Graph neural network-based state estimation method and apparatus, device and medium

Abstract

A graph neural network-based state estimation method and apparatus, a device and a medium. The method comprises: constructing a bus branch graph of a power system, and calculating the amount of unbalance of active power of at least one bus node (S110); inputting into a graph neural network the bus branch graph and the amount of unbalance of active power of each bus node, and determining a respective equivalent unit configuration result of each bus (S120); configuring equivalent units for the buses on the basis of the equivalent unit configuration results, and acquiring the power of devices in the power system (S130); and estimating the states of the devices in the power system according to the power of the devices, and calculating estimated device voltage values of the devices (S140).

Description

State estimation method, device, equipment and medium based on graph neural network

[0001] Related applications

[0002] This application claims priority to the Chinese patent application filed with the China Patent Office on October 30, 2023, with application number 2023114271597 and application name “State estimation method, device, equipment and medium based on graph neural network”, all contents of which are incorporated by reference into this application. Technical Field

[0003] The present application relates to the field of artificial intelligence technology, and in particular to a state estimation method, apparatus, device, and medium based on a graph neural network. Background Art

[0004] In power system state estimation, measurement data errors often have a serious impact on the convergence of state estimation. To address this issue, during state estimation debugging, adding equivalent units to buses with unbalanced active power measurements is often used to improve the convergence of state estimation.

[0005] Currently, it is usually determined based solely on the experience of the commissioning personnel on which buses to add equivalent units and how to optimize the configuration of equivalent units.

[0006] The traditional configuration method of equivalent units has low configuration efficiency and can only solve the local problem of state estimation convergence. It cannot globally solve the problem of active power measurement imbalance in the power system, and the convergence of state estimation is still difficult to guarantee.

[0007] Summary of the Invention

[0008] According to various embodiments disclosed in the present application, a state estimation method, apparatus, device, and medium based on a graph neural network are provided.

[0009] According to one aspect of the present application, a state estimation method based on a graph neural network is provided, the method comprising:

[0010] Constructing a busbar branch diagram of the power system and calculating the active power imbalance of at least one busbar node;

[0011] The bus branch diagram and the active power imbalance of each bus node are input into the graph neural network to determine the equivalent unit configuration results of each bus;

[0012] Based on the equivalent unit configuration results, configure the equivalent unit for the bus and obtain the equipment power in the power system; and

[0013] Based on the equipment power, the state of the equipment in the power system is estimated and the equipment voltage estimation value of the equipment is calculated.

[0014] According to another aspect of the present application, a state estimation device based on a graph neural network is provided, the device comprising:

[0015] A bus branch diagram construction module is used to construct a bus branch diagram of the power system and calculate the active power imbalance of at least one bus node;

[0016] The equivalent unit configuration result determination module is used to input the bus branch diagram and the active power imbalance of each bus node into the graph neural network to determine the equivalent unit configuration result of each corresponding bus;

[0017] The equipment power acquisition module is used to configure the equivalent units for the bus based on the equivalent unit configuration results and obtain the equipment power in the power system; and

[0018] The state estimation module is used to estimate the state of the equipment in the power system according to the equipment power and calculate the equipment voltage estimation value of the equipment.

[0019] According to another aspect of the present application, an electronic device is provided, the electronic device comprising:

[0020] at least one processor; and

[0021] A memory communicatively connected to at least one processor; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor performs the following steps when executed:

[0022] Constructing a busbar branch diagram of the power system and calculating the active power imbalance of at least one busbar node;

[0023] Inputting the bus branch diagram and the active power imbalance of each bus node into the graph neural network to determine the equivalent unit configuration results of each corresponding bus;

[0024] Based on the equivalent unit configuration result, configure the equivalent unit for the bus and obtain the equipment power in the power system; and

[0025] According to the device power, a state estimation is performed on the device in the power system, and an estimated device voltage value of the device is calculated.

[0026] According to another aspect of the present application, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to cause a processor to implement the following steps when executed:

[0027] Constructing a busbar branch diagram of the power system and calculating the active power imbalance of at least one busbar node;

[0028] Inputting the bus branch diagram and the active power imbalance of each bus node into the graph neural network to determine the equivalent unit configuration results of each corresponding bus;

[0029] Based on the equivalent unit configuration result, configure the equivalent unit for the bus and obtain the equipment power in the power system; and

[0030] According to the device power, a state estimation is performed on the device in the power system, and an estimated device voltage value of the device is calculated.

[0031] The details of one or more embodiments of the present application are set forth in the accompanying drawings and the description below. Other features and advantages of the present application will become apparent from the description, drawings, and claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0033] FIG1 is a flowchart of a state estimation method based on a graph neural network according to one or more embodiments;

[0034] FIG2 is a diagram illustrating an information transfer process of a graph neural network according to one or more embodiments;

[0035] FIG3 is a flowchart of a state estimation method based on a graph neural network according to one or more embodiments;

[0036] FIG4 is a busbar branch diagram of a power system according to one or more embodiments;

[0037] FIG5 is a schematic diagram showing the equivalent unit configuration results of each busbar determined by the neural network in one or more embodiments;

[0038] FIG6 is a schematic diagram of the structure of a state estimation device based on a graph neural network according to one or more embodiments;

[0039] Figure 7 is a structural diagram of an electronic device according to a state estimation method based on a graph neural network in one or more embodiments. DETAILED DESCRIPTION

[0040] In order to make the technical solutions and advantages of this application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0041] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in a sequence other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0042] Figure 1 is a flowchart of a graph neural network-based state estimation method provided according to some embodiments of the present application. Embodiments of the present application are applicable to situations where state estimation of power systems is performed based on graph neural networks. The method can be performed by a graph neural network-based state estimation device, which can be implemented in hardware and / or software. The graph neural network-based state estimation device can be configured in an electronic device that carries the graph neural network-based state estimation function.

[0043] Referring to FIG1 , the state estimation method based on graph neural network includes:

[0044] S110: Construct a busbar branch diagram of the power system, and calculate the active power imbalance of at least one busbar node.

[0045] The bus branch diagram can be used to characterize the equipment connection relationship between each bus in the power system. The bus branch diagram can reflect the distribution network structure in the power system. Optionally, the bus branch diagram can be used to characterize the equipment connection relationship between each bus in the power system in real time. The bus branch diagram can include bus nodes and branches. When performing state estimation on the power system, each bus node needs to maintain active measurement balance, which can be understood as requiring each bus to maintain active power balance. Bus active power balance can be understood as the difference between the active power of the equipment output by the bus and the active power of the equipment input to the bus node is zero. Bus active power imbalance can be understood as the difference between the active power of the equipment output by the bus and the active power of the equipment input to the bus is not zero. The active power imbalance of the bus node can be the difference between the active power of the equipment input to the bus and the active power of the equipment output by the bus.

[0046] Specifically, the device connection relationships within the power system can be acquired in real time. Based on these device connection relationships, the device connection relationships between each bus in the power system and between each bus can be determined, generating a bus branch diagram for the power system. The active power of the devices input to the bus and the active power of the devices output from the bus are acquired, and the difference between the active power of the devices input to the bus and the output power of the devices output from the bus is calculated to obtain the active power imbalance at each bus node.

[0047] S120: Input the bus branch diagram and the active power imbalance of each bus node into the graph neural network to determine the equivalent unit configuration results of each corresponding bus.

[0048] Graph neural networks can be used to determine whether a busbar in a power system is configured with an equivalent generator set. Graph neural networks can be pre-trained. Compared to traditional neural networks (e.g., convolutional neural networks), graph neural networks can process real-time measurement data in power systems and take into account complex distribution network structures. The equivalent generator set configuration result can indicate whether the busbar needs to be configured with an equivalent generator set. For example, the equivalent generator set configuration result may include configuring an equivalent generator set or not configuring an equivalent generator set.

[0049] Specifically, the active power imbalance at each bus node can be used as the attribute information of each bus node in the bus branch diagram. The bus branch diagram and the active power imbalance at each bus node are input into the graph neural network, and the equivalent unit configuration results for each corresponding bus are output.

[0050] S130. Based on the equivalent unit configuration result, configure the equivalent unit for the bus and obtain the equipment power in the power system.

[0051] Equivalent units can be used to achieve active power balance on the corresponding bus. Optionally, the equivalent unit configuration result may also include the type of equivalent unit and the active power of the equivalent unit. Specifically, when the active power imbalance on the bus is greater than zero, the type of equivalent unit may be a load unit; when the active power imbalance on the bus is less than zero, the type of equivalent unit may be a generator unit. Accordingly, the active power of the equivalent unit can be determined based on the active power imbalance on each bus. Equipment power may be a power measurement of equipment in the power system. Exemplarily, equipment power may include equipment active power and equipment reactive power. Compared to when no equivalent units are configured on the bus, the bus may have an active power imbalance, and the measured equipment power may be inaccurate. After configuring the bus with equivalent units, the equipment power in the power system is obtained, and the measured equipment power is more accurate. Accordingly, when estimating the state of the power system, the estimated equipment voltage is also more accurate.

[0052] Specifically, if the equivalent unit configuration result is to configure an equivalent unit, then the equivalent unit is configured for the corresponding bus; if the equivalent unit configuration result is not to configure an equivalent unit, then the equivalent unit is not configured for the corresponding bus. After the equivalent unit configuration is completed for each bus in the power system, the equipment power of the power system is obtained.

[0053] S140 . Estimating the state of the equipment in the power system according to the equipment power, and calculating an estimated value of the equipment voltage of the equipment.

[0054] The estimated device voltage can be an estimate of the device voltage when the busbar active power is balanced in the power system. Due to errors or duplication in measurement data in power systems, the accuracy of device voltage measurements is low. However, by estimating the device state in the power system and determining the estimated device voltage, the obtained device voltage estimate is more accurate than the measured device voltage, facilitating subsequent power system analysis.

[0055] Specifically, the state of the equipment in the power system can be estimated based on the equipment power, and the equipment voltage estimation value of the equipment can be calculated.

[0056] In power system state estimation, poor data and network topology connectivity errors often severely impact state estimation convergence. This poor data can include erroneous or duplicated measurements. To address this issue, during state estimation commissioning, adding equivalent units to buses experiencing active power imbalance (i.e., active power imbalance) can improve state estimation convergence. However, the decision on which buses to add equivalent units and how to optimize their configuration often relies solely on the commissioning personnel's experience, lacking a systematic and efficient optimization solution. However, state estimation methods based on commissioning personnel's experience often suffer from inefficient configuration of equivalent units. The configuration of equivalent units often only addresses local state estimation convergence issues and fails to globally address active power imbalance within the power system. Deep learning methods using artificial intelligence are a common solution to these complex problems. However, traditional convolutional neural networks (CNNs) cannot process the real-time changing measurement data in power systems and cannot account for the complex structure of distribution networks.

[0057] The technical solution of the embodiment of the present application constructs a bus branch diagram of the power system and calculates the active power imbalance of at least one bus in the power system, inputs the bus branch diagram and the active power imbalance of each bus into a graph neural network, determines the equivalent unit configuration result of each bus, configures the bus with an equivalent unit based on the equivalent unit configuration result, obtains the equipment power in the power system, performs state estimation on the equipment in the power system according to the equipment power, calculates the equipment voltage estimation value of the equipment, and utilizes the graph neural network to adapt to the characteristics of real-time changes and complex structures of data in the power system, determines the equivalent unit configuration result of the power system, and solves the low configuration efficiency of the traditional equivalent unit configuration method, which can only solve the local problem of state estimation convergence and cannot globally solve the problem of active measurement imbalance in the system, and the convergence of state estimation is still difficult to guarantee, thereby improving the configuration efficiency of the equivalent unit, globally solving the problem of active measurement imbalance, and ensuring the convergence of state estimation.

[0058] In some embodiments of the present application, the bus branch diagram and the active power imbalance of each bus node are input into the graph neural network to determine the equivalent unit configuration results of each corresponding bus, including: inputting the bus node branch diagram and the active power imbalance of each bus node into the graph neural network; through the first convolution layer of the graph neural network, for each bus node, the active power imbalance of the bus node and the active power imbalance of other bus nodes connected to the bus node are weighted averaged to obtain the weighted average result of the first convolution layer; through the first convolution layer of the graph neural network, the product between the weighted average result of the first convolution layer and the weight matrix coefficient of the first convolution layer is calculated to obtain the feature data output by the bus node in the first convolution layer; through other convolution layers of the graph neural network, for each bus node, the bus node in the previous convolution is weighted averaged. The characteristic data output by the layer and the characteristic data output by other bus nodes connected to the bus node in the previous convolution layer are weighted averaged to obtain the weighted average results of other convolution layers; through other convolution layers of the graph neural network, the product between the weighted average results of other convolution layers and the weight matrix coefficients of other convolution layers is calculated to obtain the characteristic data output by the bus node in other convolution layers, until the characteristic data output by the last convolution layer is obtained; the characteristic data output by the last convolution layer is compared with the preset characteristic data threshold; when the characteristic data output by the last convolution layer is greater than or equal to the preset characteristic data threshold, it is determined that the equivalent unit configuration result of the corresponding bus is configured with an equivalent unit; when the characteristic data output by the last convolution layer is less than the preset characteristic data threshold, it is determined that the equivalent unit configuration result of the corresponding bus is not configured with an equivalent unit.

[0059] A graph neural network (GNN) can include multiple convolutional layers (i.e., information transfer layers). Each convolutional layer implements the graph convolution process by transferring data, thereby embedding the data's pattern features into the neurons of each convolutional layer, i.e., each bus node in the bus branch graph. Training a graph neural network essentially involves adjusting the weight matrix coefficients of the convolutional layer to obtain an optimized graph neural network. The weighted average result can be used to aggregate the feature data of the bus node and other bus nodes connected to it. This allows the distribution network structure to be taken into account and the features of other bus nodes connected to it to be learned. The weight matrix coefficients can be used to transfer information about the feature data. The feature data can be the output data of each convolutional layer. It can be understood that the feature data is the feature extraction result of the convolutional layer. Preset feature data thresholds can be used to determine the equivalent unit configuration results for the corresponding bus.

[0060] Specifically, the bus node branch diagram and the active power imbalance of each bus node can be input into the graph neural network. Through the first convolutional layer of the graph neural network, for each bus node, the active power imbalance of the bus node and the active power imbalance of other bus nodes connected to the bus node are weighted averaged to obtain the weighted average result of the first convolutional layer. Through the first convolutional layer of the graph neural network, the product of the weighted average result of the first convolutional layer and the weight matrix coefficient of the first convolutional layer can be calculated to obtain the feature data output by the bus node in the first convolutional layer. Through other convolutional layers of the graph neural network, for each bus node, the feature data output by the bus node in the previous convolutional layer and the feature data output by other bus nodes connected to the bus node in the previous convolutional layer can be weighted averaged to obtain the weighted average results of other convolutional layers. Through other convolutional layers of the graph neural network, the product of the weighted average results of other convolutional layers and the weight matrix coefficients of other convolutional layers can be calculated to obtain the feature data output by the bus node in other convolutional layers, until the feature data output by the last convolutional layer is obtained. The feature data output by the last convolution layer can be compared with a preset feature data threshold. When the feature data output by the last convolution layer is greater than or equal to the preset feature data threshold, the equivalent unit configuration result of the corresponding bus is determined to be configured with an equivalent unit; when the feature data output by the last convolution layer is less than the preset feature data threshold, the equivalent unit configuration result of the corresponding bus is determined to be not configured with an equivalent unit.

[0061] For example, Figure 2 is a diagram of the information transmission process of a graph neural network. As shown in Figure 2, the bus node 1 has the characteristic data of the kth layer. Combined with the characteristic data of busbar nodes 2, 3, and 4 connected to it The characteristic data of busbar node 1 at layer k+1 is formed by aggregation The convolutional layer of the graph neural network learns the characteristics of the data by weighting the bus node and other bus nodes connected to the bus node through the weight matrix coefficient.

[0062] Taking busbar node v as an example, the following formula can be used to calculate the characteristic data of busbar node v at layer k+1:

[0063] in, is the characteristic data of busbar node v at the k+1th layer; is the characteristic data of busbar node v at the kth layer; ω is the busbar node connected to busbar node v; N(v) is the set of other busbar nodes connected to busbar node v; Aggregate is the characteristic data of other bus nodes connected to bus node v; (k) Characteristic data for the busbar node and other busbar nodes connected to it The weighted average result of Update (k) The feature data of the kth layer is convolved and passed to the k+1th layer of the graph neural network by multiplying the weight matrix coefficient system of the graph neural network.

[0064] The following formula can also be used to express the characteristic data of busbar node v at layer k+1: The calculation process:

[0065] in, is the characteristic data of busbar node v at the k+1 layer; W k+1 is the weight matrix coefficient transferred from the kth layer to the k+1th layer; ω is the busbar node connected to the busbar node v; N(v) is the set of other busbar nodes connected to the busbar node v; is the characteristic data of other busbar nodes connected to busbar node v; c ω,v is the normalization coefficient of other busbar nodes connected to the busbar node.

[0066] The output layer of the graph neural network can output the equivalent unit configuration result Y of each bus. Taking the bus node v as an example, the output layer outputs the equivalent unit configuration result Y of the corresponding bus v The feature data of the output of the last layer of the convolution layer, that is, the Nth convolution layer, can be obtained. Decide.

[0067] The following formula can be used to express the equivalent unit configuration result Yv of the bus corresponding to the bus node v:

[0068] Among them, Yv =1 means that the equivalent unit configuration result of the bus corresponding to the bus node v is the equivalent unit configuration; v =0 can indicate that the equivalent unit configuration result of the bus corresponding to bus node v is no equivalent unit configuration; Represents the feature data of the output of the Nth convolutional layer; ε is the preset feature data threshold.

[0069] This solution uses a graph neural network to determine the equivalent unit configuration results of the bus, realizes the processing of real-time changing measurement data in the power system, takes into account the complex distribution network structure, and further improves the efficiency and accuracy of the equivalent unit configuration results.

[0070] In some embodiments of the present application, while configuring equivalent units for the bus based on the equivalent unit configuration results and obtaining the equipment power in the power system, it also includes: obtaining the equipment voltage measurement value in the power system; after performing state estimation on the equipment in the power system and calculating the equipment voltage estimation value of the equipment, it includes: calculating the equipment voltage difference between the equipment voltage estimation value and the equipment voltage measurement value; when the equipment voltage difference is greater than or equal to a preset voltage difference threshold, issuing a measurement error message of the equipment to prompt that there is an error in the equipment power, equipment voltage measurement value or equipment connection relationship corresponding to the equipment.

[0071] The device voltage measurement value may be the actual voltage measurement value of the device in the power system. The device voltage estimation value may be the estimated voltage value of the device in the power system. The device voltage difference value may be the difference between the device voltage estimation value and the device voltage measurement value. The preset voltage difference value threshold may be a preset maximum value of the device voltage difference value. The preset voltage difference value threshold may be used to characterize the accuracy of the device voltage estimation value. The measurement error information may be used to indicate an error in the measurement data of the device in the power system. Optionally, the measurement error information may be in the form of voice, text, or graphics. Exemplarily, the measurement error information may be "There is a measurement error in the measurement data of XX device in the power system!", wherein the measurement data may include device power, device voltage measurement value, or device connection relationship.

[0072] This solution estimates the state of the equipment in the power system, calculates the equipment voltage estimate of the equipment, and then calculates the equipment voltage difference between the equipment voltage estimate and the equipment voltage measurement value. When the equipment voltage difference is greater than or equal to a preset voltage difference threshold, it issues a measurement error message for the equipment to prompt that there is an error in the equipment power, equipment voltage measurement value or equipment connection relationship corresponding to the equipment. This realizes the feedback of the measurement data of the equipment in the power system, and then realizes the correction of the equipment and equipment connection relationship in the power system, further improving the accuracy of the state estimation of the power system.

[0073] Figure 3 is a flow chart of a state estimation method based on a graph neural network provided according to some embodiments of the present application. Based on the above embodiments, the embodiments of the present application concretize "constructing a bus branch diagram of the power system" as "obtaining the device connection relationship of the equipment in the power system; according to the device connection relationship, at least one node device belonging to the same bus and the same bus are determined as a bus node; the branch equipment connected to the bus is determined as a branch; according to the device connection relationship, the bus node and the connected branch are connected to generate a bus branch diagram of the power system", which realizes the real-time acquisition of device data in the power system, takes into account the complex structure of the power system, improves the construction efficiency and accuracy of the bus branch diagram, and enhances the comprehensiveness of the data on which the state estimation of the power system depends. It should be noted that for the parts not described in detail in the embodiments of the present application, please refer to the description of other embodiments.

[0074] Referring to FIG3 , the state estimation method based on graph neural network includes:

[0075] S310: Obtain device connection relationships of devices in the power system.

[0076] Equipment in a power system can include busbars, node devices, and branch devices. Equipment connection relationships can be defined as the connection relationships between busbars, node devices, and branch devices. Equipment connection relationships can be defined as connected or connected.

[0077] Specifically, the device connection relationship of devices in the power system can be obtained in real time.

[0078] S320. According to the device connection relationship, at least one node device belonging to the same bus and the same bus are determined as a bus node.

[0079] A busbar can be used to power devices in a power system. A busbar can be connected to node devices or branch devices. Node devices belonging to the same busbar have the same voltage. Node devices belonging to the same busbar can be connected in parallel on the same busbar. Busbars can be connected to each other via branch devices. The voltages of different busbars can be the same or different. For example, a node device can include at least one of a generator set, a load set, a capacitor, a reactor, and a compensator.

[0080] Specifically, at least one node device connected in parallel to the same bus can be determined as at least one node device belonging to the same bus based on the device connection relationship. At least one node device belonging to the same bus and the same bus can be jointly determined as a bus node.

[0081] S330. Determine the branch equipment connected to the busbar as a branch.

[0082] Branch circuit equipment can be used to connect busbars. Branch circuit equipment can include branch circuit equipment that is connected to the busbar and branch circuit equipment that is disconnected from the busbar. Exemplarily, the branch circuit equipment can include at least one of a transmission line, a transformer, and a switch.

[0083] Specifically, the branch devices connected to the busbar can be determined as branches in the busbar branch diagram.

[0084] S340. Connect the bus nodes and connected branches according to the device connection relationship, generate a bus branch diagram of the power system, and calculate the active power imbalance of at least one bus in the power system.

[0085] Specifically, busbar nodes and corresponding connected branches can be connected based on device connection relationships to generate a busbar branch diagram for the power system. The active power of devices input to the busbar and the output power of devices output from the busbar can be obtained, and the difference between the active power of devices input to the busbar and the output power of devices output from the busbar can be calculated to obtain the active power imbalance of at least one bus in the power system.

[0086] S350: Input the bus branch diagram and the active power imbalance of each bus node into the graph neural network to determine the equivalent unit configuration results of each corresponding bus.

[0087] S360: Based on the equivalent unit configuration result, configure the equivalent unit for the bus and obtain the equipment power in the power system.

[0088] S370: Estimating the state of the equipment in the power system according to the equipment power, and calculating the estimated equipment voltage value of the equipment.

[0089] Optionally, the distribution network model file in the power system can be analyzed. The distribution network equipment model can be established in real time in the graph database. The distribution network equipment model can be used to store the measurement data of the distribution network. Among them, the node branch diagrams of devices such as the transmission line model, transformer model, busbar model (BUS), generator set model (Unit), load unit model (Load), capacitor model, reactance model, compensator model and switch model can be established in the graph database respectively. Exemplarily, the transmission line model can be an AC transmission line model (ACline_dot). The transformer model can be a three-winding transformer model (Three_Port_Transformer) or a two-set transformer model (Two_Port_Transformer). The capacitor model and the reactor model can be a shunt capacitor reactor model (C_P). The compensator model can be a series compensator model (C_S). The switch model can include a switch (Breaker) and a disconnector (Disconnector). In the graph database, the node branch graph G(V,E) of the power system can be defined by nodes (vertex, V) and edges (Edge, E). By defining devices such as generator sets, load units, capacitors, reactors and compensators as nodes V; and defining transmission lines, transformers and switches as edges E, a node branch graph consisting of nodes V and edges E can be generated. At least one node belonging to the same bus and the same bus can be determined as a bus node, and a bus branch graph can be further generated. For example, Figure 4 is a bus branch graph of a power system. As shown in Figure 4, the bus branch graph includes 5 bus nodes. Each bus node in the figure represents a bus, and each edge in the figure reflects the equipment connection relationship between the buses. The active power imbalance of the bus can be stored as parameter information in the attribute information of the bus node and branch. Among them, data such as x1, x2, x3... represent the data set that needs to be input into the graph neural network (GNN) to determine the equivalent unit configuration results of the bus.

[0090] The technical solution of the embodiment of the present application is to concretize the construction of the bus branch diagram of the power system into obtaining the device connection relationship of the equipment in the power system, and according to the device connection relationship, at least one node device belonging to the same bus and the same bus are determined as a bus node, and the branch equipment connected to the bus is determined as a branch. According to the device connection relationship, the bus node and the connected branch are connected to generate a bus branch diagram of the power system, and the device connection relationship in the power system obtained in real time is used to construct the bus branch diagram, thereby realizing real-time acquisition of device data in the power system, taking into account the complex structure of the power system, improving the construction efficiency and accuracy of the bus branch diagram, and enhancing the comprehensiveness of the data relied on for the state estimation of the power system.

[0091] In some embodiments of the present application, the node device includes at least one of a generator set, a load set, a capacitor, a reactor, and a compensator; the branch device includes at least one of a transmission line, a transformer, and a switch.

[0092] Generator sets can be used to generate equipment active power. Generator sets can input equipment active power to the bus. Load sets can consume equipment active power. Load sets can receive equipment active power output from the bus. Capacitors can be used to store electrical energy, with the characteristics of charging and discharging, passing AC, and blocking DC. Capacitors can input equipment active power to the bus or receive active power output from the bus. Reactors can be used to limit short-circuit current or higher-order harmonics in the power grid. Reactors can input active power to the bus or receive active power output from the bus. Compensators can input active power to the bus or receive active power output from the bus. Transmission lines can be used to transmit electrical energy. Transformers can be used to change the voltage in the power grid. Switches can be used to open and close lines.

[0093] This solution simplifies the screening process of node devices and branch devices by concretizing the node devices as at least one of the generator sets, load units, capacitors, reactors and compensators, and concretizing the branch devices as at least one of the transmission lines, transformers and switches, thereby further improving the efficiency of constructing the busbar branch diagram.

[0094] In some embodiments of the present application, the active power imbalance of at least one bus node is calculated, including: obtaining the device connection relationship and the active power of the equipment in the power system; determining each device connected to the bus based on the device connection relationship; calculating the device active power of each device connected to the bus, and determining the active power imbalance of the bus as the active power imbalance of the bus node.

[0095] Specifically, the device connection relationships and active power of devices in the power system can be obtained. Based on the device connection relationships, each device connected to the bus can be detected. The active power of each device connected to the bus is calculated. The difference between the active power of the device input to the bus and the active power of the device output from the bus is calculated to obtain the active power imbalance of the bus, which is used as the active power imbalance of the bus node.

[0096] This solution obtains the device connection relationship and device active power of the equipment in the power system, determines the devices connected to the bus based on the device connection relationship, calculates the device active power of each device connected to the bus, and determines the active power imbalance of the bus. This is used as the active power imbalance of the bus node, further improving the calculation efficiency and accuracy of the active power imbalance of the bus.

[0097] In some embodiments of the present application, the active power of the device includes at least one unit device active power corresponding to each unit moment in the time period to be measured; the active power imbalance includes at least one unit active power imbalance corresponding to each unit moment in the time period to be measured.

[0098] The time period to be measured may be a time period for estimating the state of the power system. Optionally, the time period to be measured may be a time period including the current moment, to enable real-time state estimation of the power system. The unit moment may be a data sampling moment. Optionally, data sampling may be performed at preset time intervals within the time period to be measured. For example, the time period to be measured may be 24 hours, and the preset time interval may be 15 minutes. The unit device active power may be the device active power corresponding to the unit moment. The unit active power imbalance may be the active power imbalance corresponding to the unit moment.

[0099] Specifically, the device connection relationships of the devices in the power system and the active power of at least one device unit corresponding to each unit moment during the time period to be measured can be obtained. Based on the device connection relationships, each device connected to the bus can be detected. The active power of at least one device unit corresponding to each unit moment during the time period to be measured for each device connected to the bus is calculated. The difference between the active power of at least one device unit corresponding to each unit moment during the time period to be measured input to the bus and the active power of at least one device unit corresponding to each unit moment during the time period to be measured output from the bus is calculated to obtain the active power imbalance of at least one device unit corresponding to each unit moment during the time period to be measured for the bus.

[0100] For example, FIG5 is a schematic diagram of a graph neural network for determining the equivalent unit configuration results of each bus. As shown in FIG5 , the structure of the graph neural network is given. The bus branch diagram and the corresponding data set x1, x2, x3… (i.e., the active power imbalance of the bus at each unit time in the time period to be measured) can be used as the input of the graph neural network. After being processed by the hidden layer of the graph neural network (GNN), the output Y is the classification prediction of the equivalent unit configuration results of each corresponding bus. The input data of the graph neural network is the active measurement imbalance of each bus node. For example, the input data set {x1, x2, x3,…} of bus node 1 represents the active measurement imbalance of the bus node at different unit times {t1, t2, t3,…} respectively. The output Y of the graph neural network is the classification identifier of the bus node, which is used to identify whether the corresponding bus is set with an equivalent unit.

[0101] This solution further improves the calculation efficiency and accuracy of the active power imbalance of the bus by concretizing the active power of the equipment as at least one unit equipment active power corresponding to each unit moment in the time period to be measured, and concretizing the active power imbalance as at least one unit active power imbalance corresponding to each unit moment in the time period to be measured.

[0102] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0103] Figure 6 is a schematic diagram of the structure of a graph neural network-based state estimation device provided according to some embodiments of the present application. Embodiments of the present application are applicable to situations where state estimation of power systems is performed based on graph neural networks. The device can execute a graph neural network-based state estimation method. The device can be implemented in hardware and / or software and can be configured in an electronic device that carries a graph neural network-based state estimation function.

[0104] Referring to FIG6 , the state estimation apparatus based on a graph neural network comprises a bus branch diagram construction module 610, an equivalent unit configuration result determination module 620, a device power acquisition module 630, and a state estimation module 640. The bus branch diagram construction module 610 is configured to construct a bus branch diagram for the power system and calculate the active power imbalance of at least one bus node; the equivalent unit configuration result determination module 620 is configured to input the bus branch diagram and the active power imbalance of each bus node into the graph neural network to determine the equivalent unit configuration results for each corresponding bus; the device power acquisition module 630 is configured to configure equivalent units for the bus based on the equivalent unit configuration results and obtain the device power in the power system; and the state estimation module 640 is configured to perform state estimation on the devices in the power system based on the device power and calculate the device voltage estimate of the devices.

[0105] The technical solution of the embodiment of the present application is to construct a bus branch diagram of the power system and calculate the active power imbalance of at least one bus node, input the bus branch diagram and the active power imbalance of each bus node into the graph neural network, determine the equivalent unit configuration results of each corresponding bus, configure equivalent units for the bus based on the equivalent unit configuration results, and obtain the equipment power in the power system, perform state estimation on the equipment in the power system according to the equipment power, and calculate the equipment voltage estimation value of the equipment. The use of the graph neural network can adapt to the characteristics of real-time changes and complex structures of data in the power system, determine the equivalent unit configuration results of the power system, and solve the problems of low configuration efficiency of traditional equivalent unit configuration methods, which can only solve local problems of state estimation convergence, and cannot globally solve the problem of active measurement imbalance in the system, and the convergence of state estimation is still difficult to guarantee. The configuration efficiency of the equivalent units is improved, the problem of active measurement imbalance is globally solved, and the convergence of state estimation is guaranteed.

[0106] In some embodiments of the present application, the bus branch diagram construction module 610 includes: a device connection relationship acquisition unit, which is used to obtain the device connection relationship of the equipment in the power system; a same bus node determination unit, which is used to determine at least one node device belonging to the same bus and the same bus as a bus node based on the device connection relationship; a branch determination unit, which is used to determine the branch device connected to the bus as a branch; and a bus branch diagram construction unit, which is used to connect the bus node and the connected branch according to the device connection relationship to generate a bus branch diagram of the power system.

[0107] In some embodiments of the present application, the node device includes at least one of a generator set, a load set, a capacitor, a reactor, and a compensator; the branch device includes at least one of a transmission line, a transformer, and a switch.

[0108] In some embodiments of the present application, the bus branch diagram construction module 610 includes: an equipment active power acquisition unit, which is used to obtain the device connection relationship and equipment active power of the equipment in the power system; a bus connection device determination unit, which is used to determine each device connected to the bus based on the device connection relationship; an active power imbalance calculation unit, which is used to calculate the device active power of each device connected to the bus, and determine the active power imbalance of the bus as the active power imbalance of the bus node.

[0109] In some embodiments of the present application, the active power of the device includes at least one unit device active power corresponding to each unit moment in the time period to be measured; the active power imbalance includes at least one unit active power imbalance corresponding to each unit moment in the time period to be measured.

[0110] In some embodiments of the present application, the equivalent unit configuration result determination module 620 includes: a data input unit for inputting the bus node branch diagram and the active power imbalance of each bus node into the graph neural network; a first convolutional layer weighting unit for performing weighted averaging on the active power imbalance of the bus node and the active power imbalance of other bus nodes connected to the bus node for each bus node through the first convolutional layer of the graph neural network to obtain the weighted average result of the first convolutional layer; a first convolutional layer data output unit for calculating the product between the weighted average result of the first convolutional layer and the weight matrix coefficient of the first convolutional layer through the first convolutional layer of the graph neural network to obtain the feature data output by the bus node in the first convolutional layer; other convolutional layer weighting units for performing weighted averaging on the feature data output by the bus node in the previous convolutional layer and the feature data output by other bus nodes connected to the bus node for each bus node through the other convolutional layers of the graph neural network. The feature data output by the previous convolution layer is weighted averaged to obtain the weighted average results of other convolution layers; the data output units of other convolution layers are used to calculate the product between the weighted average results of other convolution layers and the weight matrix coefficients of other convolution layers through other convolution layers of the graph neural network, and obtain the feature data output by the bus node in other convolution layers until the feature data output by the last convolution layer is obtained; the preset feature data threshold comparison unit is used to compare the feature data output by the last convolution layer with the preset feature data threshold; the first equivalent unit configuration result determination unit is used to determine that the equivalent unit configuration result of the corresponding bus is a configured equivalent unit when the feature data output by the last convolution layer is greater than or equal to the preset feature data threshold; the second equivalent unit configuration result determination unit is used to determine that the equivalent unit configuration result of the corresponding bus is not configured with an equivalent unit when the feature data output by the last convolution layer is less than the preset feature data threshold.

[0111] In some embodiments of the present application, the device further includes: an equipment voltage measurement value acquisition module, which is used to configure equivalent units for the bus based on the equivalent unit configuration results, and obtain the equipment power in the power system while obtaining the equipment voltage measurement value in the power system; an equipment voltage difference calculation module, which is used to perform state estimation on the equipment in the power system, calculate the equipment voltage estimate of the equipment, and then calculate the equipment voltage difference between the equipment voltage estimate and the equipment voltage measurement value; a measurement error information prompt module, which is used to issue measurement error information of the equipment when the equipment voltage difference is greater than or equal to a preset voltage difference threshold, to prompt that there is an error in the corresponding equipment power, equipment voltage measurement value or equipment connection relationship of the equipment.

[0112] The state estimation device based on graph neural network provided in the embodiment of the present application can execute the state estimation method based on graph neural network provided in any embodiment of the present application, and has the corresponding functional modules and beneficial effects of the execution method.

[0113] In the technical solutions of the embodiments of the present application, the acquisition, storage and application of the equipment power in the power system, the equipment connection relationship of the equipment in the power system, the active power of the equipment and the voltage measurement values ​​of the equipment in the power system, etc., all comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0114] FIG7 shows a block diagram of an electronic device 700 that can be used to implement an embodiment of the present application. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present application described and / or required herein.

[0115] As shown in Figure 7, the electronic device 700 includes at least one processor 701, and a memory connected to the at least one processor 701, such as a read-only memory (ROM) 702, a random access memory (RAM) 703, etc., wherein the memory stores a computer program that can be executed by the at least one processor, and the processor 701 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 702 or the computer program loaded from the storage unit 708 into the random access memory (RAM) 703. In the RAM 703, various programs and data required for the operation of the electronic device 700 can also be stored. The processor 701, ROM 702 and RAM 703 are connected to each other via a bus 704. An input / output (I / O) interface 705 is also connected to the bus 704.

[0116] Multiple components in the electronic device 700 are connected to the I / O interface 705, including an input unit 706, such as a keyboard, a mouse, etc.; an output unit 707, such as various types of displays, speakers, etc.; a storage unit 708, such as a magnetic disk, an optical disk, etc.; and a communication unit 709, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 709 allows the electronic device 700 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0117] The processor 701 can be a variety of general-purpose and / or specialized processing components with processing and computing capabilities. Some examples of the processor 701 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The processor 701 executes the various methods and processes described above, such as the state estimation method based on the graph neural network.

[0118] In some embodiments, the state estimation method based on the graph neural network can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as the storage unit 708. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 700 via the ROM 702 and / or the communication unit 709. When the computer program is loaded into the RAM 703 and executed by the processor 701, one or more steps of the state estimation method based on the graph neural network described above can be performed. Alternatively, in other embodiments, the processor 701 can be configured to execute the state estimation method based on the graph neural network by any other appropriate means (for example, by means of firmware).

[0119] Various embodiments of the systems and techniques described above can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0120] Computer programs for implementing the methods of the present application may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the computer program is executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer program may be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0121] In the context of the present application, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. A computer-readable storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable signal medium. A more specific example of a machine-readable storage medium can include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0122] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0123] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0124] A computing system may include a client and a server. The client and server are generally remote from each other and typically interact via a communication network. This client-server relationship arises through computer programs running on the respective computers and establishing a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, a host product within a cloud computing service ecosystem that addresses the management difficulties and limited scalability of traditional physical hosts and VPS (Virtual Private Server) services.

[0125] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this application can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solution of this application can be achieved. This is not limited herein.

[0126] The above specific embodiments do not constitute a limitation on the scope of protection of this application. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application shall be included within the scope of protection of this application.

Claims

1. A state estimation method based on graph neural network, wherein: The method comprises: constructing a bus branch diagram of the power system and calculating the active power unbalance of at least one bus node; Inputting the bus branch diagram and the active power unbalance of each bus node into the graph neural network to determine the equivalent unit configuration results of each corresponding bus; Based on the equivalent unit configuration result, configure the bus with an equivalent unit and obtain the equipment power in the power system; and According to the device power, a state estimation is performed on the device in the power system, and an estimated value of the device voltage of the device is calculated.

2. The method according to claim 1, wherein: The constructing of a busbar branch diagram of the power system comprises: Obtaining device connection relationships of devices in the power system; According to the device connection relationship, at least one node device belonging to the same bus and the same bus are determined as a bus node; Determine the branch circuit equipment connected to the busbar as a branch circuit; and According to the equipment connection relationship, the bus nodes and the connected branches are connected to generate a bus branch diagram of the power system.

3. The method according to claim 2, wherein: The node equipment includes at least one of a generator set, a load set, a capacitor, a reactor and a compensator; the branch equipment includes at least one of a transmission line, a transformer and a switch.

4. The method according to claim 1, wherein: The calculating of the active power unbalance amount of at least one bus node comprises: Obtain the device connection relationship and device active power of the devices in the power system; According to the device connection relationship, determining each of the devices connected to the bus; and The active power of each device connected to the bus is calculated to determine the active power imbalance of the bus as the active power imbalance of the bus node.

5. The method according to claim 4, wherein: The equipment active power includes at least one unit equipment active power corresponding to each unit moment in the time period to be measured; the active power unbalance includes at least one unit active power unbalance corresponding to each unit moment in the time period to be measured.

6. The method according to claim 1, wherein: The step of inputting the bus branch diagram and the active power unbalance of each bus node into a graph neural network to determine the equivalent unit configuration results of each corresponding bus includes: The bus branch diagram and the active power unbalance of each bus node are input into the graph neural network. In the network; Through the first convolution layer of the graph neural network, for each of the bus nodes, the active power imbalance of the bus node and the active power imbalance of other bus nodes connected to the bus node are weighted averaged to obtain a weighted average result of the first convolution layer; By using the first convolution layer of the graph neural network, the product between the weighted average result of the first convolution layer and the weight matrix coefficient of the first convolution layer is calculated to obtain the feature data output by the bus node in the first convolution layer; Through the other convolutional layers of the graph neural network, for each of the bus nodes, the feature data output by the bus node in the previous convolutional layer and the feature data output by other bus nodes connected to the bus node in the previous convolutional layer are weighted averaged to obtain the weighted average result of the other convolutional layers; By using other convolutional layers of the graph neural network, the product between the weighted average result of the other convolutional layers and the weight matrix coefficients of the other convolutional layers is calculated to obtain the feature data output by the bus node in the other convolutional layers, until the feature data output by the last convolutional layer is obtained; Comparing the feature data output by the last convolutional layer with a preset feature data threshold; When the characteristic data output by the last convolutional layer is greater than or equal to a preset characteristic data threshold, determining that the equivalent unit configuration result of the corresponding bus is a configured equivalent unit; and When the characteristic data output by the last convolutional layer is less than a preset characteristic data threshold, it is determined that the equivalent unit configuration result of the corresponding bus is not to configure the equivalent unit.

7. The method according to claim 1, wherein: While configuring the busbar with an equivalent unit based on the equivalent unit configuration result and obtaining the power of equipment in the power system, the method further includes: Obtain voltage measurements of equipment in power systems; After the state of the equipment in the power system is estimated and the estimated value of the equipment voltage of the equipment is calculated, the method includes: calculating a device voltage difference between the device voltage estimate and the device voltage measurement; and When the device voltage difference is greater than or equal to a preset voltage difference threshold, measurement error information of the device is issued to prompt that there is an error in the device power, device voltage measurement value or device connection relationship corresponding to the device.

8. A state estimation device based on a graph neural network, wherein: The device comprises: A bus branch diagram construction module is used to construct a bus branch diagram of the power system and calculate the active power unbalance of at least one bus node; The equivalent unit configuration result determination module is used to input the bus branch diagram and the active power imbalance of each bus node into the graph neural network to determine the equivalent unit configuration of each corresponding bus. Set the result; an equipment power acquisition module, configured to configure an equivalent unit for the bus based on the equivalent unit configuration result, and to acquire the equipment power in the power system; and The state estimation module is used to perform state estimation on the equipment in the power system according to the equipment power and calculate the equipment voltage estimation value of the equipment.

9. An electronic device, wherein: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the state estimation method based on graph neural network described in any one of claims 1-7.

10. A computer-readable storage medium, wherein: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the state estimation method based on a graph neural network according to any one of claims 1 to 7 when executed.

Images