Power system state estimation method, device and equipment and storage medium

Abstract

The invention discloses a power system state estimation method, device and equipment and a storage medium, and belongs to the technical field of power system state monitoring. The method comprises the following steps: acquiring measurement data of a power system and a measurement missing mask matrix corresponding to the measurement data; performing mask embedding correction processing on the measurement data according to the measurement missing mask matrix to generate measurement embedding correction data; inputting the measurement embedded correction data into a system state estimation model to obtain initial state estimation data; and normalizing the initial state estimation data to generate a state estimation result. By using the technical scheme provided by the invention, the accuracy and robustness of state estimation of the power system can be improved.

Description

Technical Field

[0001] This invention relates to the field of power system condition monitoring technology, and in particular to a power system condition estimation method, apparatus, equipment and storage medium. Background Technology

[0002] The core task of power system state estimation is to calculate the state variables of the entire network (voltage amplitude, voltage phase angle, branch power, etc.) based on the redundant measurement data collected in real time (node ​​voltage amplitude, node injected active / reactive power, branch active / reactive power, etc.), so as to provide the dispatch center with a complete and reliable profile of the power grid operation.

[0003] Current state estimation methods generally employ weighted least squares (WLS) and its improved forms based on power flow equations. In recent years, researchers have begun to explore new paradigms for end-to-end mapping using deep learning. For example, they abstract the power grid as a graph structure and use graph convolutional networks (GCNs) to aggregate information from neighboring nodes to characterize node features, thereby directly inferring the state of the entire network.

[0004] However, existing technologies have the following drawbacks in practical applications: First, the numerical iterative method based on WLS is computationally burdensome. For large-scale power grids, each estimation requires a large number of floating-point operations, resulting in excessive CPU time consumption, which makes it difficult to meet the stringent time limit requirements of modern power grids for millisecond-level real-time closed-loop control. Second, the WLS objective function lacks robustness to bad data with large residuals. A single bad measurement can cause significant drift in the state vector, and an excessively high measurement missing rate will lead to a sharp increase in the probability of convergence failure. Third, the power grid topology may change at any time due to branch switching, bus splitting, etc., requiring the regeneration of the Jacobian matrix and correction of the factor table. Existing power systems cannot automatically adapt to this and can only rely on offline maintenance, which reduces the continuity and reliability of state estimation. In addition, existing methods are based on strict physical models and lack the ability to compensate for unmodeled micro-variables such as line temperature and parameter aging. Parameter errors will directly be converted into state errors, causing the estimation accuracy to gradually decrease with the years of operation.

[0005] Therefore, a more reliable solution is needed. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the prior art and provide a power system state estimation method, apparatus, device and storage medium, which can improve the accuracy and robustness of power system state estimation and improve the real-time performance of online power system state estimation.

[0007] To achieve the above objectives, the present invention employs the following technical solution:

[0008] On one hand, the present invention provides a power system state estimation method, the method comprising:

[0009] Obtain the measurement data of the power system and the measurement missing mask matrix corresponding to the measurement data;

[0010] Based on the measurement missing mask matrix, the measurement data is subjected to mask embedding correction processing to generate measurement embedding correction data;

[0011] The measurement embedding correction data is input into the system state estimation model to obtain the initial state estimation data;

[0012] The initial state estimation data is normalized to generate state estimation results.

[0013] In some possible implementations, the step of performing mask embedding correction processing on the measurement data based on the measurement missing mask matrix to generate measurement embedding correction data includes:

[0014] Based on the measurement missing mask matrix, the measurement data is subjected to mask embedding processing to generate a measurement embedding correction vector;

[0015] Based on the measurement embedding correction vector, the measurement data is corrected to generate measurement embedding correction data.

[0016] In some possible implementations, the step of performing mask embedding processing on the measurement data based on the measurement missing mask matrix to generate a measurement embedding correction vector includes:

[0017] The measurement data is projected into a high-dimensional embedding space to obtain an initial embedding vector;

[0018] The initial embedding vector is superimposed with the position embedding vector of the measurement data to generate a measurement embedding vector.

[0019] Based on the measurement missing mask matrix, the effective measurement embedding vectors in the measurement embedding vectors are aggregated to generate a system cross-section summary vector.

[0020] The system cross-section summary vector and the location embedding vector are fused to generate the measurement embedding correction vector.

[0021] In some possible implementations, the system state estimation model includes an input layer, multiple hidden layers containing light residual bottleneck modules, and an output layer; the step of inputting the measurement embedding correction data into the system state estimation model to obtain initial state estimation data includes:

[0022] The measurement embedding correction data is processed through the input layer to obtain the measurement correction vector;

[0023] The measurement correction vector is processed by feature extraction through the multiple hidden layers containing light residual bottleneck modules to generate measurement embedding correction features.

[0024] The output layer performs state estimation processing based on the measurement embedding correction features to generate the initial state estimation data.

[0025] In some possible implementations, the lightweight residual bottleneck module includes a dimensionality reduction module, an activation module, and a residual connection module. The step of performing feature extraction processing on the measurement correction vector through the multiple hidden layers containing the lightweight residual bottleneck module to generate measurement embedding correction features includes:

[0026] The measurement correction vector is input into the hidden layer containing the light residual bottleneck module for feature extraction processing to generate the current embedded correction feature;

[0027] The current embedding correction feature is input into the next hidden layer containing a light residual bottleneck module for feature extraction processing, until all hidden layers containing light residual bottleneck modules are passed, and the measurement embedding correction data is processed for feature extraction to obtain the measurement embedding correction feature.

[0028] The step of inputting the measurement correction vector into the hidden layer containing the light residual bottleneck module for feature extraction to generate the current embedded correction feature includes:

[0029] The measurement correction vector is input into the dimensionality reduction module for dimensionality reduction processing to generate dimensionality reduction embedding correction features;

[0030] The dimensionality reduction embedding correction features are input into the activation module for nonlinear activation processing to generate activation embedding correction features;

[0031] The activated embedding correction feature is input into the residual connection module for dimensionality upscaling and is added to the measurement correction vector by residual addition to generate the current embedding correction feature.

[0032] In some possible implementations, the normalization process of the initial state estimation data to generate state estimation results includes:

[0033] The initial state estimation data is compared with a preset state range to generate a state comparison result;

[0034] If the state comparison result indicates that the initial state estimation data exceeds the preset state range, the initial state estimation data is truncated to generate the state estimation result.

[0035] In some possible implementations, acquiring the power system measurement data includes:

[0036] Obtain the raw measurement data of the power system;

[0037] The original measurement data is standardized to generate the measurement data.

[0038] On the other hand, a power system state estimation device is provided, the device comprising:

[0039] The data acquisition module is used to acquire the measurement data of the power system and the measurement missing mask matrix corresponding to the measurement data;

[0040] The mask embedding correction module is used to perform mask embedding correction processing on the measurement data according to the measurement missing mask matrix, and generate measurement embedding correction data.

[0041] The state estimation determination module is used to input the measurement embedding correction data into the system state estimation model to obtain initial state estimation data; and to perform normalization processing on the initial state estimation data to generate state estimation results.

[0042] On the other hand, an electronic device is provided, the device including a processor and a memory, the memory storing at least one instruction and at least one program, the at least one instruction and the at least one program being loaded and executed by the processor to implement the power system state estimation method as described above.

[0043] On the other hand, a computer-readable storage medium is provided, wherein at least one instruction and at least one program are stored therein, the at least one instruction and the at least one program being loaded and executed by a processor to implement the power system state estimation method as described above.

[0044] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

[0045] In this invention, by acquiring the measurement data of the power system and the corresponding measurement missing mask matrix, and then performing mask embedding correction processing on the measurement data based on the measurement missing mask matrix to generate measurement embedding correction data, the robustness and accuracy of state estimation can be improved even when the data is incomplete or contains noise. Next, the measurement embedding correction data is input into the system state estimation model to obtain initial state estimation data. The initial state estimation data is then normalized to generate state estimation results. This allows the system state estimation model to realize and improve the real-time performance of online power system state estimation, as well as enhance the accuracy, reliability, and efficiency of power system state estimation. Attached Figure Description

[0046] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 This is a flowchart illustrating a power system state estimation method provided in an embodiment of the present invention;

[0048] Figure 2 This is a schematic diagram of a process provided in this embodiment of the invention to perform mask embedding correction processing on measurement data based on a measurement missing mask matrix to generate measurement embedding correction data.

[0049] Figure 3 This is a schematic diagram of the structure of a power system state estimation device provided in an embodiment of the present invention. Detailed Implementation

[0050] To enable those skilled in the art to better understand the technical solutions of this invention, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0051] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0052] In this embodiment of the invention, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0053] Various exemplary embodiments, features, and aspects of the present invention will now be described in detail with reference to the accompanying drawings. The same reference numerals in the drawings denote elements that have the same or similar functions. Although various aspects of the embodiments are shown in the drawings, they are not necessarily drawn to scale unless specifically indicated otherwise.

[0054] The term “exemplary” as used herein means “serving as an example, embodiment, or illustration.” Any embodiment illustrated herein as “exemplary” is not necessarily to be construed as superior to or better than other embodiments.

[0055] In this document, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent three cases: A alone, A and B simultaneously, and B alone. Furthermore, the term "at least one" in this document means any combination of at least two of any one or more elements. For example, including at least one of A, B, and C can mean including any one or more elements selected from the set consisting of A, B, and C.

[0056] Furthermore, to better illustrate the present invention, numerous specific details are set forth in the following detailed embodiments. Those skilled in the art will understand that the present invention can be practiced without certain specific details. In some instances, methods, means, elements, and circuits well known to those skilled in the art have not been described in detail in order to highlight the spirit of the invention.

[0057] Figure 1 This is a flowchart illustrating a power system state estimation method according to an embodiment of the present invention. This specification provides the method operation steps as shown in the embodiment or flowchart, but based on conventional or non-inventive labor, more or fewer operation steps may be included. The order of steps listed in the embodiment is merely one possible execution order among many and does not represent the only execution order. In actual system or server product execution, the method can be executed sequentially according to the embodiment or the accompanying drawings, or in parallel (e.g., in a parallel processor or multi-threaded processing environment). Specifically, as shown... Figure 1 As shown, the above method may include:

[0058] S101: Obtain the measurement data of the power system and the measurement missing mask matrix corresponding to the measurement data;

[0059] In one specific embodiment, the measurement data can be standardized measurement data collected in real time by measurement units deployed in the power system. Optionally, the measurement units may include phasor measurement units, remote terminal units, etc., and the measurement data may include bus voltage amplitude, bus voltage phase angle, line active power, line reactive power, generator active power output, generator reactive power output, load active power, load reactive power, etc. Optionally, the measurement data may include at least one target measurement data. Specifically, the measurement data may be a vector or matrix composed of at least one target measurement data according to a time window or cross-section, denoted as... M represents the total number of target measurement data.

[0060] The measurement missing mask matrix can be a matrix with the same dimensions as the measurement data, generated based on the communication status of the data acquisition system and the quality inspection results of the measurement data. Optionally, the measurement missing mask matrix can reflect the quality verification results of the measurement data. Optionally, the measurement missing mask matrix is ​​used to mark whether each target measurement data point is valid. Specifically, if the i-th target measurement data has been successfully acquired and passed verification, the i-th target measurement data is valid. If the i-th target measurement data is lost or determined to be an outlier and is therefore removed, the i-th target measurement data is invalid. Specifically, if the system has 100 measurement points, and the measurement data for the 5th, 12th, and 30th targets are missing, then the corresponding mask vector... middle, , , All other positions are 1.

[0061] In an optional embodiment, the acquisition of power system measurement data described above may include:

[0062] Acquire raw measurement data of the power system;

[0063] The raw measurement data is standardized to generate measurement data.

[0064] In one specific embodiment, the raw measurement data can be unstandardized data that is collected in real time and directly by measurement units deployed in the power system.

[0065] The above standardization process for the raw measurement data can be performed using Z-score standardization, as shown in the following formula:

[0066] ;

[0067] in, Indicates measurement data, This represents the raw measurement data; This indicates element-wise multiplication; and Let represent the mean vector and standard deviation vector of each measurement in the raw measurement data obtained based on historical data statistics, respectively; where, for measurements marked as invalid (i.e. ), corresponding The data can be filled with 0, the mean, or other preset values; alternatively, other standardization methods such as maximum-minimum standardization (normalization to the [0,1] interval) can be used. The specific standardization method can be selected according to the distribution characteristics of the original measurement data.

[0068] In the above embodiments, the dimensional and order-of-magnitude differences between different types of measurements (such as voltage and power) can be eliminated by standardizing the original measurement data, thereby improving the stability and reliability of subsequent model processing.

[0069] S102: Based on the measurement missing mask matrix, perform mask embedding correction processing on the measurement data to generate measurement embedding correction data;

[0070] In one specific embodiment, the measurement embedding correction data is obtained by performing mask embedding correction processing on the measurement data based on the measurement missing mask matrix.

[0071] In an optional embodiment, the above-described mask embedding correction processing of the measurement data based on the measurement missing mask matrix to generate measurement embedding correction data may include:

[0072] Based on the measurement missing mask matrix, the measurement data is masked and embedded to generate a measurement embedding correction vector.

[0073] Based on the measurement embedding correction vector, the measurement data is corrected to generate measurement embedding correction data.

[0074] In one specific embodiment, the measurement embedding correction vector is obtained by masking the measurement data based on the measurement missing mask matrix. Optionally, the measurement embedding correction vector can be used to correct the measurement data.

[0075] In an optional embodiment, the above-described mask embedding process of the measurement data based on the measurement missing mask matrix to generate the measurement embedding correction vector may include:

[0076] The measurement data is projected into a high-dimensional embedding space to obtain the initial embedding vector;

[0077] The initial embedding vector is superimposed with the position embedding vector of the measurement data to generate the measurement embedding vector.

[0078] Based on the measurement missing mask matrix, the valid measurement embedding vectors in the measurement embedding vectors are aggregated to generate system cross-section summary vectors.

[0079] The system cross-section summary vector and the location embedding vector are fused to generate a measurement embedding correction vector.

[0080] In a specific embodiment, the above-described projection of measurement data into a high-dimensional embedding space to obtain an initial embedding vector can be expressed as follows:

[0081] ;

[0082] in, This represents the initial embedding vector corresponding to the i-th target measurement data. Indicates the embedding dimension. This represents the measurement data of the i-th target; and These are learnable parameters;

[0083] Optionally, the measurement data can be... Each target measurement data The initial embedding vector is obtained by projecting a linear transformation onto a high-dimensional embedding space through a learnable fully connected layer. .

[0084] In one specific embodiment, superimposing the initial embedding vector with the location embedding vector of the measurement data can associate the measurement data with a learnable location vector, thereby enabling the model to perceive the physical location of the measurement data in the system. Specifically, superimposing the initial embedding vector with the location embedding vector of the measurement data can associate each target measurement data with a learnable location vector, thereby enabling the model to perceive the physical location of each target measurement data in the system (e.g., which line or bus it belongs to). Specifically, the measurement embedding vector generated by superimposing the initial embedding vector with the location embedding vector of the measurement data can be expressed as follows:

[0085] ;

[0086] in, It can represent the measurement embedding vector corresponding to the i-th target measurement data. This can represent the initial embedding vector corresponding to the i-th target measurement data. It can represent the position embedding vector corresponding to the i-th target measurement data.

[0087] Optionally, the location embedding vector of the measurement data can be determined through learning and optimization during the training process of the system state estimation model after random initialization.

[0088] In a specific embodiment, the effective measurement embedding vector can be the embedded measurement corresponding to the effective measurement marked in the measurement missing mask matrix within the measurement embedding vector. The system profile summary vector can be a summary vector characterizing the global information of a power system profile (snapshot). The above-mentioned aggregation processing of the effective measurement embedding vectors in the measurement embedding vector based on the measurement missing mask matrix to generate the system profile summary vector may include: filtering out all embedding vectors corresponding to effective measurements from the measurement embedding vector based on the measurement missing mask matrix to obtain the effective embedding vector. The effective embedding vectors are aggregated to generate system cross-sectional summary vectors. Optionally, masked mean pooling can be used to aggregate the effective embedding vectors, as shown in the following equation:

[0089] ;

[0090] Where s represents the system cross-section summary vector; This represents the mask vector corresponding to the measurement data of the i-th target.

[0091] Optionally, weighted pooling and multi-head self-attention mechanisms can be used for aggregation processing. The weights in weighted pooling can be dynamically generated by a learnable small network based on the position or importance of the measurement data. The multi-head self-attention mechanism allows information to be exchanged between the position embedding vectors of all valid measurements, thereby generating more expressive system cross-sectional summary vectors.

[0092] In one specific embodiment, the system cross-section summary vector and the location embedding vector of the measurement data are fused to generate a measurement embedding correction vector for the measurement data. Optionally, the system cross-section summary vector and the location embedding vector of each target measurement data are fused to generate a correction vector for each target measurement data, which can inject global system information into the representation of each measurement.

[0093] Optionally, a small multilayer perceptron can be used to fuse the system cross-sectional summary vector with the location embedding vector of the measurement data, as shown in the following equation:

[0094] ;

[0095] in, This represents the correction vector corresponding to the i-th target measurement data. This represents a vector concatenation operation; This represents a multilayer perceptron. This fusion MLP can contain two hidden layers and use the ReLU activation function.

[0096] In a specific embodiment, measurement data can be corrected using a measurement embedding correction vector to generate measurement embedding correction data. Specifically, the measurement embedding correction vector can be projected back to the dimension corresponding to the measurement data using a fully connected projection network to obtain the projected measurement embedding correction vector. The projected measurement embedding correction vector is then added to the measurement data to obtain the measurement embedding correction data, as shown in the following formula:

[0097] ;

[0098] ;

[0099] in, This represents the projected measurement embedding correction vector. This represents the measurement embedding correction vector. This indicates that the measurement data has been embedded and corrected. and This indicates that the parameters in the projected fully connected network have been learned and optimized. The projected fully connected network is used to adjust the dimension of the measurement embedding correction vector to the dimension corresponding to the measurement data.

[0100] In one specific embodiment Figure 2 This is a schematic diagram of a process provided in this embodiment of the invention for performing mask embedding correction processing on measurement data based on a measurement missing mask matrix to generate measurement embedding correction data; as shown below. Figure 2 As shown, the above-mentioned mask embedding correction processing of measurement data based on the measurement missing mask matrix, generating measurement embedding correction data, may include:

[0101] S201: Project the measurement data into a high-dimensional embedding space to obtain the initial embedding vector;

[0102] S202: Superimpose the initial embedding vector with the position embedding vector of the measurement data to generate the measurement embedding vector;

[0103] S203: Based on the measurement missing mask matrix, aggregate the valid measurement embedding vectors in the measurement embedding vectors to generate system cross-section summary vectors;

[0104] S204: The system cross-section summary vector and the location embedding vector are fused to generate a measurement embedding correction vector;

[0105] S205: Based on the measurement embedding correction vector, the measurement data is corrected to generate measurement embedding correction data.

[0106] In the above embodiments, a measurement embedding correction vector is determined through a mask embedding mechanism. Then, the measurement data is corrected according to the measurement embedding correction vector. This can integrate global information at the power system level into the correction of a single measurement, thereby effectively compensating for information loss caused by missing or abnormal data. This facilitates the model's identification and compensation of missing or abnormal measurement data, thereby improving the model's accuracy and efficiency in data processing. Furthermore, it effectively enhances the robustness of state estimation in the case of incomplete or noisy measurement data, thereby improving the accuracy and robustness of state estimation.

[0107] S103: Input the measurement embedding correction data into the system state estimation model to obtain the initial state estimation data;

[0108] In one specific embodiment, the system state estimation model can be used to estimate the power system state, improving the real-time performance and continuity of the power system state estimation. The initial state estimation data can be obtained by processing measurement embedding correction data according to the system state estimation model; the initial state estimation data can be data characterizing the power system state.

[0109] In an optional embodiment, the above system state estimation model may include an input layer, multiple hidden layers containing light residual bottleneck modules, and an output layer;

[0110] The above-mentioned input of measurement embedding correction data into the system state estimation model to obtain initial state estimation data may include:

[0111] The measurement embedding correction data is processed through the input layer to obtain the measurement correction vector;

[0112] The measurement correction vector is processed by feature extraction through at least one hidden layer containing a light residual bottleneck module to generate measurement embedding correction features.

[0113] The output layer performs state estimation processing based on the measurement embedding correction features to generate initial state estimation data.

[0114] In a specific embodiment, the model structure of the system state estimation model can be set according to actual application requirements. Optionally, the system state estimation model can be a lightweight deep neural network model, whose backbone network can be a multilayer perceptron, and also includes hidden layers with light residual bottleneck modules. Optionally, the input layer can be an input fully connected layer, and the output layer can be an output fully connected layer. Measurement embedding correction data is taken as input, passes through a fully connected layer, and then sequentially through multiple cascaded hidden layers containing light residual bottleneck modules. The output of the last hidden layer containing the light residual bottleneck module is fed into an output fully connected layer, with an output dimension of 2N, where N can be the number of buses in the power system. Optionally, the output 2N-dimensional vector is the initial state estimation data.

[0115] In an optional embodiment, the aforementioned lightweight residual bottleneck module may include a dimensionality reduction module, an activation module, and a residual connection module.

[0116] The above-mentioned method uses multiple hidden layers containing light residual bottleneck modules to perform feature extraction on the measurement correction vector, generating measurement embedding correction features that can include:

[0117] The measurement correction vector is input into the hidden layer containing the light residual bottleneck module for feature extraction to generate the current embedded correction feature;

[0118] The current embedding correction feature is input into the next hidden layer containing a light residual bottleneck module for feature extraction processing, until all hidden layers containing light residual bottleneck modules are passed, and the measurement embedding correction data is processed for feature extraction processing to obtain the measurement embedding correction feature.

[0119] Specifically, the measurement correction vector is input into the hidden layer containing the light residual bottleneck module for feature extraction, generating the current embedded correction features, including:

[0120] The measurement correction vector is input into the dimension reduction module for dimension reduction processing to generate dimension reduction embedded correction features;

[0121] The dimension-reduced embedding correction features are input into the activation module for nonlinear activation processing to generate activated embedding correction features.

[0122] The activated embedding correction feature is input into the residual connection module for dimensionality upscaling and is added to the measurement correction vector by residual addition to generate the current embedding correction feature.

[0123] In a specific embodiment, the dimensionality reduction module is used to perform dimensionality reduction processing. Optionally, the dimensionality reduction module can be a first fully connected layer, the output dimension of which is smaller than the input dimension. The measurement correction vector is input to the first fully connected layer for dimensionality reduction processing to obtain the dimensionality-reduced embedding correction feature. The activation module can be used to perform nonlinear activation processing. Optionally, the dimensionality-reduced embedding correction feature is input to the activation module, and a nonlinear activation function can be used for nonlinear activation processing to generate the activation embedding correction feature. Optionally, the nonlinear activation function can be set according to actual application requirements, such as ReLU, GELU, Sigmoid, Tanh, Swish, etc. The residual connection module can be used to perform dimensionality-up projection and residual connection processing. Optionally, the residual connection module can be a second fully connected layer, the output dimension of which is larger than the input dimension. The activation embedding feature is input to the second fully connected layer for dimensionality-up processing to obtain the dimensionality-up embedding correction feature. The dimensionality-up embedding correction feature and the measurement correction vector are added together by residual processing to obtain the current embedding correction feature.

[0124] Optionally, the current embedded feature is input to the next hidden layer containing the light residual bottleneck module for feature extraction processing, until each hidden layer containing the light residual bottleneck module processes the embedded correction feature output by the previous hidden layer containing the light residual bottleneck module, and the last hidden layer containing the light residual bottleneck module outputs the measurement embedded correction feature.

[0125] In the above embodiments, a lightweight residual bottleneck structure is introduced into the MLP backbone network to alleviate the gradient vanishing problem during model training, promote the effective propagation of gradients, and thus make model training more stable and convergent, thereby improving the accuracy of model state estimation and enhancing the precision and reliability of power system state estimation. Furthermore, by combining a mask embedding mechanism and using an MLP optimized with a residual bottleneck as the backbone network, high performance can be achieved without excessively increasing the number of model parameters, thereby improving the efficiency of model state estimation, reducing computational resource requirements, and making it suitable for online state estimation scenarios with high real-time requirements.

[0126] S104: Normalize the initial state estimation data to generate state estimation results.

[0127] In one specific embodiment, the state estimation result is obtained by normalizing the initial state estimation data.

[0128] In an optional embodiment, the above-described normalization process for the initial state estimation data to generate the state estimation result may include:

[0129] The initial state estimation data is compared with the preset state range to generate a state comparison result.

[0130] If the state comparison result indicates that the initial state estimation data exceeds the preset state range, the initial state estimation data is truncated to generate the state estimation result.

[0131] In one specific embodiment, the preset state range can be set according to actual application requirements. Optionally, the preset state range can be determined based on historical long-term operating data. Optionally, if the state comparison result indicates that the initial state estimation data exceeds the preset state range, the initial state estimation data is truncated to the upper or lower limit of the preset state range to generate a state estimation result; if the state comparison result indicates that the initial state estimation data does not exceed the preset state range, the initial state estimation data is determined as the state estimation result. Optionally, the state estimation result can be used for subsequent power grid security and stability analysis and control decisions.

[0132] Optionally, the process of normalizing the initial state estimation data to generate the state estimation result may also include: using the initial state estimation data as initial values, running a fast AC power flow calculation (such as the PQ decomposition method) for iterative correction until the result fully satisfies the power flow equation, thereby generating the state estimation result.

[0133] In the above embodiments, by post-processing the initial state estimation data, it can be ensured that it conforms to the physical laws of the power system, thereby improving the accuracy and reliability of the power system state estimation.

[0134] As can be seen from the technical solutions provided in the embodiments of this specification above, this specification obtains the measurement data of the power system and the corresponding measurement missing mask matrix. Then, based on the measurement missing mask matrix, it performs mask embedding correction processing on the measurement data to generate measurement embedding correction data. This can improve the robustness and accuracy of state estimation when the data is incomplete or contains noise. Next, the measurement embedding correction data is input into the system state estimation model to obtain initial state estimation data. The initial state estimation data is then normalized to generate state estimation results. The system state estimation model can realize and improve the real-time performance of online state estimation of the power system, as well as improve the accuracy, reliability, and efficiency of power system state estimation.

[0135] This invention also provides a power system state estimation device, correspondingly, Figure 3This is a schematic diagram of the structure of a power system state estimation device provided in an embodiment of the present invention; as shown below. Figure 3 As shown, the above-mentioned device includes:

[0136] Data acquisition module 310 is used to acquire measurement data of the power system and the measurement missing mask matrix corresponding to the measurement data;

[0137] The mask embedding correction module 320 is used to perform mask embedding correction processing on the measurement data according to the measurement missing mask matrix to generate measurement embedding correction data.

[0138] The state estimation determination module 330 is used to input the measurement embedding correction data into the system state estimation model to obtain initial state estimation data; and to perform normalization processing on the initial state estimation data to generate state estimation results.

[0139] This invention also provides an electronic device, which includes a processor and a memory. The memory stores at least one instruction, at least one program, a code set, or an instruction set. The at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by the processor to implement the power system state estimation method as described in any of the method embodiments.

[0140] Embodiments of the present invention also provide a computer storage medium, which may be disposed in a server to store at least one instruction, at least one program, code set, or instruction set for implementing the method embodiments. The at least one instruction, the at least one program, the code set, or the instruction set is loaded and executed by the processor to implement the power system state estimation method as described in any of the method embodiments.

[0141] Optionally, in embodiments of the present invention, the storage medium may be located at at least one of a plurality of network servers in a computer network. Optionally, in embodiments of the present invention, the storage medium may include, but is not limited to, various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0142] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0143] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more flowcharts and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0144] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more flowcharts and / or boxes Figure 1 The function specified in one or more boxes.

[0145] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more flowcharts and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0146] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of an instruction containing one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than those shown in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0147] Finally, it should be noted that the embodiments of the present invention have been described above in conjunction with the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. A power system state estimation method, characterized in that, The method includes: Obtain the measurement data of the power system and the measurement missing mask matrix corresponding to the measurement data; Based on the measurement missing mask matrix, the measurement data is subjected to mask embedding correction processing to generate measurement embedding correction data; The measurement embedding correction data is input into the system state estimation model to obtain the initial state estimation data; The initial state estimation data is normalized to generate state estimation results.

2. The power system state estimation method according to claim 1, characterized in that, The step of performing mask embedding correction processing on the measurement data based on the measurement missing mask matrix to generate measurement embedding correction data includes: Based on the measurement missing mask matrix, the measurement data is subjected to mask embedding processing to generate a measurement embedding correction vector. Based on the measurement embedding correction vector, the measurement data is corrected to generate measurement embedding correction data.

3. The power system state estimation method according to claim 2, characterized in that, The step of performing mask embedding processing on the measurement data based on the measurement missing mask matrix to generate a measurement embedding correction vector includes: The measurement data is projected into a high-dimensional embedding space to obtain an initial embedding vector; The initial embedding vector is superimposed with the position embedding vector of the measurement data to generate a measurement embedding vector. Based on the measurement missing mask matrix, the effective measurement embedding vectors in the measurement embedding vectors are aggregated to generate a system cross-section summary vector. The system cross-section summary vector and the location embedding vector are fused to generate the measurement embedding correction vector.

4. The power system state estimation method according to claim 1, characterized in that, The system state estimation model includes an input layer, multiple hidden layers containing light residual bottleneck modules, and an output layer. The step of inputting the measurement embedding correction data into the system state estimation model to obtain initial state estimation data includes: The measurement embedding correction data is processed through the input layer to obtain the measurement correction vector; The measurement correction vector is processed by feature extraction through the multiple hidden layers containing light residual bottleneck modules to generate measurement embedding correction features. The output layer performs state estimation processing based on the measurement embedding correction features to generate the initial state estimation data.

5. The power system state estimation method according to claim 4, characterized in that, The lightweight residual bottleneck module includes a dimensionality reduction module, an activation module, and a residual connection module. The measurement correction vector is processed through the multiple hidden layers containing the lightweight residual bottleneck module to generate measurement embedding correction features, including: The measurement correction vector is input into the hidden layer containing the light residual bottleneck module for feature extraction processing to generate the current embedded correction feature; The current embedding correction feature is input into the next hidden layer containing a light residual bottleneck module for feature extraction processing, until all hidden layers containing light residual bottleneck modules are passed, and the measurement embedding correction data is processed for feature extraction to obtain the measurement embedding correction feature. The step of inputting the measurement correction vector into the hidden layer containing the light residual bottleneck module for feature extraction to generate the current embedded correction feature includes: The measurement correction vector is input into the dimensionality reduction module for dimensionality reduction processing to generate dimensionality reduction embedding correction features; The dimensionality reduction embedding correction features are input into the activation module for nonlinear activation processing to generate activation embedding correction features; The activated embedding correction feature is input to the residual connection module for dimensionality upscaling and is added to the measurement correction vector by residual addition to generate the current embedding correction feature.

6. The power system state estimation method according to claim 1, characterized in that, The step of normalizing the initial state estimation data to generate state estimation results includes: The initial state estimation data is compared with a preset state range to generate a state comparison result; If the state comparison result indicates that the initial state estimation data exceeds the preset state range, the initial state estimation data is truncated to generate the state estimation result.

7. The power system state estimation method according to claim 1, characterized in that, The acquisition of power system measurement data includes: Obtain the raw measurement data of the power system; The original measurement data is standardized to generate the measurement data.

8. A power system state estimation device, characterized in that, The device includes: The data acquisition module is used to acquire the measurement data of the power system and the measurement missing mask matrix corresponding to the measurement data; The mask embedding correction module is used to perform mask embedding correction processing on the measurement data according to the measurement missing mask matrix, and generate measurement embedding correction data. The state estimation determination module is used to input the measurement embedding correction data into the system state estimation model to obtain initial state estimation data; and to perform normalization processing on the initial state estimation data to generate state estimation results.

9. An electronic device comprising a processor and a memory, the memory storing at least one instruction and at least one program, the at least one instruction and the at least one program being loaded and executed by the processor to implement the power system state estimation method as claimed in any one of claims 1 to 7.

10. A computer storage medium storing at least one instruction and at least one program, the at least one instruction and the at least one program being loaded and executed by a processor to implement the power system state estimation method as claimed in any one of claims 1 to 7.

Images