coupling of multiple factors, such as aging of the transformer body and insulation medium, changes in secondary load, frequency offset and harmonic components, all of which can cause time-varying shifts in ratio error and phase error. In new energy power plants, the randomness of output and the use of power electronic equipment make these offsets exhibit stronger non-stationarity and scenario-dependent characteristics, making it difficult for offline calibration results to be applicable to all scenarios, and traditional periodic verification and spot checks are unable to reflect the online status in a timely manner.

[0004] In current engineering practice, one type of method relies on periodic offline calibration or on-site sampling comparison. While this provides a benchmark-level reference, it requires power outages or special access conditions, limiting the frequency of calibrations and resulting in high costs. Another type of method relies on online error characteristic detection devices or secondary side measurements collected at the metering cabinet, using thresholds to trigger alarms for out-of-limit data. These online methods often rely on single-point signals or short-time window statistical data, making it difficult to isolate error drift caused by the superposition of multiple factors under complex operating conditions, thus limiting the interpretability of the algorithm results. Currently, while machine learning and deep learning have certain advantages in modeling, they use single-point measurement data or simply splice together data from multiple measurement points. The model may exhibit unexplained biases under unseen operating conditions or noise disturbances, resulting in poor accuracy and interpretability of error assessments. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide a method and system for evaluating the error of high-voltage transformers in new energy power plants, which is used to solve the problem of poor accuracy and interpretability of current voltage transformer error evaluation.

[0006] In a first aspect of the present invention, a method for evaluating the error of a high-voltage voltage transformer in a new energy power station is provided, comprising:

[0007] The voltage phasors of all measurement nodes within a predetermined time period are collected simultaneously at substations and new energy power plants, and the parameters of voltage transformer equipment and line parameters between measurement nodes are obtained to construct a measurement dataset.

[0008] Based on the power system line topology and the voltage phasors, equipment parameters, and line parameters between measurement nodes in the measurement dataset, the corresponding adjacency matrix and node feature matrix are constructed by extracting line impedance and node operating condition features, and the time sequence graph structure dataset is obtained based on the adjacency matrix and node feature matrix.

[0009] Construct a graph neural network and train the graph neural network based on the aforementioned temporal graph structure dataset;

[0010] Input the voltage phasors of each measurement node, and determine the voltage transformer state based on the output of the trained graph neural network and a preset threshold.

[0011] In a second aspect of the present invention, a high-voltage transformer error assessment system for new energy power stations is provided, comprising:

[0012] The data acquisition module is used to simultaneously collect voltage phasors of all measurement nodes within a predetermined time period at substations and new energy power plants, and to obtain voltage transformer equipment parameters and line parameters between measurement nodes to construct a measurement dataset.

[0013] The dataset construction module is used to construct the corresponding adjacency matrix and node feature matrix by extracting line impedance and node operating condition features based on the power system line topology and the voltage phasors, equipment parameters and line parameters between measurement nodes in the measurement dataset. The time-series graph structure dataset is obtained based on the adjacency matrix and node feature matrix.

[0014] The model training module is used to construct a graph neural network and train the graph neural network based on the time-series graph structure dataset;

[0015] The error assessment module is used to input the voltage phasors of each measurement node and, based on the output of the trained graph neural network, determine the state of the voltage transformer by using a preset threshold.

[0016] In a third aspect of the present invention, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein the processor executes the computer program to implement the steps of the method as described in the first aspect of the present invention.

[0017] In a fourth aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the method provided in the first aspect of the present invention.

[0018] In this embodiment of the invention, an adjacency matrix is ​​constructed using system topology and line impedance. Time-series measurement data and equipment parameters are used as node features to construct a time-series graph structure dataset, which is then input into a graph neural network. Through training the graph neural network, the ratio error and phase error estimates of all nodes at the same time are finally output. This not only enables online alarm of voltage transformer errors but also fully utilizes the coupling relationship determined by system topology and line parameters to improve the interpretability and robustness of error assessment in complex electrical scenarios. At the same time, the time-series graph structure dataset constructed based on adjacency coupling strength and node operating condition characteristics can ensure the accuracy of voltage transformer error assessment. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 A flowchart illustrating an error assessment method for high-voltage transformers in new energy power plants, provided as an embodiment of the present invention;

[0021] Figure 2 This is a schematic diagram of the structure of a high-voltage voltage transformer error assessment system for a new energy power station, provided in one embodiment of the present invention.

[0022] Figure 3 This is a schematic diagram of the structure of an electronic device provided in one embodiment of the present invention. Detailed Implementation

[0023] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0024] It should be understood that the terms "comprising" and other similar expressions in the specification, claims, and accompanying drawings of this invention are intended to cover a non-exclusive inclusion, such as a process, method, system, or apparatus that includes a series of steps or units and is not limited to the listed steps or units. Furthermore, "first" and "second" are used to distinguish different objects and are not intended to describe a specific order.

[0025] Please see Figure 1 The present invention provides a flowchart illustrating a method for evaluating the error of high-voltage voltage transformers in new energy power stations, comprising:

[0026] S101. Simultaneously collect the voltage phasors of all measurement nodes within a predetermined time period at the substation and new energy power station, and obtain the voltage transformer equipment parameters and the line parameters between the measurement nodes to construct a measurement dataset.

[0027] The measurement node refers to the location where voltage measurement needs to be performed, i.e., the location where voltage transformers are installed in substations or new energy power plants; the equipment parameters may include equipment type (CVT or PT), years of operation, load conditions, etc., and may also include nameplate parameters of high voltage transformers, such as turns ratio, accuracy class, capacity, etc.; the line parameters are a set of equivalent circuit parameters characterizing the electromagnetic characteristics of power lines, i.e., the line parameters between measurement nodes, including resistance, reactance, conductance, etc.

[0028] The measurement nodes include at least a measurement point, a busbar, and the secondary side of a voltage transformer; the line parameters include at least the resistance and reactance between the measurement nodes.

[0029] In power systems, a measuring point usually refers to the monitoring point of a current transformer (CT) or voltage transformer (PT), used to collect current or voltage signals from the primary system; a busbar refers to the main line in the power system, which undertakes the task of transmitting electrical energy; the secondary side refers to the output terminal of the transformer that converts the high voltage / current on the primary side into a standard signal, mainly used for measurement, protection and control systems.

[0030] At substations and new energy power plants, voltage phasor data of all nodes, as well as high-voltage transformer parameters and line parameters between nodes, can be collected within a predetermined time period T (assuming that the system topology and all line parameters remain unchanged during this time period) to obtain a measurement dataset.

[0031] Discrete time Representing the sampling sequence, in terms of node set This represents the equivalent measurement nodes of the measuring point / busbar / VT secondary side, denoted as . .Will The voltage phasor obtained by measuring at time t is denoted as a complex vector. , of which Voltage of each node .Will The node operating condition characteristics, composed of factors such as transformer type and voltage level, are represented as a real number matrix. Collect system topology and line parameters. Based on the resistance and reactance in the line parameters, the line impedance can be calculated, forming a set of impedance parameters.

[0032] ;

[0033] In the formula, express time, Indicates the first 1 node Indicates the first The impedance between nodes Indicates the first The resistance between nodes Indicates the first Reactance between nodes.

[0034] Therefore, all time-time measurement datasets for: .

[0035] Optionally, the theoretical error value is obtained based on the voltage vector of the measurement node through an error evaluation algorithm, and the theoretical error value is used as the data label of the corresponding sampling time in the measurement dataset to obtain a label dataset. The error evaluation algorithm is an error prediction algorithm based on deep learning.

[0036] The voltage phasor data from the measurement dataset is input into an existing error evaluation algorithm to obtain theoretical error values. These theoretical error values ​​are then used as labels to construct a labeled dataset.

[0037] exist At each time step, output the theoretical ratio and phase error label for each node. This leads to the construction of a dataset of all time-labeled moments. , .

[0038] The error assessment algorithm can be a publicly available transformer error assessment algorithm (e.g., CN118330538A), which uses a large amount of monitoring data from voltage transformers and selects relevant measurement data as samples to directly train a deep learning model to obtain a prediction model, enabling error prediction for voltage transformers. This embodiment optimizes voltage transformer error assessment based on existing error prediction algorithms, not only reducing the training difficulty of subsequent graph neural networks and decreasing the model training workload, but also effectively improving the accuracy of error assessment.

[0039] S102. Based on the power system line topology and the voltage phasors, equipment parameters and line parameters between measurement nodes in the measurement data, the corresponding adjacency matrix and node feature matrix are constructed by extracting line impedance and node operating condition features, and the time sequence graph structure dataset is obtained based on the adjacency matrix and node feature matrix.

[0040] Based on the topology of each line in the power system and the line parameters between nodes, line impedance can be extracted to construct an adjacency matrix to represent the coupling relationship between lines; based on the voltage phasors and equipment parameters of nodes, a node feature matrix can be constructed to represent the operating characteristics of each node.

[0041] Specifically, based on the power system line topology and the line parameters between the measurement nodes in the measurement data, the line impedance is extracted to construct an adjacency matrix, and the adjacency matrix is ​​normalized to obtain a normalized adjacency matrix.

[0042] Based on the time-series voltage phasors and voltage transformer equipment parameters in the measurement dataset, a node feature matrix is ​​constructed by extracting node operating condition features;

[0043] The adjacency matrix and node feature matrix are combined to construct a time-series graph structure dataset.

[0044] The adjacency matrix can characterize the coupling strength between measurement nodes (i.e., between voltage transformers) and can be represented by the line admittance magnitude (admittance and impedance are reciprocals of each other); the node feature matrix can characterize the node operating conditions and can be used for equipment parameters and voltage vector representation. Combining the normalized adjacency matrix and the node feature matrix can yield a time-series graph structure dataset.

[0045] The node feature matrix includes equipment parameters, such as equipment type (CVT or PT), transformer service life, transformer load condition (light load, heavy load), and parameters characterizing equipment attributes from the equipment nameplate, as well as time-series voltage amplitude and phase. These equipment parameters are encoded as state representations; for example, the equipment type can be set with specific values ​​or characters, and the service life can be given a specific value. The final node feature matrix consists of the time-series voltage amplitude, phase, and the aforementioned equipment parameters.

[0046] The node feature matrix is ​​represented as ,node The characteristics are ={ } Selecting time-series voltage characteristics, the dataset at time t includes... The voltage data from time t to time t, the window size is We.

[0047] Constructing an adjacency matrix using line impedance The matrix elements are ,express Time of the first The coupling strength between nodes is then analyzed, and a normalized adjacency matrix is ​​obtained by symmetric normalization. , .

[0048] ;

[0049] In the formula, For degree matrix, , Defined as:

[0050]

[0051] Then in The dataset for constructing a time series graph structure is as follows: .

[0052] S103. Construct a graph neural network and train the graph neural network based on the time-series graph structure dataset;

[0053] Graph neural networks (GNNs) are algorithms based on deep learning that process graph-structured data. They can perform data prediction by extracting features from nodes, edges, and the graph as a whole. GNNs can be trained on time-series graph datasets to predict errors at each node.

[0054] The graph neural network is a stacked GCN / GAT structure, and its loss function consists of label data loss and physical prior loss, expressed as follows:

[0055]

[0056] In the formula, Let t represent the total loss, and T represent the total time. express Error estimates for all nodes at time 1. express The theoretical error value at time 10:00 This represents the loss value calculated using KVL constraints. This represents the sum of the squares of the components. Indicates the weight.

[0057] A graph neural network (GNN) is constructed by using stacked GCNs (Graph Convolutional Networks) / GATs (Graph Attention Networks), and its transfer formula is as follows:

[0058] ;

[0059] ;

[0060] In the formula, Indicates the first Hidden features of the layer Indicates the first The learnable weight matrix of the layer, This represents the activation function, which can be the ReLU function.

[0061] The output of this network is Error estimation of all nodes at time step .

[0062] node The true phasor estimate of the voltage at point A is written under a small error approximation as follows: , express Time Node The true voltage phasor estimate.

[0063] To construct KVL constraints for the obtained voltage true phasor estimate, the first step is to construct... The directed incidence matrix of edges and nodes at different times ( This represents the number of edges. Each edge can be oriented in any direction. The head of the edge is incremented by 1, and the tail is incremented by 1. 1) Then the voltage vector on the edge is given by the node voltage. In the basic loop-edge incidence matrix of the selected loop ( This represents the number of basic loops in the circuit. Each row corresponds to one basic loop. The number of loops is incremented by 1 along the direction of the loop and decremented by 1 in the opposite direction. 1. (Not denoteed as 0 in a loop). Ideally, the equations should satisfy the following KVL constraints:

[0064] ;

[0065] The total loss function can be obtained using tags and electrical constraints: .

[0066] The first term of the loss function represents the sum of squared differences between the error estimate and the error label at all time points, and the second term represents the physical prior loss based on KVL constraints. The weight of this item can be 1.

[0067] The number of training rounds is set to Epoch. If the number of training rounds epoch is less than Epoch, the training is iterated continuously to obtain the trained graph neural network, which is used for node error prediction.

[0068] It is understandable that Kirchhoff's Voltage Law (KVL) and the power / current balance relationship at nodes impose structural constraints on the voltage phasors of nodes / loops; line impedance / admittance parameters determine the coupling strength between different measurement points. This kind of prior knowledge, closely integrated with the model learning process, can suppress unreasonable results under conditions of incomplete data and uncertain operating conditions, improve the model's generalization ability and interpretability, and further ensure the accuracy and reliability of error estimation.

[0069] S104. Input the voltage phasors of each measurement node, and determine the voltage transformer state based on the output of the trained graph neural network and a preset threshold.

[0070] The voltage phasors of each measurement node are acquired in real time and input into a trained graph neural network. The trained graph neural network outputs the ratio error and voltage phase error of each node. Based on the ratio error and voltage phase error of each node output by the graph neural network, and combined with a preset threshold, the state of the voltage transformer is determined.

[0071] Specifically, a first threshold and a second threshold are preset, where the first threshold is a ratio error threshold and the second threshold is a voltage phase error threshold;

[0072] If the ratio error in the output of the graph neural network is less than the first threshold, or the voltage phase error is less than the second threshold, then the voltage transformer error is determined to be within the limit range.

[0073] If the ratio error in the output of the graph neural network is greater than the first threshold but less than twice the first threshold, or the voltage phase error is greater than the second threshold but less than twice the second threshold, then an error warning will be issued for the voltage transformer.

[0074] If the ratio error in the output of the graph neural network is greater than twice the first threshold, or the voltage phase error is greater than twice the second threshold, then the voltage transformer is determined to be out of tolerance.

[0075] Optionally, if the results of the amplitude and phase judgments are inconsistent, the higher priority should be used as the final output result according to the priority principle of "out of tolerance > warning > within limits". For example, if the ratio error in the output result of the graph neural network is greater than twice the first threshold and the voltage phase error is less than twice the second threshold, the output result should be out of tolerance.

[0076] For example, for each time node Error state assessment can be obtained using the following formula:

[0077] ;

[0078] In the formula, Indicates the proportional error threshold. Indicates the voltage phase error threshold. This represents the proportional error of the i-th node at time t. This represents the voltage phase error of the i-th node at time t.

[0079] In this embodiment, an adjacency matrix is ​​constructed using the system topology and line impedance. Time-series measurement data and equipment parameters are used as node features to construct a time-series graph structure dataset, which is then input into a graph neural network. During training, error labels obtained from existing error assessment algorithms are used as constraints, combined with Kirchhoff's Voltage Law (KVL) constraints, to train the graph neural network model. This not only enables online voltage transformer error estimation but also provides strong interpretability and robustness, while ensuring the accuracy and reliability of the error assessment results. Compared to traditional offline detection and timed detection schemes, this approach offers lower implementation costs, higher real-time performance, and wider coverage.

[0080] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0081] Figure 2 A schematic diagram of a high-voltage voltage transformer error assessment system for new energy power stations, provided as an embodiment of the present invention, is shown. The system includes:

[0082] The data acquisition module 210 is used to simultaneously acquire the voltage phasors of all measurement nodes within a predetermined time period in substations and new energy power plants, and to obtain the voltage transformer equipment parameters and the line parameters between measurement nodes to construct a measurement dataset.

[0083] The measurement nodes include at least a measurement point, a busbar, and the secondary side of a voltage transformer; the line parameters include at least the resistance and reactance between the measurement nodes.

[0084] Optionally, the construction of the measurement dataset further includes:

[0085] The theoretical error value is obtained based on the voltage vector of the measurement node through an error evaluation algorithm, and the theoretical error value is used as the data label of the corresponding sampling time in the measurement dataset to obtain a label dataset. The error evaluation algorithm is an error prediction algorithm based on deep learning.

[0086] The dataset construction module 220 is used to construct the corresponding adjacency matrix and node feature matrix by extracting line impedance and node operating condition features based on the power system line topology and the voltage phasors, equipment parameters and line parameters between measurement nodes in the measurement dataset, and to obtain the time sequence graph structure dataset based on the adjacency matrix and node feature matrix.

[0087] The dataset construction module 220 includes:

[0088] The first construction unit is used to extract line impedance and construct an adjacency matrix based on the power system line topology and the line parameters between the measurement nodes in the measurement data. The adjacency matrix is ​​then normalized to obtain a normalized adjacency matrix.

[0089] The second construction unit is used to construct a node feature matrix by extracting node operating condition features based on the time-series voltage phasors and voltage transformer equipment parameters in the measurement dataset.

[0090] The combination building unit is used to combine the adjacency matrix and the node feature matrix to construct a time-series graph structure dataset.

[0091] The model training module 230 is used to construct a graph neural network and train the graph neural network based on the time-series graph structure dataset;

[0092] Optionally, the graph neural network is a stacked GCN / GAT structure, and the loss function of the graph neural network is the label data loss and the physical prior loss, expressed as follows:

[0093]

[0094] In the formula, Let t represent the total loss, and T represent the total time. express Error estimates for all nodes at time 1. express The theoretical error value at time 10:00 This represents the loss value calculated using KVL constraints. This represents the sum of the squares of the components. Indicates the weight.

[0095] The error assessment module 240 is used to input the voltage phasors of each measurement node and, based on the output of the trained graph neural network, determine the state of the voltage transformer by means of a preset threshold.

[0096] Optionally, determining the voltage transformer state based on the output of the trained graph neural network and a preset threshold includes:

[0097] A first threshold and a second threshold are preset, wherein the first threshold is a ratio error threshold and the second threshold is a voltage phase error threshold;

[0098] If the ratio error in the output of the graph neural network is less than the first threshold, or the voltage phase error is less than the second threshold, then the voltage transformer error is determined to be within the limit range.

[0099] If the ratio error in the output of the graph neural network is greater than the first threshold but less than twice the first threshold, or the voltage phase error is greater than the second threshold but less than twice the second threshold, then an error warning will be issued for the voltage transformer.

[0100] If the ratio error in the output of the graph neural network is greater than twice the first threshold, or the voltage phase error is greater than twice the second threshold, then the voltage transformer is determined to be out of tolerance.

[0101] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the system and modules described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0102] Figure 3 This is a schematic diagram of an electronic device according to an embodiment of the present invention. The electronic device is used for voltage transformer error assessment. Figure 3 As shown, the electronic device 3 in this embodiment includes: a memory 310, a processor 320, and a system bus 330. The memory 310 includes an executable program 3101 stored thereon. As those skilled in the art will understand, Figure 3 The electronic device structure shown does not constitute a limitation on the electronic device and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0103] The following is combined Figure 3 A detailed introduction to each component of the electronic device:

[0104] The memory 310 can be used to store software programs and modules. The processor 320 executes various functional applications and data processing of the electronic device by running the software programs and modules stored in the memory 310. The memory 310 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, application programs required for at least one function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the electronic device (such as cached data), etc. In addition, the memory 310 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0105] The memory 310 contains an executable program 3101 for error evaluation. This executable program 3101 can be divided into one or more modules / units, which are stored in the memory 310 and executed by the processor 320 to achieve online evaluation of voltage transformer errors, etc. The one or more modules / units can be a series of computer program instruction segments capable of performing specific functions, describing the execution process of the executable program 3101 in the electronic device 3. For example, the executable program 3101 can be divided into functional modules such as a data acquisition module, a dataset construction module, a model training module, and an error evaluation module.

[0106] The processor 320 is the control center of the electronic device. It connects various parts of the electronic device via various interfaces and lines. By running or executing software programs and / or modules stored in the memory 310, and by calling data stored in the memory 310, it performs various functions and processes data, thereby monitoring the overall status of the electronic device. Optionally, the processor 320 may include one or more processing units; preferably, the processor 320 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, application programs, etc., and the modem processor mainly handles wireless communication. It is understood that the modem processor may also not be integrated into the processor 320.

[0107] The system bus 330 is used to connect various functional components within the computer, transmitting data, address, and control information. Its type can be, for example, a PCI bus, an ISA bus, or a CAN bus. Instructions from the processor 320 are transmitted to the memory 310 via the bus, and the memory 310 sends data back to the processor 320. The system bus 330 is responsible for data and instruction exchange between the processor 320 and the memory 310. Of course, the system bus 330 can also connect to other devices, such as network interfaces and display devices.

[0108] In this embodiment of the invention, the executable program executed by the processor 320 included in the electronic device includes:

[0109] The voltage phasors of all measurement nodes within a predetermined time period are collected simultaneously at substations and new energy power plants, and the parameters of voltage transformer equipment and line parameters between measurement nodes are obtained to construct a measurement dataset.

[0110] Based on the power system line topology and the voltage phasors, equipment parameters, and line parameters between measurement nodes in the measurement dataset, the corresponding adjacency matrix and node feature matrix are constructed by extracting line impedance and node operating condition features, and the time sequence graph structure dataset is obtained based on the adjacency matrix and node feature matrix.

[0111] Construct a graph neural network and train the graph neural network based on the aforementioned temporal graph structure dataset;

[0112] Input the voltage phasors of each measurement node, and determine the voltage transformer state based on the output of the trained graph neural network and a preset threshold.

[0113] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0114] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0115] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for evaluating the error of high-voltage voltage transformers in new energy power stations, characterized in that, include: The voltage phasors of all measurement nodes within a predetermined time period are collected simultaneously at substations and new energy power plants, and the parameters of voltage transformer equipment and line parameters between measurement nodes are obtained to construct a measurement dataset. Based on the power system line topology and the voltage phasors, equipment parameters, and line parameters between measurement nodes in the measurement dataset, the corresponding adjacency matrix and node feature matrix are constructed by extracting line impedance and node operating condition features, and the time sequence graph structure dataset is obtained based on the adjacency matrix and node feature matrix. Among them, based on the power system line topology and the line parameters between the measurement nodes in the measurement data, the line impedance is extracted to construct the adjacency matrix, and the adjacency matrix is ​​normalized to obtain the normalized adjacency matrix. Based on the time-series voltage phasors and voltage transformer equipment parameters in the measurement dataset, a node feature matrix is ​​constructed by extracting node operating condition features; A temporal graph structure dataset is constructed by combining the adjacency matrix and the node feature matrix; The adjacency matrix represents the coupling strength between the measured nodes, and the node feature matrix represents the node operating condition. Construct a graph neural network and train the graph neural network based on the aforementioned temporal graph structure dataset; The graph neural network is a stacked GCN / GAT structure, and its loss function consists of label data loss and physical prior loss, expressed as follows: In the formula, Let t represent the total loss, and T represent the total time. express Error estimates for all nodes at time 1. express The theoretical error value at time 10:00 This represents the loss value calculated using KVL constraints. This represents the sum of the squares of the components. Indicates weight; Among them, the voltage true value phasor estimation To construct KVL constraints, first construct the edge-node directed incidence matrix at time t. In the formula, M represents the number of edges. Each edge can be chosen in any direction, with the head of the edge marked as +1 and the tail as -1. Then, the voltage phasor on the edge is given by the node voltage. In the basic loop-edge incidence matrix of the selected loop In the formula, Let $\mathbf{ ... ; Input the voltage phasors of each measurement node, and determine the voltage transformer state based on the output of the trained graph neural network and a preset threshold.

2. The method according to claim 1, characterized in that, The measurement node includes at least a measuring point, a busbar, and the secondary side of a voltage transformer; The line parameters include at least the resistance and reactance between the measured nodes.

3. The method according to claim 1, characterized in that, The construction of the measurement dataset also includes: The theoretical error value is obtained from the voltage phasor of the measurement node through an error evaluation algorithm, and the theoretical error value is used as the data label of the corresponding sampling time in the measurement dataset to obtain the label dataset. The error evaluation algorithm is an error prediction algorithm based on deep learning.

4. The method according to claim 1, characterized in that, The method of determining the voltage transformer state based on the output of the trained graph neural network and a preset threshold includes: A first threshold and a second threshold are preset, wherein the first threshold is a ratio error threshold and the second threshold is a voltage phase error threshold; If the ratio error in the output of the graph neural network is less than the first threshold, or the voltage phase error is less than the second threshold, then the voltage transformer error is determined to be within the limit range. If the ratio error in the output of the graph neural network is greater than the first threshold but less than twice the first threshold, or the voltage phase error is greater than the second threshold but less than twice the second threshold, then an error warning will be issued for the voltage transformer. If the ratio error in the output of the graph neural network is greater than twice the first threshold, or the voltage phase error is greater than twice the second threshold, then the voltage transformer is determined to be out of tolerance.

5. A system for evaluating the error of high-voltage voltage transformers in new energy power stations, characterized in that, include: The data acquisition module is used to simultaneously collect voltage phasors of all measurement nodes within a predetermined time period at substations and new energy power plants, and to obtain voltage transformer equipment parameters and line parameters between measurement nodes to construct a measurement dataset. The dataset construction module is used to construct the corresponding adjacency matrix and node feature matrix by extracting line impedance and node operating condition features based on the power system line topology and the voltage phasors, equipment parameters and line parameters between measurement nodes in the measurement dataset. The time-series graph structure dataset is obtained based on the adjacency matrix and node feature matrix. Among them, based on the power system line topology and the line parameters between the measurement nodes in the measurement data, the line impedance is extracted to construct the adjacency matrix, and the adjacency matrix is ​​normalized to obtain the normalized adjacency matrix. Based on the time-series voltage phasors and voltage transformer equipment parameters in the measurement dataset, a node feature matrix is ​​constructed by extracting node operating condition features; A temporal graph structure dataset is constructed by combining the adjacency matrix and the node feature matrix; The adjacency matrix represents the coupling strength between the measured nodes, and the node feature matrix represents the node operating condition. The model training module is used to construct a graph neural network and train the graph neural network based on the time-series graph structure dataset; The graph neural network is a stacked GCN / GAT structure, and its loss function consists of label data loss and physical prior loss, expressed as follows: In the formula, Let t represent the total loss, and T represent the total time. express Error estimates for all nodes at time 1. express The theoretical error value at time 10:00 This represents the loss value calculated using KVL constraints. This represents the sum of the squares of the components. Indicates weight; Among them, the voltage true value phasor estimation To construct KVL constraints, first construct the edge-node directed incidence matrix at time t. In the formula, M represents the number of edges. Each edge can be chosen in any direction, with the head of the edge marked as +1 and the tail as -1. Then, the voltage phasor on the edge is given by the node voltage. In the basic loop-edge incidence matrix of the selected loop In the formula, Let $\mathbf{ ... ; The error assessment module is used to input the voltage phasors of each measurement node and, based on the output of the trained graph neural network, determine the state of the voltage transformer by using a preset threshold.

6. The system according to claim 5, characterized in that, The construction of the measurement dataset also includes: The theoretical error value is obtained from the voltage phasor of the measurement node through an error evaluation algorithm, and the theoretical error value is used as the data label of the corresponding sampling time in the measurement dataset to obtain the label dataset. The error evaluation algorithm is an error prediction algorithm based on deep learning.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the error assessment method for high-voltage transformers in new energy power stations as described in any one of claims 1 to 4.

8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed, it implements the steps of the error assessment method for high-voltage voltage transformers in new energy power stations as described in any one of claims 1 to 4.

Images