Static voltage stability evaluation method and system for power system

Abstract

The invention discloses an electric power system static voltage stability evaluation method and system, and the method comprises the steps: obtaining operation scene data generated in a new energy station and a load operation interval in an electric power system, calculating a minimum model feature value in each operation scene, and building a first data set; calculating the causal intensity of the new energy power generation power and load power with the minimum mode characteristic value, and performing characteristic screening according to the causal intensity to form key characteristic variables; taking the key feature variables as input, taking the minimum model feature value as output, and establishing a prediction model for integrating an attention mechanism into a reverse feature enhancement time convolution network; and predicting and evaluating the static voltage stability of the power system based on the prediction model. According to the method, the data utilization rate can be improved by constructing a reverse feature extraction branch, key information can be selectively obtained by integrating a multi-head attention mechanism, the mapping relation between input data and output data is obtained through training, and static voltage stability evaluation can be rapidly and accurately achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of power system stability control, and in particular to a method and system for evaluating static voltage stability of a power system. Background Art

[0002] With the continuous increase in the scale of power systems and the large-scale access of new energy sources, the supply and demand balance of power systems faces severe challenges, and the operation of power systems is getting closer and closer to its stability limit. Since the 1970s, large-scale power outages caused by voltage collapse have occurred frequently, seriously threatening the safe and stable operation of power systems. Common methods for static voltage stability analysis of power systems include sensitivity method, PV curve and QV curve method, continuous power flow method, L index and characteristic structure analysis. The effectiveness of analysis methods based on physical models is greatly affected by modeling accuracy and system scale. At the same time, with the large-scale and extensive access of new energy sources, analysis based on deterministic physical models is difficult to adapt to the random changes in the operating state of the power system.

[0003] At present, with the rapid development of big data and artificial intelligence technology, power system stability analysis based on deep learning methods has received widespread attention. This method is no longer limited by the differences in research objectives and is applicable to various complex power system operation scenarios. It can characterize the complex temporal and spatial coupling laws of power systems and is an important way to study and reveal the operation behavior of new power systems. However, due to the large scale of actual power systems, the number of nodes is usually tens of thousands, resulting in an exponential growth in the dimensions of historical data sets. Processing such a large-scale data set will increase the cost of computing and storage, and will make the deep learning model extremely complex. In addition, not all variables have the same impact on voltage stability. Selecting important feature variables can save measurement costs and improve the computational efficiency of deep learning models. Therefore, it is necessary to screen out key factors from the operating variables of the power system, and use key factors combined with deep learning model methods to predict the voltage stability margin of the power system. Summary of the invention

[0004] The purpose of this section is to summarize some aspects of embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the specification abstract and the invention title of this application to avoid blurring the purpose of this section, the specification abstract and the invention title, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0005] In view of the above existing problems, the present invention is proposed.

[0006] Therefore, the present invention provides a method and system for evaluating the static voltage stability of a power system to solve the problem that existing evaluation methods are unable to reasonably select input signals from massive measurement data and fully exploit the characteristics of underlying data, thus restricting the evaluation performance of the model.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] In a first aspect, an embodiment of the present invention provides a method for evaluating static voltage stability of a power system, including: obtaining operation scenario data generated by a new energy station and a load operation interval in the power system, and calculating the minimum modulus eigenvalue under each operation scenario to establish a first data set;

[0009] Calculate the causal strength of renewable energy power generation and load power with the minimum modulus eigenvalue, and perform feature screening according to the causal strength to form key feature variables;

[0010] Taking the key feature variables as input and the minimum modulus eigenvalue as output, a prediction model integrating the attention mechanism into the inverse feature enhanced temporal convolutional network is established;

[0011] Based on the prediction model, the static voltage stability of the power system is predicted and evaluated.

[0012] As a preferred scheme of the static voltage stability assessment method of the power system described in the present invention, wherein: the first data set includes: a data set containing {new energy generation power, load power, minimum modulus eigenvalue} established based on the operating scenario data, and the data set is divided into a training set, a validation set and a test set in proportion.

[0013] As a preferred solution of the static voltage stability evaluation method of the power system described in the present invention, wherein: calculating the minimum modulus eigenvalue under each operation scenario includes: setting the active output fluctuation range of the new energy station and the load, randomly generating multiple groups of operating conditions within the fluctuation range, selecting several groups of conditions after removing the power imbalance scenario, linearizing the power system flow equation, and calculating the minimum modulus eigenvalue expressed as:

[0014]

[0015] Among them, ΔP s and ΔQ s are the deviations of node active power and reactive power, ΔV and Δθ are the deviations of node voltage amplitude and phase angle, H, N, M and L are the Jacobian matrices J s submatrix of ;

[0016] Assume that the Jacobian matrix J of the convergent power flow equation s With non-zero and distinct eigenvalues ​​λ1,λ2,…,λn , take the eigenvalue with the smallest modulus value as the minimum modulus eigenvalue λ min , select the minimum modulus eigenvalue λ min As an indicator of voltage stability margin.

[0017] As a preferred solution of the method for evaluating the static voltage stability of a power system described in the present invention, wherein: calculating the causal strength of the renewable energy power generation and load power with the minimum modulus eigenvalue, and performing feature screening according to the causal strength to form key feature variables comprises: taking the renewable energy power generation or load power as the first sequence, and taking the minimum modulus eigenvalue as the second sequence;

[0018] The first sequence and the second sequence are decomposed into two groups of intrinsic mode function components, and the instantaneous phase correlation between each pair of intrinsic mode function components is calculated as:

[0019]

[0020] in, is the phase difference of the jth pair of intrinsic mode function components at time t, A is the first sequence, and B is the second sequence;

[0021] Each modal subcomponent IMF is removed in turn and re-decomposed. After removing an IMF in a given time series, it is re-decomposed to generate a new set of intrinsic mode function components IMFB', and the instantaneous phase correlation Coh (S Aj ,S′ Bj ).

[0022] As a preferred solution of the method for evaluating the static voltage stability of a power system according to the present invention, it further comprises: using Coh(S Aj ,S B ' j The absolute causal strength D(S) of the first sequence on the second sequence is determined by the variance-weighted Euclidean distance of Aj →S Bj ) is expressed as:

[0023]

[0024]

[0025]

[0026] Among them, W j is the weight, Var Aj is the variance of the jth modal subcomponent of the first sequence;

[0027] like Then the first sequence has a positive causal effect on the second sequence, when When it is bigger, the cause and effect is stronger;

[0028] like Then the second sequence has a reverse causal effect on the first sequence, when The bigger it is, the stronger the cause and effect.

[0029] As a preferred solution of the static voltage stability assessment method of the power system described in the present invention, wherein: establishing a prediction model that integrates the attention mechanism into the inverse feature enhanced time convolution network includes: the inverse feature enhanced time convolution network includes a residual branch, an inverse feature enhanced branch and a time convolution branch, and the learner of the inverse feature enhanced time convolution network adopts the Adam gradient descent algorithm;

[0030] The multi-head attention mechanism layer is embedded before the output layer of the inverse feature enhancement temporal convolutional network to assign different weights to obtain input information;

[0031] The key characteristic variables are taken as input, and the minimum modulus eigenvalue is taken as output. Training is performed based on the training set and the validation set data to obtain a prediction model with optimal weight parameter configuration to predict the minimum modulus eigenvalue.

[0032] In a second aspect, the present invention provides a power system static voltage stability assessment system, comprising:

[0033] A data generation module, used to obtain operation scenario data generated by new energy stations and load operation intervals in the power system, calculate the minimum modulus eigenvalue under each operation scenario, and establish a first data set;

[0034] A key variable screening module is used to calculate the causal strength of the new energy power generation and load power with the minimum modulus eigenvalue, and to perform feature screening according to the causal strength to form key feature variables;

[0035] A modeling module, for establishing a prediction model that incorporates an attention mechanism into an inverse feature enhanced temporal convolutional network, using the key feature variables as input and the minimum modulus eigenvalue as output;

[0036] The prediction module is used to predict and evaluate the static voltage stability of the power system based on the prediction model.

[0037] As a preferred solution of the power system static voltage stability assessment system of the present invention, the modeling module includes:

[0038] The reverse feature enhanced temporal convolutional network includes a residual branch, a reverse feature enhanced branch and a temporal convolutional branch, and the learner of the reverse feature enhanced temporal convolutional network adopts the adam gradient descent algorithm;

[0039] The multi-head attention mechanism layer is embedded before the output layer of the inverse feature enhancement temporal convolutional network to assign different weights to obtain input information;

[0040] The key characteristic variables are taken as input, and the minimum modulus eigenvalue is taken as output. Training is performed based on the training set and the validation set data to obtain a prediction model with optimal weight parameter configuration to predict the minimum modulus eigenvalue.

[0041] In a third aspect, the present invention provides a computing device, comprising:

[0042] Memory and processor;

[0043] The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions. When the computer executable instructions are executed by the processor, the steps of the power system static voltage stability assessment method are implemented.

[0044] In a fourth aspect, the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the power system static voltage stability assessment method.

[0045] Compared with the prior art, the present invention has the following beneficial effects: without having to master the system topology and line parameters, the present invention can effectively capture the key factors affecting voltage stability, quantify the degree of causality, reduce the original feature set according to the causal strength, form a key feature subset, reduce the difficulty of model training, and improve computational efficiency by relying only on limited system historical operation data. By constructing a time convolution model with inverse feature enhancement and constructing an inverse feature extraction branch, the utilization rate of data can be improved, and the multi-head attention mechanism can be added to automatically and selectively focus on the key information in the high-dimensional features. By training to obtain the mapping relationship between input data and output data, static voltage stability evaluation can be achieved quickly and accurately. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0047] Figure 1 A method flow chart of a method and system for evaluating static voltage stability of a power system according to an embodiment of the present invention;

[0048] Figure 2 A schematic diagram of a multi-region interconnected system of a method and system for evaluating static voltage stability of a power system according to an embodiment of the present invention;

[0049] Figure 3 A schematic diagram of the structure of a RFETCN-MSA deep learning model of a method and system for evaluating static voltage stability of a power system according to an embodiment of the present invention;

[0050] Figure 4 A prediction error result diagram of a method for evaluating static voltage stability of a power system and a FBF model of the system according to an embodiment of the present invention;

[0051] Figure 5 A prediction error result diagram of an SVM model of a method and system for evaluating static voltage stability of a power system according to an embodiment of the present invention;

[0052] Figure 6 A prediction error result diagram of a CNN model of a method and system for evaluating static voltage stability of a power system according to an embodiment of the present invention;

[0053] Figure 7 A prediction error result diagram of an LSTM model of a method and system for evaluating static voltage stability of a power system according to an embodiment of the present invention;

[0054] Figure 8 A prediction error result diagram of a GRU model of a method and system for evaluating static voltage stability of a power system according to an embodiment of the present invention;

[0055] Fig. 9 A prediction error result diagram of a TCN model of a method and system for evaluating static voltage stability of a power system according to an embodiment of the present invention;

[0056] Fig.10 A prediction error result diagram of a RFETCN model of a method and system for evaluating static voltage stability of a power system according to an embodiment of the present invention;

[0057] Fig.11 This is a prediction error result diagram of the prediction model of the present invention for a method and system for evaluating static voltage stability of an electric power system described in one embodiment of the present invention. DETAILED DESCRIPTION

[0058] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0059] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0060] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0061] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.

[0062] At the same time, in the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the directions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0063] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0064] Example 1

[0065] Reference Figure 1 to Figure 3 , is an embodiment of the present invention, which provides a method for evaluating static voltage stability of a power system, comprising:

[0066] S1: Obtaining operation scenario data generated by new energy stations and load operation intervals in the power system, and calculating the minimum modulus eigenvalue under each operation scenario to establish a first data set;

[0067] It should be noted that the method of generating system operation scenario data in new energy stations and load operation intervals in the power system can be random.

[0068] In an embodiment of the present application, the first data set includes: a data set containing {new energy generation power, load power, minimum modulus eigenvalue} established based on operating scenario data, and the data set is divided into a training set, a validation set, and a test set in proportion.

[0069] In an optional embodiment, the first data set may be divided into a training set, a validation set, and a test set in a ratio of 8:1:1.

[0070] In an alternative embodiment, referring to Figure 2 A multi-region interconnected simulation system is constructed; among them, G2 is a doubly fed wind farm, the other units are synchronous generators, the load nodes are 6 to 9, and the synchronous generator G3 is a balancing unit. The system base capacity is taken as 100MVA, and the active output of G1 is 6p.u. under steady-state conditions, and the active output of G4 is 6p.u. Under steady-state conditions, the active output fluctuation range of the wind farm and the load is as shown in Table 1:

[0071] Table 1 Wind farm output and load fluctuation range

[0072] Power / pu <![CDATA[P w ]]> <![CDATA[Q w ]]> <![CDATA[P6]]> <![CDATA[P7]]> <![CDATA[P8]]> <![CDATA[P9]]> Fluctuation range 6～9 2～3 2～4 9～10 9～10 4～6

[0073] The reactive output is between 0p.u. and 2p.u., and 1200 groups of operating conditions are randomly generated within the range of wind power output and load output in Table 1. After removing the power imbalance scenario, several groups of conditions are selected.

[0074] In the embodiment of the present application, calculating the minimum modulus eigenvalue under each operation scenario includes: setting the active output fluctuation range of the new energy station and the load, randomly generating multiple groups of operating conditions within the fluctuation range, selecting several groups of conditions after removing the power imbalance scenario, linearizing the power system flow equation, and calculating the minimum modulus eigenvalue as:

[0075]

[0076] Among them, ΔP s and ΔQ s are the deviations of node active power and reactive power, ΔV and Δθ are the deviations of node voltage amplitude and phase angle, H, N, M and L are the Jacobian matrices J s submatrix of ;

[0077] Assume that the Jacobian matrix J of the convergent power flow equation s With non-zero and distinct eigenvalues ​​λ1,λ2,…,λ n , take the eigenvalue with the smallest modulus value as the minimum modulus eigenvalue λ min, select the minimum modulus eigenvalue λ min As an indicator of voltage stability margin.

[0078] S2: Calculate the causal strength of renewable energy power generation and load power with the minimum modulus eigenvalue, and perform feature screening based on the causal strength to form key feature variables;

[0079] In an optional embodiment, a causal decomposition method is used to calculate the causal strength of renewable energy power generation and load power relative to the minimum modulus eigenvalue. The causal decomposition method removes and re-decomposes modal subcomponents (Intrinsic Mode Functions, IMF) based on the Galilei principle.

[0080] In the embodiment of the present application, the causal strength of the renewable energy power generation and load power and the minimum modulus eigenvalue is calculated, and the feature screening is performed according to the causal strength to form the key feature variable, including: taking the renewable energy power generation or load power as the first sequence, and taking the minimum modulus eigenvalue as the second sequence;

[0081] The first sequence and the second sequence are decomposed into two groups of intrinsic mode function components, and the instantaneous phase correlation between each pair of intrinsic mode function components is calculated as:

[0082]

[0083] in, is the phase difference of the jth pair of intrinsic mode function components at time t, A is the first sequence, and B is the second sequence;

[0084] Each modal subcomponent IMF is removed in turn and re-decomposed. After removing an IMF in a given time series, it is re-decomposed to generate a new set of intrinsic mode function components IMFB', and the instantaneous phase correlation Coh (S Aj ,S′ Bj ).

[0085] In an optional embodiment, an ensemble empirical mode decomposition (EEMD) method may be applied to decompose the first sequence and the second sequence.

[0086] In the embodiment of the present application, it also includes: using Coh (S Aj ,S′ Bj The absolute causal strength D(S) of the first sequence on the second sequence is determined by the variance-weighted Euclidean distance of Aj →S Bj ) is expressed as:

[0087]

[0088]

[0089]

[0090] Among them, W j is the weight, Var Aj is the variance of the jth modal subcomponent of the first sequence;

[0091] like Then the first sequence has a positive causal effect on the second sequence, when When it is bigger, the cause and effect is stronger;

[0092] like Then the second sequence has a reverse causal effect on the first sequence, when The bigger it is, the stronger the cause and effect.

[0093] In an optional embodiment, the causal strength results of each variable and the minimum modulus eigenvalue are shown in Table 2:

[0094] Table 2 Causal strength of each variable and the minimum modulus eigenvalue

[0095]

[0096] From Table 2, we can see that according to the causal decomposition index The causal strength ranking of each variable is Q w >P w >Q6>P7>P6>Q7>Q8>Q9>P9>P8. If the actual power system installs measurement equipment at all nodes, it will lead to huge costs. In order to ensure the observability of the system and save costs, measurements are generally arranged at important nodes. In this embodiment, the first six variables with high causal strength or correlation can be selected as input, that is, {Q w ,P w ,Q6,P7,P6,Q7}, deep learning modeling of the minimum modulus eigenvalues ​​in different scenarios.

[0097] It should be noted that compared with the existing correlation analysis method, the causal decomposition method can accurately capture the causal relationship between variables and stability margin, effectively reflect the causal strength, and accurately identify the key influencing factors in the power system. Adding an inverse feature extraction module to the temporal convolutional network model effectively improves the feature extraction capability of the temporal convolutional network model, alleviates the gradient vanishing and gradient explosion problems of conventional models, and greatly improves the prediction accuracy. The prediction model incorporates the attention mechanism, which can adaptively focus on the key information in the high-dimensional features and realize the accurate evaluation of the static voltage stability margin of the power system.

[0098] S3: Taking the key feature variables as input and the minimum modulus eigenvalue as output, a prediction model is established that incorporates the attention mechanism into the inverse feature enhanced temporal convolutional network;

[0099] In an embodiment of the present application, establishing a prediction model that incorporates an attention mechanism into a reverse feature enhanced temporal convolutional network includes: the reverse feature enhanced temporal convolutional network includes a residual branch, a reverse feature enhanced branch, and a temporal convolutional branch, and the learner of the reverse feature enhanced temporal convolutional network adopts an adam gradient descent algorithm;

[0100] The multi-head attention mechanism layer is embedded before the output layer of the inverse feature enhancement temporal convolutional network to assign different weights to obtain input information;

[0101] Taking the key feature variables as input and the minimum modulus eigenvalue as output, training is performed based on the training set and validation set data to obtain the prediction model with the optimal weight parameter configuration to predict the minimum modulus eigenvalue.

[0102] In an alternative embodiment, see Figure 3 , establish a reverse feature enhanced temporal convolutional network RFETCN, and integrate the attention mechanism MSA to build a prediction model, namely the RFETCN-MSAy model. In the RFETCN-MSA deep learning model, the learner adopts the Adam gradient descent algorithm with a maximum number of iterations of 5000. A multi-head attention mechanism layer is embedded before the output layer, and different weights are assigned to selectively focus on important parts of the input information, so as to process large-scale data more efficiently.

[0103] S4: Predictive evaluation of the static voltage stability of the power system based on the prediction model.

[0104] In an optional example, the selected key feature variables are used as input, the minimum modulus eigenvalue is used as output, and training is performed based on the training set and validation set data to obtain the RFETCN-MSA model parameters. The test set data is input into the trained RFETCN-MSA model to evaluate the performance of the model.

[0105] The evaluation indicators of model performance include the mean absolute error e, which is used to reflect the average accuracy of the model. MAE ; Mean deviation error e MBE , ideally e MBE Should be close to zero. If it is positive, it means the predicted value is too high, and if it is negative, it means the predicted value is too low. RMSE , giving higher weight to large errors and focusing on evaluating the model's fit to large errors; the mean absolute percentage error e MAPE , focuses on evaluating the relative prediction accuracy of the model to understand the model's fitting effect on different data sets; the determination coefficient R 2, focusing on evaluating the overall fit of the model. The calculation of each error is as follows:

[0106]

[0107]

[0108]

[0109]

[0110]

[0111] Among them, λ i is the actual eigenvalue, n is the total number of eigenvalues, is the predicted feature value.

[0112] The RFETCN-MSA model parameters can be set as: convolution kernel size is 3×1, the number of convolution kernels is 16, the spatial loss factor is 0.001, the learning rate is 0.01, the number of attention heads of the multi-head attention mechanism is 2, and the number of channels of the query key is 56.

[0113] The above is a schematic scheme of a method for evaluating static voltage stability of a power system in this embodiment. It should be noted that the technical scheme of the system for evaluating static voltage stability of a power system and the technical scheme of the method for evaluating static voltage stability of a power system are of the same concept, and the details not described in detail in the technical scheme of the system for evaluating static voltage stability of a power system in this embodiment can be found in the description of the technical scheme of the method for evaluating static voltage stability of a power system.

[0114] In this embodiment, a static voltage stability assessment system for a power system includes:

[0115] A data generation module, used to obtain operation scenario data generated by new energy stations and load operation intervals in the power system, calculate the minimum modulus eigenvalue under each operation scenario, and establish a first data set;

[0116] The key variable screening module is used to calculate the causal strength of the new energy power generation and load power with the minimum modulus eigenvalue, and to screen the features according to the causal strength to form the key feature variables;

[0117] The modeling module is used to build a prediction model that incorporates the attention mechanism into the inverse feature enhanced temporal convolutional network with key feature variables as input and the minimum modulus eigenvalue as output;

[0118] The prediction module is used to predict and evaluate the static voltage stability of the power system based on the prediction model.

[0119] In the embodiment of the present application, the modeling module includes:

[0120] The reverse feature enhanced temporal convolutional network includes a residual branch, a reverse feature enhanced branch and a temporal convolutional branch, and the learner of the reverse feature enhanced temporal convolutional network adopts the adam gradient descent algorithm;

[0121] The multi-head attention mechanism layer is embedded before the output layer of the inverse feature enhancement temporal convolutional network to assign different weights to obtain input information;

[0122] Taking the key feature variables as input and the minimum modulus eigenvalue as output, training is performed based on the training set and validation set data to obtain the prediction model with the optimal weight parameter configuration to predict the minimum modulus eigenvalue.

[0123] This embodiment further provides a computing device, which is applicable to the case of the static voltage stability evaluation method of the power system, and includes:

[0124] Memory and processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the static voltage stability assessment method of the power system as proposed in the above embodiment.

[0125] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for evaluating the static voltage stability of a power system proposed in the above embodiment is implemented.

[0126] The storage medium proposed in this embodiment and the method for implementing the static voltage stability assessment of the power system proposed in the above embodiment belong to the same inventive concept. The technical details not fully described in this embodiment can be referred to the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0127] Through the above description of the implementation methods, the technicians in the relevant field can clearly understand that the present invention can be implemented by means of software and necessary general hardware, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ReadOnly, Memory, ROM), random access memory (RandomAccess Memory, RAM), flash memory (FLASH), hard disk or optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform the methods of various embodiments of the present invention.

[0128] Example 2

[0129] Reference Figures 4 to 11, which is an embodiment of the present invention. This embodiment is different from the first embodiment in that this embodiment verifies the prediction performance of the model established by the present invention by constructing 7 comparative models.

[0130] In this embodiment, according to the test scenario data, the RFETCN-MSA deep learning network is used to predict the minimum modulus eigenvalue. The comparison models include machine learning models: Support Vector Machine (SVM) and Radial Basis Function (RBF); deep learning models: Convolutional Neural Network (CNN), Gated Recurrent Unit (GRU) and Long Short-Term Memory (LSTM); and derivative models of deep learning models: Temporal Convolutional Network (TCN) and Reverse Feature Enhancement Temporal Convolutional Network (RFETCN).

[0131] The minimum modulus eigenvalue prediction errors of different models are shown in Table 3:

[0132] Table 3 Minimum modulus eigenvalue prediction error of different models

[0133]

[0134]

[0135] The prediction errors of each model can be found in Figures 4 to 11 As shown in Table 3, compared with the traditional machine learning models RBF and SVM, the average absolute percentage error of the method of the present invention is reduced by 50.27% and 48.92% respectively; compared with the deep learning models CNN, LSTM and GRU, the average absolute percentage error is reduced by 30.95%, 29.9% and 29.62% respectively. Therefore, the model of the present invention has a higher static voltage stability evaluation accuracy than the traditional machine learning model and other deep learning models.

[0136] The present invention greatly improves the feature extraction capability of the model of the present invention by adding an inverse feature enhancement model. Compared with the simple TCN model, the average absolute percentage error is reduced by 17.55%. In addition, since the model integrates the multi-head self-attention mechanism, the model can more accurately weight the features and focus on key features. Compared with the RFETCN model, the average absolute percentage error is reduced by 4.21%, further improving the prediction accuracy of the model.

[0137] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for evaluating static voltage stability of a power system, characterized in that: include: Obtaining operation scenario data generated by new energy stations and load operation intervals in the power system, and calculating the minimum modulus eigenvalue under each operation scenario to establish a first data set; Calculate the causal strength of renewable energy power generation and load power with the minimum modulus eigenvalue, and perform feature screening according to the causal strength to form key feature variables; Taking the key feature variables as input and the minimum modulus eigenvalue as output, a prediction model integrating the attention mechanism into the inverse feature enhanced temporal convolutional network is established; Based on the prediction model, the static voltage stability of the power system is predicted and evaluated.

2. The method for evaluating static voltage stability of a power system according to claim 1, characterized in that: The first data set includes: a data set including {new energy generation power, load power, minimum modulus eigenvalue} established based on the operation scenario data, and the data set is divided into a training set, a validation set and a test set in proportion.

3. The method for evaluating static voltage stability of a power system according to claim 2, characterized in that: Calculating the minimum modulus eigenvalue under each operation scenario includes: setting the active output fluctuation range of the new energy station and the load, randomly generating multiple groups of operating conditions within the fluctuation range, selecting several groups of conditions after removing the power imbalance scenario, linearizing the power system flow equation, and calculating the minimum modulus eigenvalue as: Where ΔP s and ΔQ s are the deviations of node active power and reactive power, ΔV and Δθ are the deviations of node voltage amplitude and phase angle, H, N, M and L are the Jacobian matrices J s submatrix of ; Assume that the Jacobian matrix J of the convergent power flow equation s With non-zero and distinct eigenvalues ​​λ1,λ2,…,λ n , take the eigenvalue with the smallest modulus value as the minimum modulus eigenvalue λ min , select the minimum modulus eigenvalue λ min As an indicator of voltage stability margin.

4. The method for evaluating static voltage stability of a power system according to claim 3, characterized in that: Calculating the causal strength of renewable energy power generation and load power and the minimum modulus eigenvalue, and performing feature screening according to the causal strength to form key feature variables includes: taking renewable energy power generation or load power as the first sequence, and taking the minimum modulus eigenvalue as the second sequence; The first sequence and the second sequence are decomposed into two groups of intrinsic mode function components, and the instantaneous phase correlation between each pair of intrinsic mode function components is calculated as: in, is the phase difference of the jth pair of intrinsic mode function components at time t, A is the first sequence, and B is the second sequence; Each modal subcomponent IMF is removed in turn and re-decomposed. After removing an IMF in a given time series, it is re-decomposed to generate a new set of intrinsic mode function components IMFB', and the instantaneous phase correlation Coh (S Aj ,S′ Bj ).

5. The method for evaluating static voltage stability of a power system according to claim 4, characterized in that: Also includes: Using Coh (S Aj ,S′ Bj The variance-weighted Euclidean distance of the first sequence determines the absolute causal strength D(S Aj →S Bj ) is expressed as: Among them, W j is the weight, Var Aj is the variance of the jth modal subcomponent of the first sequence; like Then the first sequence has a positive causal effect on the second sequence, when When it is bigger, the cause and effect is stronger; like Then the second sequence has a reverse causal effect on the first sequence, when The bigger it is, the stronger the cause and effect.

6. The method for evaluating static voltage stability of a power system according to claim 5, characterized in that: Establishing a prediction model that integrates the attention mechanism into the reverse feature enhanced temporal convolutional network includes: the reverse feature enhanced temporal convolutional network includes a residual branch, a reverse feature enhanced branch and a temporal convolutional branch, and the learner of the reverse feature enhanced temporal convolutional network adopts the adam gradient descent algorithm; The multi-head attention mechanism layer is embedded before the output layer of the inverse feature enhancement temporal convolutional network to assign different weights to obtain input information; The key characteristic variables are taken as input, and the minimum modulus eigenvalue is taken as output. Training is performed based on the training set and the validation set data to obtain a prediction model with optimal weight parameter configuration to predict the minimum modulus eigenvalue.

7. A static voltage stability assessment system for a power system, characterized in that: include: A data generation module, used to obtain operation scenario data generated by new energy stations and load operation intervals in the power system, calculate the minimum modulus eigenvalue under each operation scenario, and establish a first data set; A key variable screening module is used to calculate the causal strength of the new energy power generation and load power with the minimum modulus eigenvalue, and to perform feature screening according to the causal strength to form key feature variables; A modeling module, for establishing a prediction model that incorporates an attention mechanism into an inverse feature enhanced temporal convolutional network, using the key feature variables as input and the minimum modulus eigenvalue as output; The prediction module is used to predict and evaluate the static voltage stability of the power system based on the prediction model.

8. The power system static voltage stability assessment system according to claim 7, characterized in that: The modeling module includes: The reverse feature enhanced temporal convolutional network includes a residual branch, a reverse feature enhanced branch and a temporal convolutional branch, and the learner of the reverse feature enhanced temporal convolutional network adopts the adam gradient descent algorithm; The multi-head attention mechanism layer is embedded before the output layer of the inverse feature enhancement temporal convolutional network to assign different weights to obtain input information; The key characteristic variables are taken as input, and the minimum modulus eigenvalue is taken as output. Training is performed based on the training set and the validation set data to obtain a prediction model with optimal weight parameter configuration to predict the minimum modulus eigenvalue.

9. An electronic device, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the method for evaluating static voltage stability of a power system described in any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the method for evaluating static voltage stability of a power system as claimed in any one of claims 1 to 6.

Images