Section static voltage stability margin monitoring model construction method and related device

Abstract

The invention discloses a section static voltage stability margin monitoring model construction method and a related device, and the method comprises the steps: constructing a dot product graph attention network module based on a PaddlePaddle frame, and embedding edge features in the dot product graph attention network module; performing weighting processing on each channel in the original feature map by constructing a channel attention mechanism module; according to the method, evaluators of key nodes to be monitored in a power system are constructed respectively, and the construction process of each evaluator comprises the steps that a dot product graph attention network module and a channel attention mechanism module are combined, and multiple full-connection layers are stacked alternately at the same time; and forming a voltage stability margin monitoring model by the evaluators of the to-be-monitored key nodes, and performing training and parameter adjustment on the voltage stability margin monitoring model to obtain a final voltage stability margin monitoring model. Therefore, the problem that accurate monitoring of the node-level voltage stability margin in a cross-operation mode cannot be realized in the prior art is solved.

Description

Technical Field

[0001] The present application relates to the technical field of voltage stability analysis, and in particular to a method for constructing a section static voltage stability margin monitoring model and a related device. Background Art

[0002] With the expansion of power grid scale and the rapid growth of power load, the voltage stability problem of power system has become increasingly prominent. Static voltage stability margin is an important indicator to measure the ability of the system to resist voltage collapse under the current operating state, which has important guiding significance for the planning, design, operation and dispatch of power system.

[0003] Traditional voltage stability margin assessment methods, such as static voltage stability margin assessment based on the continuous power flow method, generally rely on complex mathematical models and iterative calculations, which are often computationally expensive and time-consuming. They cannot respond quickly to real-time changes in the power grid status and are difficult to effectively support online monitoring and emergency decision-making of the power system.

[0004] However, the existing voltage stability margin assessment method based on deep learning, 2. Voltage stability margin assessment based on deep learning: Due to the high complexity and nonlinear characteristics of the power system, the generalization performance of the existing deep learning model is weak, and the assessment accuracy under certain working conditions may be low, which cannot meet the needs of high-precision assessment. Secondly, the structure of the deep learning model often requires the input data dimension to be consistent. Specifically, each layer of neurons (or nodes) in the neural network is connected to the neurons of the previous layer through weights to learn the features in the input data. This connection mode requires that the input data of each layer must have consistent dimensions to ensure that the network can correctly execute the forward propagation and backpropagation algorithms, thereby optimizing and learning parameters. In the evaluation of the static voltage stability margin of the power system, the network topology of the power system will change with the operation mode of the power grid. The number of nodes, the number of lines and their interconnection methods of the system may change, which will cause the dimension of the input data to change accordingly. This change directly violates the requirement of the deep learning model for the consistency of the input data dimension, making the originally trained model unable to be directly applied to the new system state. Therefore, it is difficult for general deep learning models to flexibly adapt to changes in the network topology of the power system, and it is impossible to realize static voltage stability margin monitoring under cross-operation modes. Summary of the invention

[0005] The present application provides a method for constructing a section static voltage stability margin monitoring model and related devices, which are used to solve the problem that the prior art is unable to flexibly adapt to changes in the network topology of the power system, resulting in the inability to achieve accurate monitoring of node-level voltage stability margin under cross-operating modes.

[0006] In view of this, the first aspect of the present application provides a method for constructing a section static voltage stability margin monitoring model, the method comprising:

[0007] S1. Build a dot product graph attention network module based on the PaddlePaddle framework, and embed edge features in the dot product graph attention network module;

[0008] S2, by constructing a channel attention mechanism module, each channel in the original feature map is weighted;

[0009] S3, respectively constructing an evaluator for each key node to be monitored in the power system, wherein the construction process of each evaluator includes: combining the dot product graph attention network module with the channel attention mechanism module, and alternately stacking multiple fully connected layers;

[0010] S4. The evaluators of the key nodes to be monitored are combined into a voltage stability margin monitoring model, and the voltage stability margin monitoring model is trained and parameterized to obtain a final voltage stability margin monitoring model. The system network topology information and real-time power flow data are used as inputs of the voltage stability margin monitoring model, and the static voltage stability margin index of the key nodes to be monitored is used as the output of the voltage stability margin monitoring model.

[0011] Optionally, the channel attention mechanism module is constructed to perform weighted processing on each channel in the original feature map, including:

[0012] Perform global average pooling on each feature map to compress the spatial dimension of the feature map into a single value, thereby obtaining global information on each channel; use a fully connected layer and activation function to process the global information and generate an attention weight for each channel to reflect the model's evaluation of the importance of different channel features; multiply the attention weight by the original feature map to achieve feature recalibration.

[0013] Optionally, the system network topology information and real-time power flow data include:

[0014] An adjacency matrix used to store the numbering information of the nodes at the beginning and end of the system branch; an edge feature matrix used to store the system branch admittance amplitude, the head-end active power and the head-end reactive power; a node feature matrix used to store the system node's active power generation, reactive power generation, load active power, load reactive power, reactive power compensation, voltage amplitude and voltage angle; a node type used to store the node type label; a label used to store the static voltage stability margin index of the node to be monitored; a node name containing the node name information; a sample ID containing the sample number information.

[0015] Optionally, the static voltage stability margin index is expressed as:

[0016] ;

[0017] In the formula, is the static stability limit power of the key node to be monitored, It is the operating power of the key node to be monitored under the initial section.

[0018] Optionally, the training and parameter adjustment of the voltage stability margin monitoring model includes: adjusting the learning rate using a cosine annealing strategy.

[0019] Optionally, step S4 further includes:

[0020] Two actual data sections are selected as the basic data of the training set and the validation set for sample expansion, and the mean square error (mse) is used as the loss function to verify the effect of the voltage stability margin monitoring model.

[0021] A second aspect of the present application provides a section static voltage stability margin monitoring model construction system, the system comprising:

[0022] A first construction unit is used to construct a dot product graph attention network module based on the PaddlePaddle framework, and embed edge features in the dot product graph attention network module;

[0023] The second construction unit is used to perform weighted processing on each channel in the original feature map by constructing a channel attention mechanism module;

[0024] A third construction unit is used to respectively construct an evaluator for each key node to be monitored in the power system, wherein the construction process of each evaluator includes: combining the dot product graph attention network module with the channel attention mechanism module, and alternately stacking multiple fully connected layers;

[0025] The fourth construction unit is used to form a voltage stability margin monitoring model by combining the evaluators of the key nodes to be monitored, training and adjusting the parameters of the voltage stability margin monitoring model to obtain the final voltage stability margin monitoring model, and using the system network topology information and real-time power flow data as the input of the voltage stability margin monitoring model, and using the static voltage stability margin index of the key nodes to be monitored as the output of the voltage stability margin monitoring model.

[0026] Optionally, it also includes: a verification unit, which is used to select two actual data section data as basic data of the training set and the verification set respectively for sample expansion, and use the mean square error MSE as the loss function to verify the effect of the voltage stability margin monitoring model.

[0027] A third aspect of the present application provides a device for constructing a section static voltage stability margin monitoring model, the device comprising a processor and a memory:

[0028] The memory is used to store program code and transmit the program code to the processor;

[0029] The processor is used to execute the steps of the method for constructing a section static voltage stability margin monitoring model as described in the first aspect according to the instructions in the program code.

[0030] A fourth aspect of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium is used to store program code, and the program code is used to execute the section static voltage stability margin monitoring model construction method described in the first aspect above.

[0031] It can be seen from the above technical solutions that this application has the following advantages:

[0032] The present application provides a method for constructing a section static voltage stability margin monitoring model. The voltage stability margin monitoring model is based on the PaddlePaddle framework. By constructing a dot product graph attention network module, it deeply analyzes the interaction relationship between nodes in the power system and focuses on the impact of key nodes on the system voltage stability. It further introduces a channel attention mechanism to intelligently adjust the node feature importance weights according to system state changes, dynamically optimize node feature expressions, and improve model accuracy. It further realizes the construction of a node-level voltage stability margin evaluator to achieve a refined assessment of the voltage stability margin of key nodes. Finally, multiple node-level voltage stability margin evaluators are combined into a voltage stability margin monitoring model, which can adapt to system topology changes in real time and achieve accurate monitoring of node-level voltage stability margins across operating modes. The present application combines deep learning with static voltage stability margin assessment to achieve efficient and accurate assessment of voltage stability margins at key nodes in the power system, providing strong support and guarantee for the safe and stable operation and emergency dispatch of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 A method for constructing a section static voltage stability margin monitoring model provided in an embodiment of the present application;

[0034] Figure 2 A schematic diagram of the channel attention mechanism provided in an embodiment of the present application;

[0035] Figure 3 A schematic diagram of model input characteristics provided in an embodiment of the present application;

[0036] Figure 4 The cosine annealing strategy learning rate curve provided in the embodiment of the present application varies with the round number;

[0037] Figure 5 This is a schematic diagram of the model verification effect provided in the embodiments of the present application;

[0038] Figure 6 A system for building a section static voltage stability margin monitoring model is provided in an embodiment of the present application. DETAILED DESCRIPTION

[0039] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0040] See also Figure 1 , a method for constructing a section static voltage stability margin monitoring model provided in an embodiment of the present application includes:

[0041] Step 101: Build a dot product graph attention network module based on the PaddlePaddle framework, and embed edge features in the dot product graph attention network module.

[0042] It should be noted that this application constructs a dot product graph attention network module based on the PaddlePaddle framework, uses the attention mechanism to capture the mutual influence and change relationship between key nodes in the power system, and embeds edge features to enhance the sensitivity to the grid topology structure, so that the model can more sensitively capture the dynamic changes of the voltage stability margin caused by topology changes in the grid (such as line switching, faults, etc.).

[0043] Among them, the core idea of ​​the dot product graph attention network (PD-GAT) is to simulate the human attention mechanism by introducing dot product operations, so that the model can intelligently focus on the nodes and their connections in the graph that are most critical or have the highest information content for the target task. Specifically, in the PD-GAT model, each node not only uses its own feature information, but also combines the feature information of the neighboring nodes directly connected to it to calculate an attention weight through dot product operations. This process allows the model to automatically learn and quantify the relative importance of connections between different nodes without the need for additional complex transformations or parameters. After obtaining the attention weights, PD-GAT uses these weights to perform weighted aggregation on the features of neighboring nodes. This dot product-based weighted aggregation strategy, on the basis of efficiently fusing the feature information of neighboring nodes, also reflects the strength of the association between them and the current node, thereby more effectively integrating the local information structure in the graph, highlighting the nodes and connections in the graph that are critical to voltage stability assessment, and improving the accuracy and efficiency of data analysis.

[0044] In the application scenario of static voltage stability margin assessment of power systems, due to the complexity and dynamics of the power system network topology, traditional methods often find it difficult to fully capture and effectively process these characteristics. PD-GAT, with its unique dot product attention mechanism, can flexibly and adaptively adjust the attention to different nodes and connections, and accurately identify the key nodes and paths that affect voltage stability. This feature greatly enhances the evaluation capability and robustness of the model in a complex power system environment, making the monitoring of static voltage stability margin more accurate and reliable. Therefore, in this application, the dot product graph attention network (PD-GAT) is used as the core model architecture, aiming to deeply explore the complex characteristics of the power system and significantly improve the accuracy and overall performance of the static voltage stability margin assessment model.

[0045] At the same time, in addition to node features, this application also embeds edge features in the attention network, transforms the edge features through a specific embedding layer, and then participates in the attention weight calculation together with the node features. In this way, the model can more sensitively capture the dynamic changes in the voltage stability margin caused by topological changes (such as changes in operating mode) in the power grid.

[0046] Step 102: By constructing a channel attention mechanism module, each channel in the original feature map is weighted.

[0047] It should be noted that this application introduces a channel attention mechanism to automatically analyze and evaluate the importance of node feature channels according to changes in system status, and accordingly perform refined and adaptive corrections to feature channel weights to optimize the feature expression capabilities of the model.

[0048] Step 103: construct an evaluator for each key node to be monitored in the power system respectively, wherein the construction process of each evaluator includes: combining a dot product graph attention network module with a channel attention mechanism module, and alternately stacking multiple fully connected layers.

[0049] It should be noted that the present application constructs a node voltage stability margin evaluator with high precision and strong generalization by deeply combining multiple dot product graph attention network modules with channel attention mechanism modules and alternately stacking multiple layers of fully connected layers.

[0050] Specific:

[0051] This application forms the core architecture of the evaluator feature extraction module by combining the above-mentioned graph attention module with the channel attention module. This architecture constructs a node-level voltage stability margin evaluator with high precision and strong generalization by alternately stacking multiple fully connected layers and the final fully connected layer output module.

[0052] Furthermore, in this application, the node-level voltage stability margin evaluator parameter settings are shown in Table 1:

[0053] Table 1 Node level voltage stability margin evaluator parameter setting table

[0054]

[0055] Among them, the left side of Table 1 represents the names of each layer in the network. GAT, CA, and fc represent the attention layer, channel attention layer, and fully connected layer, respectively. The numbers behind them are used as the names of the layers for distinction. The right side of Table 1 represents the parameter design of each layer. The three parameters of the GAT attention layer represent the node input, node output, and number of attention heads, respectively. The two parameters of the channel attention layer and the fully connected layer represent the node input and node output, respectively. For the parameters with arrows '→' in the middle, it represents the internal transformation, such as 'Submodule 1: CA1…45 / 2→2 / 45' means that within this layer, the node input dimension 45 is first processed and reduced to 2, and then 2 is processed and increased to 45, that is, the final node output dimension is 45. The above parameter setting table reflects the overall model structure of the node-level voltage stability margin controller.

[0056] Step 104, the evaluators of the key nodes to be monitored are combined into a voltage stability margin monitoring model, the voltage stability margin monitoring model is trained and the parameters are adjusted to obtain the final voltage stability margin monitoring model, and the system network topology information and real-time power flow data are used as the input of the voltage stability margin monitoring model, and the static voltage stability margin index of the key nodes to be monitored is used as the output of the voltage stability margin monitoring model.

[0057] It should be noted that the present application combines multiple node-level voltage stability margin evaluators to form a voltage stability margin monitoring model, so that it can adapt to system topology changes in real time, continuously optimize model parameters and improve model performance through training, and realize accurate and real-time monitoring of voltage stability margins of multiple key nodes under cross-operation modes of new energy transmission sections.

[0058] It is understandable that the voltage stability margin monitoring model is composed of multiple node-level voltage stability margin evaluator groups. Taking the example of 130 key nodes to be monitored in the system, the voltage stability margin monitoring model proposed in this application is composed of 130 node-level voltage stability margin evaluators, which are trained and model parameter tuned as a whole. The voltage stability margin model simultaneously outputs the voltage stability margin indicators of 130 key nodes, where each evaluator corresponds to the node to be monitored one by one, and only outputs the voltage stability margin of the node.

[0059] The advantage of the above model is that it can adapt to changes in system topology in real time. For example, if the number of nodes in the original 130 systems decreases or increases due to changes in operating modes, the model only needs to reduce or increase the number of node-level voltage stability margin evaluators accordingly, and the overall training and tuning process does not change. If the original 130 nodes to be monitored are reduced to 121 due to topology changes, the model can automatically adjust the number of evaluators to 121 based on the read-in system topology, and realize accurate monitoring of node-level voltage stability margins under different operating modes.

[0060] In one embodiment, step 102 includes:

[0061] Global average pooling is performed on each feature map to compress the spatial dimension of the feature map into a single value, thereby obtaining the global information on each channel; the global information is processed using a fully connected layer and an activation function to generate an attention weight for each channel to reflect the model's evaluation of the importance of different channel features; the attention weight is multiplied by the original feature map to achieve feature recalibration.

[0062] It should be noted that this application introduces a channel attention mechanism to enable the model to adaptively adjust the degree of emphasis on different channel features according to the global information of the feature map, thereby improving the ability to capture key information and the ability to express tasks.

[0063] Specifically, the channel attention mechanism includes the following steps:

[0064] As attached Figure 2 As shown, first, for each feature map (dimension N Feature F) performs global average pooling (F squeeze ), compressing the spatial dimensions of the feature map into a single value (dimension 1 Feature F), thereby obtaining global information on each channel. Then, using the fully connected layer and activation function (F excitation , w is the weight matrix) processes the global information and generates the attention weight of each channel (dimension 1 Feature F) to reflect the model's evaluation of the importance of different channel features. Finally, the generated attention weight is multiplied by the original feature map (F scale ), realize feature recalibration (dimension N This process is equivalent to weighting each channel in the original feature map, so that the model can pay more attention to those channel features that are more critical to the task, while relatively ignoring relatively unimportant features, effectively improving the model's feature expression ability.

[0065] In one embodiment, the system network topology information and real-time power flow data in step 104 include:

[0066] An adjacency matrix used to store the numbering information of the nodes at the beginning and end of the system branch; an edge feature matrix used to store the system branch admittance amplitude, the head-end active power and the head-end reactive power; a node feature matrix used to store the system node's active power generation, reactive power generation, load active power, load reactive power, reactive power compensation, voltage amplitude and voltage angle; a node type used to store the node type label; a label used to store the static voltage stability margin index of the node to be monitored; a node name containing the node name information; a sample ID containing the sample number information.

[0067] It should be noted that in the present application, the input of the model mainly integrates the system network topology information and real-time power flow data, specifically including the 'adjacency matrix': the numbering information of the nodes at the beginning and end of the storage system branch; 'edge feature matrix': the storage system branch admittance amplitude, the first-end active power, and the first-end reactive power; 'node feature matrix': the storage system node's active power, reactive power, load active power, load reactive power, reactive power compensation, voltage amplitude, voltage angle, that is, the active power growth direction; 'node type': storage node type label, the node label to be monitored is 1 (the node to be monitored in this patent is a 220KV node), and the rest are 0; 'label': storage of the static voltage stability margin index of the node to be monitored (see 6. Model output for details); 'node name': node name; 'sample ID': sample number. Figure 3 shown.

[0068] The output of the model is the static voltage stability margin index of the node. The expression of the static voltage stability margin index is:

[0069] ;

[0070] In the formula, is the static stability limit power of the key node to be monitored, It is the operating power of the key node to be monitored under the initial section.

[0071] In one embodiment, in step 104, the voltage stability margin monitoring model is trained and parameters are adjusted, including: using a cosine annealing strategy to adjust the learning rate.

[0072] It should be noted that during the model training process, this application uses the cosine annealing strategy to adjust the learning rate to help the model quickly approach the global minimum and accelerate the convergence process. It is also beneficial for the model to make fine adjustments in local areas and improve the generalization ability of the model. The learning rate change curve during model training is as follows: Figure 4 As shown:

[0073] ;

[0074] Among them, l and epoch represent the learning rate and round, lmin and lmax represent the lower and upper limits of the learning rate, T represents the restart period of the benchmark, and K i T represents the total number of training epochs required after the i-th restart. In this application, K is set i =2 i-1 , T=2.

[0075] In one embodiment, after step 104, the method further includes:

[0076] Two actual data sections are selected as the basic data of the training set and the validation set for sample expansion, and the mean square error (mse) is used as the loss function to verify the effect of the voltage stability margin monitoring model.

[0077] It should be noted that if Figure 5 As shown in the figure, two actual data sections are selected as the basic data of the training set and the validation set for sample expansion, and the mean square error mse is used as the loss function to verify the model effect. The experimental results show that the model can effectively use node-level information and accurately monitor the node voltage stability margin across operating modes. The model verification effect is shown in Table 2:

[0078] Table 2

[0079]

[0080] The mean square error expression is:

[0081] .

[0082] The present application provides a method for constructing a section static voltage stability margin monitoring model. The voltage stability margin monitoring model is based on the PaddlePaddle framework. By constructing a dot product graph attention network module, it deeply analyzes the interaction relationship between nodes in the power system and focuses on the impact of key nodes on the system voltage stability. It further introduces a channel attention mechanism to intelligently adjust the node feature importance weights according to system state changes, dynamically optimize node feature expressions, and improve model accuracy. It further realizes the construction of a node-level voltage stability margin evaluator to achieve a refined assessment of the voltage stability margin of key nodes. Finally, multiple node-level voltage stability margin evaluators are combined into a voltage stability margin monitoring model, which can adapt to system topology changes in real time and achieve accurate monitoring of node-level voltage stability margins across operating modes. The present application combines deep learning with static voltage stability margin assessment to achieve efficient and accurate assessment of voltage stability margins at key nodes in the power system, providing strong support and guarantee for the safe and stable operation and emergency dispatch of the power system.

[0083] The above is a method for constructing a section static voltage stability margin monitoring model provided in an embodiment of the present application, and the following is a system for constructing a section static voltage stability margin monitoring model provided in an embodiment of the present application.

[0084] See also Figure 6 , a section static voltage stability margin monitoring model construction system provided in an embodiment of the present application includes:

[0085] The first construction unit 201 is used to construct a dot product graph attention network module based on the PaddlePaddle framework and embed edge features in the dot product graph attention network module.

[0086] The second construction unit 202 is used to perform weighted processing on each channel in the original feature map by constructing a channel attention mechanism module.

[0087] The third construction unit 203 is used to respectively construct an evaluator for each key node to be monitored in the power system, wherein the construction process of each evaluator includes: combining a dot product graph attention network module with a channel attention mechanism module, and alternately stacking multiple fully connected layers.

[0088] The fourth construction unit 204 is used to form a voltage stability margin monitoring model by combining the evaluators of the key nodes to be monitored, train and adjust the parameters of the voltage stability margin monitoring model to obtain the final voltage stability margin monitoring model, and use the system network topology information and real-time power flow data as the input of the voltage stability margin monitoring model, and use the static voltage stability margin index of the key nodes to be monitored as the output of the voltage stability margin monitoring model.

[0089] In one embodiment, the section static voltage stability margin monitoring model construction system of the present application also includes: a verification unit, which is used to select two actual data section data as the basic data of the training set and the verification set for sample expansion, and use the mean square error MSE as the loss function to verify the effect of the voltage stability margin monitoring model.

[0090] Furthermore, an embodiment of the present application also provides a device for constructing a section static voltage stability margin monitoring model, the device comprising a processor and a memory:

[0091] The memory is used to store program code and transmit the program code to the processor;

[0092] The processor is used to execute the steps of the section static voltage stability margin monitoring model construction method as described in the above method embodiment according to the instructions in the program code.

[0093] Furthermore, a computer-readable storage medium is provided in an embodiment of the present application, and the computer-readable storage medium is used to store program code, and the program code is used to execute the section static voltage stability margin monitoring model construction method described in the above method embodiment.

[0094] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0095] The terms "first", "second", "third", "fourth", etc. (if any) in the specification 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 sequence. 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 an order other than those illustrated or described herein, for example. 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 that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0096] It should be understood that in this application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0097] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0098] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0099] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0100] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (full name in English: Read-Only Memory, English abbreviation: ROM), random access memory (full name in English: Random Access Memory, English abbreviation: RAM), disk or optical disk and other media that can store program codes.

[0101] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for constructing a section static voltage stability margin monitoring model. It is characterized in that include: S1. Build a dot product graph attention network module based on the PaddlePaddle framework, and embed edge features in the dot product graph attention network module; S2. By constructing a channel attention mechanism module, each channel in the original feature map is weighted; S3, respectively constructing an evaluator for each key node to be monitored in the power system, wherein the construction process of each evaluator includes: combining the dot product graph attention network module with the channel attention mechanism module, and alternately stacking multiple fully connected layers; S4. The evaluators of the key nodes to be monitored are combined into a voltage stability margin monitoring model, and the voltage stability margin monitoring model is trained and parameterized to obtain a final voltage stability margin monitoring model. The system network topology information and real-time power flow data are used as inputs of the voltage stability margin monitoring model, and the static voltage stability margin index of the key nodes to be monitored is used as the output of the voltage stability margin monitoring model.

2. According to the method for constructing a section static voltage stability margin monitoring model according to claim 1, It is characterized in that By constructing a channel attention mechanism module, each channel in the original feature map is weighted, including: Perform global average pooling on each feature map to compress the spatial dimension of the feature map into a single value, thereby obtaining global information on each channel; use a fully connected layer and activation function to process the global information and generate an attention weight for each channel to reflect the model's evaluation of the importance of different channel features; multiply the attention weight by the original feature map to achieve feature recalibration.

3. According to the method for constructing a section static voltage stability margin monitoring model according to claim 1, It is characterized in that The system network topology information and real-time power flow data include: An adjacency matrix used to store the numbering information of the nodes at the beginning and end of the system branch; an edge feature matrix used to store the system branch admittance amplitude, the head-end active power and the head-end reactive power; a node feature matrix used to store the system node's active power generation, reactive power generation, load active power, load reactive power, reactive power compensation, voltage amplitude and voltage angle; a node type used to store the node type label; a label used to store the static voltage stability margin index of the node to be monitored; a node name containing the node name information; a sample ID containing the sample number information.

4. According to the method for constructing a section static voltage stability margin monitoring model according to claim 1, It is characterized in that The expression of the static voltage stability margin index is: ； In the formula, is the static stability limit power of the key node to be monitored, It is the operating power of the key node to be monitored under the initial section.

5. According to the method for constructing a section static voltage stability margin monitoring model according to claim 1, It is characterized in that The training and parameter adjustment of the voltage stability margin monitoring model includes: adjusting the learning rate by adopting a cosine annealing strategy.

6. The method for constructing a section static voltage stability margin monitoring model according to claim 1, It is characterized in that Step S4, further comprising: Two actual data sections are selected as the basic data of the training set and the validation set for sample expansion, and the mean square error (mse) is used as the loss function to verify the effect of the voltage stability margin monitoring model.

7. A section static voltage stability margin monitoring model construction system, It is characterized in that include: A first construction unit is used to construct a dot product graph attention network module based on the PaddlePaddle framework, and embed edge features in the dot product graph attention network module; The second construction unit is used to perform weighted processing on each channel in the original feature map by constructing a channel attention mechanism module; A third construction unit is used to respectively construct an evaluator for each key node to be monitored in the power system, wherein the construction process of each evaluator includes: combining the dot product graph attention network module with the channel attention mechanism module, and alternately stacking multiple fully connected layers; The fourth construction unit is used to form a voltage stability margin monitoring model by combining the evaluators of the key nodes to be monitored, training and adjusting the parameters of the voltage stability margin monitoring model to obtain the final voltage stability margin monitoring model, and using the system network topology information and real-time power flow data as the input of the voltage stability margin monitoring model, and using the static voltage stability margin index of the key nodes to be monitored as the output of the voltage stability margin monitoring model.

8. The section static voltage stability margin monitoring model construction system according to claim 7, It is characterized in that Also includes: The verification unit is used to select two actual data section data as the basic data of the training set and the verification set respectively for sample expansion, and use the mean square error mse as the loss function to verify the effect of the voltage stability margin monitoring model.

9. A device for constructing a section static voltage stability margin monitoring model. It is characterized in that The device comprises a processor and a memory: The memory is used to store program code and transmit the program code to the processor; The processor is used to execute the section static voltage stability margin monitoring model construction method described in any one of claims 1-6 according to the instructions in the program code.

10. A computer-readable storage medium, It is characterized in that The computer-readable storage medium is used to store program code, and the program code is used to execute the section static voltage stability margin monitoring model construction method described in any one of claims 1-6.

Images