Method and device for evaluating stability of voltage of power system

Abstract

The application provides a power system voltage stability evaluation method and device, wherein the method comprises the following steps: obtaining a bus vector corresponding to each bus in a power system based on a pre-trained bus vector representation training model; obtaining distance data converted from time series data corresponding to each bus in the power system; and obtaining a power system voltage stability evaluation result according to the bus vector and the distance data. The method fully utilizes time series information and spatial structure information reflected by all buses in the power system during voltage instability, excavates the correlation between the buses, and organically combines the time series information and the spatial structure information in a representation learning manner, so that the instability characteristics of all load buses in the power system during voltage instability can be learned more comprehensively, and the accuracy of power system voltage stability evaluation is effectively ensured.

Description

Technical Field

[0001] This invention relates to the field of power system safety and stability assessment technology, and in particular to a method and apparatus for assessing the stability of power system voltage. Background Technology

[0002] Voltage instability in power systems can lead to large-scale blackouts, causing huge economic losses and seriously affecting social life. Utilizing node voltage indicators to achieve rapid monitoring of grid voltage stability is of great significance for ensuring the safe and stable operation of the power system.

[0003] In complex real-world power systems, information about medium- and long-term voltage instability is not only embedded in the time series of dynamic processes but also reflected in spatial structure information. Load buses at different spatial locations may exhibit drastically different roles during medium- and long-term voltage instability. Therefore, to more reliably assess medium- and long-term voltage stability, it is necessary to simultaneously integrate time series information and spatial structure information. However, due to the difficulty in modeling the spatial information of the power grid, there are few technical solutions for assessing medium- and long-term voltage stability that integrate spatiotemporal information.

[0004] Currently, by using visualization and voltage interpolation methods to analyze the changes in voltage over time in the entire region during transient voltage instability, a transient voltage stability assessment method integrating spatiotemporal information has been established. This method does indeed cleverly link spatiotemporal information, enabling system dispatchers to understand the dynamic characteristics of the system under different spatial distributions from an apparent perspective. However, interpolation based on spatial distance ignores the complex electrical relationships between system nodes, and still cannot deeply understand the role of spatial information in voltage instability.

[0005] Therefore, the inability of existing technologies to effectively integrate the temporal and spatial structural information of voltage in a power system to accurately assess the voltage stability of the power system is a critical issue that urgently needs to be addressed in the field of power system safety and stability assessment technology. Summary of the Invention

[0006] This invention provides a method and apparatus for assessing the stability of power system voltage, which overcomes the shortcomings of existing technologies that cannot effectively integrate the temporal and spatial structural information of voltage in a power system to accurately assess the voltage stability of the power system, thereby achieving stable assessment of power system voltage.

[0007] On one hand, the present invention provides a method for evaluating the stability of power system voltage, comprising: obtaining the bus vector corresponding to each bus in the power system based on a pre-trained bus vector representation training model; obtaining distance data converted from time series data corresponding to each bus in the power system; and obtaining the stability evaluation result of the power system voltage based on the bus vector and the distance data.

[0008] Furthermore, the method for assessing the stability of power system voltage also includes: acquiring bus training samples and labeling the bus training samples according to bus labels; and training the bus vector representation training model using the labeled bus training samples to obtain a trained bus vector representation training model.

[0009] Further, the step of labeling a portion of the bus training samples in the bus training samples according to the bus labels includes: extracting a target time series subsequence corresponding to the multi-dimensional time series corresponding to each bus; obtaining the best matching subsequence in the multi-dimensional time series corresponding to the target time series subsequence; calculating the bus correlation degree between each bus and the other buses in the bus training samples based on the best matching subsequence; and determining the bus label corresponding to each bus in the bus training samples based on the bus correlation degree; wherein, the target time series subsequence is the subsequence with the largest metric function value.

[0010] Furthermore, obtaining the bus vector corresponding to each bus in the power system includes: obtaining the bus vector corresponding to each bus through a trained bus vector representation model based on the one-hot encoding corresponding to each bus; wherein, the one-hot encoding corresponds one-to-one with the bus label, and the bus vector is used to characterize the proportion of the target bus affected by the other buses in the power system during the voltage instability process of the power system.

[0011] Further, the step of calculating the bus correlation degree between each bus and the other buses in the bus training sample based on the best matching subsequence includes: calculating the bus correlation degree using the bus correlation degree formula, which is as follows:

[0012]

[0013] Where p and q are the bus labels in the bus training sample, j is the bus training sample label, I represents the multi-dimensional time series data corresponding to each bus in the bus training sample, and U, P, and Q are the voltage data, active power data, and reactive power data of each bus in the bus training sample, respectively. and These are the best matching subsequences corresponding to the p-th and q-th bus lines, respectively.

[0014] Furthermore, determining the bus label corresponding to each bus in the bus training sample based on the bus correlation degree includes: discretizing the bus correlation degree to obtain the bus label.

[0015] Further, obtaining the stability assessment result of the power system voltage based on the bus vector and the distance data includes: inputting the bus vector and the distance data into a pre-trained power system voltage stability assessment model to obtain the stability assessment result.

[0016] Secondly, the present invention also provides a power system voltage stability assessment device, comprising: a first acquisition module, configured to acquire a bus vector corresponding to each bus in the power system based on a pre-trained bus vector representation training model; a second acquisition module, configured to acquire distance data converted from time series data corresponding to each bus in the power system; and an assessment module, configured to obtain a power system voltage stability assessment result based on the bus vector and the distance data.

[0017] Thirdly, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the power system voltage stability assessment method as described in any of the above.

[0018] Fourthly, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the power system voltage stability assessment method as described in any of the above.

[0019] The power system voltage stability assessment method provided by this invention obtains the bus vector corresponding to each bus in the power system, as well as the distance data converted from the time series data corresponding to each bus in the power system. Based on the obtained bus vector and distance data, the power system voltage stability is assessed to obtain the stability assessment result. This method fully utilizes the time series information and spatial structure information reflected by all buses in the power system during voltage instability, explores the correlation between buses, and organically combines the time series information and spatial structure information of instability through representation learning. This can more comprehensively reflect the instability characteristics exhibited by all load buses in the power system during voltage instability, effectively ensuring the accuracy of the power system voltage stability assessment. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in this 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 some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0021] Figure 1A schematic flowchart of the power system voltage stability assessment method provided by the present invention;

[0022] Figure 2 A schematic diagram of the structure of the bus vector representation training model provided by the present invention;

[0023] Figure 3 This is a schematic diagram of the structure of the long-term voltage stability assessment model in the power system provided by the present invention;

[0024] Figure 4 A schematic diagram of the overall process of the power system voltage stability assessment method provided by the present invention;

[0025] Figure 5 A schematic diagram of the structure of the power system voltage stability assessment device provided by the present invention;

[0026] Figure 6 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0028] Figure 1 A schematic flowchart of the power system voltage stability assessment method provided by this invention is shown. Figure 1 As shown, the method includes:

[0029] S101, based on a pre-trained bus vector representation training model, obtains the bus vector corresponding to each bus in the power system;

[0030] S102, Obtain distance data converted from time series data corresponding to each bus in the power system.

[0031] It should be noted that in real-world complex power systems, information about medium- and long-term voltage instability is not only contained in the time series of the dynamic process but also reflected in spatial structure information. Load buses at different spatial locations may exhibit drastically different roles during medium- and long-term voltage instability. Therefore, to more reliably assess medium- and long-term voltage stability, it is necessary to integrate both time series information and spatial structure information.

[0032] Understandably, power systems are complex networks with intricate relationships between nodes, making them difficult to describe directly using precise mathematical models. To analyze the relationships between buses within spatial structural information, uncover the roles of different buses during power system voltage instability, and better prevent voltage instability, power systems can draw upon two major applications of representation learning: word vector representation in natural language processing and node vectorization in social networks. This allows for the representation of buses as multi-dimensional vectors, known as bus vectors.

[0033] Specifically, the bus vector can be obtained by training a model that represents a pre-trained bus vector. This training model can be a deep learning network model or other network models, without any specific limitations here.

[0034] A power system contains multiple buses, each with a corresponding bus vector. This bus vector is multi-dimensional and can specifically represent the proportion of influence of the other buses on that bus during a power system voltage instability process.

[0035] In this step, in addition to obtaining the bus vector that takes into account the spatial structure information of the power system, it is also necessary to obtain the distance data converted from the time series data corresponding to each bus in the power system. Specifically, using the shapelet method, the most representative sub-time series, i.e., the shapelet, is extracted from the time series data corresponding to each bus. Then, based on the sub-time series corresponding to each bus, the time series data of each bus is converted into easily processed structured distance data. Specifically, the Euclidean distance between the time series data and the most representative sub-time series is calculated to obtain distance information that retains key characteristics of voltage instability, i.e., distance data.

[0036] Among them, shapelet is a continuous sub-time series that can reflect category information to the greatest extent in a time series. It is an extension of the nearest neighbor algorithm, which can extract the most representative and typical feature sub-time series from the time series data corresponding to each bus line.

[0037] S103, based on the bus vector and distance data, obtain the stability assessment results of the power system voltage.

[0038] It is understandable that, based on the bus vector corresponding to each bus in the power system obtained in the previous step S101, and the distance data obtained by converting the time series data corresponding to each bus in the power system in step S102, the voltage stability of the power system is evaluated according to the obtained bus vector and distance data, and the stability evaluation result is obtained.

[0039] The stability assessment results of the power system are used to characterize the instability features of the power system voltage. In a specific embodiment, the stability assessment results include two types: voltage stability and voltage instability. Of course, the stability assessment results can also be classified into multiple levels of categories based on actual needs.

[0040] It should be noted that the power system voltage stability assessment method provided in this embodiment of the invention is applicable to the stability assessment of transient voltage and medium- and long-term voltage, and is particularly applicable to the stability assessment of medium- and long-term voltage in power systems.

[0041] In this embodiment, by acquiring the bus vector corresponding to each bus in the power system and the distance data converted from the time-series data corresponding to each bus in the power system, and based on the acquired bus vector and distance data, the voltage stability of the power system is assessed, and the stability assessment result is obtained. This method fully utilizes the time-series information and spatial structure information reflected by all buses in the power system during voltage instability, mines the correlation between buses, and organically combines the temporal and spatial structure information of instability through representation learning. This enables a more comprehensive learning of the instability characteristics exhibited by all load buses in the power system during voltage instability, effectively ensuring the accuracy of the power system voltage stability assessment.

[0042] Based on the above embodiments, the power system voltage stability assessment method further includes: acquiring bus training samples and labeling the bus training samples according to bus labels; using the labeled bus training samples to train the bus vector representation training model to obtain the trained bus vector representation training model.

[0043] Understandably, due to the special nature of power systems, obtaining the bus vector corresponding to each bus in a power system requires improvements based on representation learning. Specifically, this can be achieved by training a model that has been trained to represent the bus vector.

[0044] To obtain a well-trained bus vector representation training model, firstly, bus training samples are acquired. In a specific embodiment, a large number of bus training samples with medium- to long-term voltage stability and / or instability are obtained through simulation. The bus training samples are then labeled according to bus tags. Here, the bus tags are labels used to classify the bus training samples. In a specific embodiment, these randomly extracted bus training samples are labeled with bus tag "1" and bus tag "0".

[0045] After labeling the bus training samples according to the bus labels, the labeled bus vector representation training model is trained using the bus training samples to obtain the trained bus vector representation training model.

[0046] The bus vector represents the training model, which can be a deep learning network model or other network models. In a specific embodiment, the bus vector represents the training model, which includes an input layer, a hidden layer, and an output layer.

[0047] After obtaining the trained bus vector representation training model, the bus vector corresponding to each bus in the power system is obtained, including: obtaining the bus vector corresponding to each bus through the trained bus vector representation model based on the one-hot encoding of each bus; wherein, the one-hot encoding corresponds one-to-one with the bus label, and the bus vector is used to characterize the proportion of the target bus affected by the other buses in the power system during the voltage instability process of the power system.

[0048] Specifically, Figure 2 A schematic diagram of the structure of the bus vector representation training model provided by this invention is shown. Figure 2 As shown, assuming there are m buses in the power system within the target area, the k-th bus can be represented using one-hot encoding as follows:

[0049] bus k =[0 [1] ,0 [2] ,…,0 [k-1] ,1 [k] ,0 [k+1] ,…,0 [m] ]

[0050] The one-hot encoding of the bus line is used as the input to train the model using the bus line vector representation. After passing through a hidden layer containing M neurons, the corresponding bus line vector can be obtained. Then, after passing through a fully connected layer and a softmax function, the corresponding bus line label can be obtained.

[0051] It's important to note here that the mother vector represents the output of the intermediate hidden layer of the training model. During the training process of the mother vector-represented training model, this mother vector representation includes two types of output results: one type is the mother vector output by the intermediate hidden layer containing M neurons. Here, j represents the label of the bus training sample; the other type is the bus label y output by the fully connected layer.

[0052] The bus labels obtained here can be used to further label the bus training samples, resulting in a large number of labeled bus training samples, which enables the bus vector representation training model to be trained better.

[0053] When formally using the bus vector representation to train the model and obtain the bus vector, only the M-dimensional bus vector output by the intermediate hidden layer is obtained.

[0054] The bus vector represents the bus vector output by the intermediate hidden layer of the training model. It represents the proportion of the k-th (k=1,2,…,m) bus affected by the other m-1 buses during the voltage instability process of the power system. The sum of these proportions is 100%.

[0055] In this embodiment, by acquiring bus training samples and labeling them according to bus labels, the labeled bus training samples are used to train the bus vector representation model, resulting in a trained bus vector representation training model. Then, through this trained bus vector representation training model, the bus vectors that reflect the spatial structure information of the bus can be better obtained.

[0056] Based on the above embodiments, further, some bus training samples in the bus training samples are labeled according to the bus labels, which includes: extracting the target time series subsequence corresponding to the multi-dimensional time series based on the multi-dimensional time series corresponding to each bus; obtaining the best matching subsequence in the multi-dimensional time series corresponding to the target time series subsequence; calculating the bus correlation degree between each bus and the other buses in the bus training samples based on the best matching subsequence; and determining the bus label corresponding to each bus in the bus training samples based on the bus correlation degree.

[0057] Understandably, before labeling a portion of the bus training samples based on the bus labels, it is necessary to obtain the bus labels. Specifically, the shapelet method is first used to extract the corresponding target time series subsequence from the multi-dimensional time series of each bus. Here, the multi-dimensional time series refers to the time series of each bus in the U, P, and Q dimensions. The target time series is the subsequence with the largest metric function value. The metric function can specifically be an indicator such as information gain, information gain ratio, or separation coefficient, all of which have a certain ability to distinguish time series.

[0058] Among them, the U-dimensional time series refers to the voltage time series corresponding to the bus, the P-dimensional time series refers to the active power time series corresponding to the bus, and the Q-dimensional time series refers to the reactive power time series corresponding to the bus.

[0059] Accordingly, the target time series subsequences corresponding to the multi-dimensional time series are extracted, including extracting the target time series subsequences corresponding to the U, P, and Q dimensions. For the p-th parent line, the corresponding target time series subsequence can be represented as...

[0060] After extracting the multi-dimensional target time series subsequences, the best matching subsequence corresponding to the target time series subsequence is obtained from the multi-dimensional time series of all the training samples of the parent vector, that is, the subsequence that is closest to the target time series subsequence in the multi-dimensional time series.

[0061] For a training sample j (j = 1, 2, ..., N) of the bus vector, the multi-dimensional time series corresponding to the p-th (p = 1, 2, ..., m) bus is: Where I∈(U,P,Q), then the multidimensional time series The corresponding best matching subsequence is

[0062] Based on the best-matching subsequence, the bus correlation degree between each bus and the other buses in the bus training sample is calculated. Taking the bus correlation degree between the p-th bus and the q-th bus in the bus training sample j as an example, the bus correlation degree can be calculated using the bus correlation degree formula, which is as follows:

[0063]

[0064] Where p and q are the bus labels in bus training sample j, j is the label of bus training sample j, I represents the multi-dimensional time series data corresponding to each bus in bus training sample j, and U, P and Q are the voltage data, active power data and reactive power data of each bus in bus training sample j, respectively. and These are the best-matching subsequences corresponding to the p-th and q-th bus lines, respectively. Dist_Improved(·) uses the formula for calculating the distance between sequences of unequal lengths.

[0065] The aforementioned bus correlation degree refers to the distance between the best matching subsequences of the p-th bus and the q-th bus in the same dimension in the bus training sample j.

[0066] It is understandable that the best-matching subsequence corresponding to a certain dimension of each bus in a sample is regarded as the most representative segment of the time series of that bus in that dimension. That is, for the instability sample, the best-matching subsequence is the time series segment that best reflects the voltage instability characteristics of the power system. Therefore, the distance between the best-matching subsequences of two buses in the same dimension can represent their correlation in the voltage instability process. Taking the distance values ​​of the best-matching subsequences in different dimensions as multi-dimensional vectors and then calculating the Euclidean distance of the multi-dimensional vectors between the two buses is to integrate information from different dimensions and obtain a more effective bus correlation value, that is, the bus correlation degree.

[0067] The training sample j contains m buses. Based on the above, the bus correlation degree between the p-th bus and the remaining m-1 buses in the training sample can be summarized as follows:

[0068] R(p|j)=[R(p,1|j),…,R(p,q-1|j),R(p,q+1|j),…R(p,m-1|j)]

[0069] The smaller the value of R(p|j), the stronger the correlation; the larger the value of R(p|j), the weaker the correlation.

[0070] However, the above R(p|j) cannot be directly used as the bus vector to represent the bus label output by the fully connected layer of the training model. Further, based on the bus correlation degree, the bus label corresponding to each bus is determined, including: discretizing the bus correlation degree to obtain the bus label.

[0071] Specifically, discretization is needed based on the aforementioned R(p|j). This involves setting the position with the smallest value in R(p|j) to 1, and setting all other positions to 0. In other words, in a training sample of a bus line, only the bus line with the highest correlation to bus line p has a label of 1, while the labels of all other buses are zero. This method may seem to lose some information by only considering the bus line with the strongest correlation and equating it with the bus line itself. However, in reality, with a sufficiently large dataset, the differences between bus lines can be fully reflected in a large number of samples.

[0072] In this embodiment, by extracting the target time series subsequence corresponding to the multi-dimensional time series, the best matching subsequence corresponding to the target time series subsequence in the multi-dimensional time series is obtained. Based on the best matching subsequence, the bus correlation degree between each bus and the other buses in the bus training samples is calculated. Then, based on the calculated bus correlation degree, the bus label corresponding to each bus is determined, and the bus training samples are labeled according to the bus label, thereby better training the bus vector representation training model.

[0073] Furthermore, based on the bus vector and distance data, the stability assessment results of the power system voltage are obtained, including: inputting the bus vector and distance data into a pre-trained power system voltage stability assessment model to obtain the stability assessment results.

[0074] Understandably, based on bus vectors and distance data, the stability assessment results of the power system voltage are obtained. Specifically, based on the bus vector corresponding to each bus, the weight value corresponding to that bus is obtained. That is, through the spatial structure information between buses in the power system, the importance of different load buses in the power system voltage instability process is obtained and converted into a practically usable weight value.

[0075] In a specific embodiment, taking the long-term voltage stability assessment in a power system as an example, the process of obtaining the stability assessment results described above can be achieved by constructing a reliable medium- and long-term voltage stability assessment model using a multi-dimensional distance-weighted summation method.

[0076] Figure 3 A schematic diagram of the structure of the long-term voltage stability assessment model for power systems provided by this invention is shown. Figure 3 As shown, the model can be divided into an input layer, a parameter layer, and an output layer.

[0077] Regarding the input layer, its task is to calculate multi-dimensional distance data. Specifically, using the shapelet method, it extracts multi-dimensional sub-time series from massive time series data, namely voltage, active power, and reactive power dimensions. Then, using the distance calculation formula between unequal time series, it calculates the distance from each dimension of the sub-time series to the corresponding dimension of the time series of other buses in the power system. All the obtained distance data is used as input to the long-term voltage stability assessment model of the power system.

[0078] Assuming there are m buses in a power system, the long-term voltage stability assessment model for the power system has m×3 inputs.

[0079] Regarding the parameter layer, this parameter layer takes m bus vectors of dimension M as parameters input into the long-term voltage stability evaluation model of the electronic system. After passing through a one-dimensional convolutional network layer, its dimension becomes M′. Then, after passing through a fully connected layer, a value can be obtained, which is the weight value corresponding to each bus.

[0080] Regarding the output layer, it takes the weight value corresponding to each bus obtained from the parameter layer and multiplies it sequentially with the distance data of the bus in different dimensions from the input layer. After multiplication, the results are summed to obtain a weighted sum. The weighted sum has three values, corresponding to the three dimensions of the bus: voltage dimension U, active power dimension P, and reactive power dimension Q. Then, it passes through a feedforward network to obtain the final output result, which indicates the category to which the corresponding bus belongs, i.e., voltage stable or voltage unstable.

[0081] In one specific embodiment, “0” and “1” represent the category to which the bus belongs, with “0” indicating voltage instability and “1” indicating voltage stability.

[0082] In this embodiment, the stability assessment result is obtained by inputting the bus vector and distance data into a pre-trained power system voltage stability assessment model. Building upon the original model that only considered time-series information, this model organically combines the bus vector, which incorporates spatial structure information, with time-series information. This allows for a more comprehensive learning of the instability characteristics exhibited by all load buses in the power system during voltage instability, effectively ensuring the accuracy of the power system voltage stability assessment.

[0083] in addition, Figure 4 A schematic diagram of the overall process of the power system voltage stability assessment method provided by the present invention is shown.

[0084] like Figure 4 As shown, the voltage stability assessment of a power system mainly includes two aspects:

[0085] On the one hand, it is necessary to obtain the bus vector by training the model through the bus vector representation. This bus vector contains the spatial structure information of each bus in the power system.

[0086] Specifically, a time series dataset, i.e. bus training samples, is obtained, and some samples in the bus training samples are labeled according to the bus labels. Then, the labeled bus training samples are used to train the constructed bus vector representation training model. The one-hot encoding of each bus is used as input, and the bus vector corresponding to each bus is obtained through the trained bus vector representation training model.

[0087] This also includes obtaining labels representing the strength of the correlation between buses, i.e., bus labels. Specifically, it involves obtaining the multi-dimensional time series corresponding to each bus, extracting the target time series subsequence corresponding to the multi-dimensional time series, and obtaining the best matching subsequence corresponding to the target time series subsequence of each dimension. Based on the best matching subsequence, the bus correlation degree between each bus and other buses in the bus training sample in each dimension can be obtained. By combining the bus correlation degree of each bus with other buses in the bus training sample in all dimensions and discretizing the bus correlation degree, the bus label corresponding to each bus can be obtained.

[0088] On the other hand, distance data is obtained by converting the time-series data corresponding to each bus in the power system. Specifically, the multidimensional shapelet method is used to convert the time-series data corresponding to each bus in the power system into easily processed structured distance data, i.e., distance data.

[0089] Finally, by fully combining the bus vector and distance data, the corresponding stability assessment results are obtained.

[0090] This invention, from the perspective of instability, obtains the impact of a single bus voltage drop on the overall regional voltage stability, thereby guiding the deployment of control devices such as reactive power capacitors and static reactive power compensators at different locations in the power grid, and further better preventing the occurrence of power system voltage instability.

[0091] It should be noted that the bus vector representation training model, power system voltage stability assessment model, and power system long-term voltage stability assessment model mentioned in the above embodiments can all be built and trained based on the TenForflow environment. Verification through practical cases shows that the results obtained from the training of these models have very high accuracy. Furthermore, they can reveal the correlation between buses during transient or long-term voltage instability processes, quantifying the impact of each bus on regional voltage instability.

[0092] Figure 5 A schematic diagram of the structure of the power system voltage stability assessment device provided by the present invention is shown. Figure 5 As shown, the device includes: a first acquisition module 501, used to acquire the bus vector corresponding to each bus in the power system based on a pre-trained bus vector representation training model; a second acquisition module 502, used to acquire distance data converted from the time series data corresponding to each bus in the power system; and an evaluation module 503, used to obtain the stability evaluation result of the power system voltage based on the bus vector and the distance data.

[0093] In this embodiment, the acquisition module 501 acquires the bus vector corresponding to each bus in the power system, and the second acquisition module 502 acquires the distance data converted from the time series data corresponding to each bus in the power system. The evaluation module 503 performs a stability evaluation of the power system voltage based on the acquired bus vectors and distance data, and obtains the stability evaluation result. This device fully utilizes the time series information and spatial structure information reflected by all buses in the power system during voltage instability, mines the correlation between buses, and organically combines the time series information and spatial structure information of instability in a representation learning manner. It can more comprehensively learn the instability characteristics exhibited by all load buses in the power system during voltage instability, effectively ensuring the accuracy of the power system voltage stability evaluation.

[0094] The power system voltage stability assessment device provided by this invention can be referred to in correspondence with the power system voltage stability assessment method described above, and will not be repeated here.

[0095] Figure 6 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 6As shown, the electronic device may include a processor 610, a communication interface 620, a memory 630, and a communication bus 640. The processor 610, communication interface 620, and memory 630 communicate with each other via the communication bus 640. The processor 610 can call logical instructions in the memory 630 to execute a power system voltage stability assessment method. This method includes: acquiring the bus vector corresponding to each bus in the power system, and distance data converted from the time series data corresponding to each bus in the power system; and obtaining the power system voltage stability assessment result based on the bus vector and distance data.

[0096] Furthermore, the logical instructions in the aforementioned memory 630 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0097] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When executed by a processor, the computer program implements a method for evaluating the stability of power system voltage provided by the above methods. The method includes: obtaining the bus vector corresponding to each bus in the power system and the distance data converted from the time series data corresponding to each bus in the power system, and obtaining the stability evaluation result of the power system voltage.

[0098] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0099] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0100] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for evaluating the stability of voltage in a power system, characterized in that, include: Based on the pre-trained bus vector representation training model, the bus vector corresponding to each bus in the power system is obtained; Obtain distance data converted from time-series data corresponding to each bus in the power system; Based on the bus vector and the distance data, the stability assessment result of the power system voltage is obtained; This also includes: acquiring bus training samples and labeling the bus training samples according to the bus labels; The bus vector representation training model is trained using labeled bus training samples to obtain a trained bus vector representation training model. The step of labeling the bus training samples according to the bus labels includes, prior to: Based on the multi-dimensional time series corresponding to each bus, extract the target time series subsequence corresponding to the multi-dimensional time series; Obtain the best matching subsequence in the multi-dimensional time series that corresponds to the target time series subsequence; Based on the best matching subsequence, calculate the bus correlation degree between each bus and the other buses in the bus training sample; Based on the bus correlation degree, determine the bus label corresponding to each bus in the bus training sample; Wherein, the target time series subsequence is the subsequence with the largest metric function value; The process of obtaining the bus vector corresponding to each bus in the power system includes: Based on the one-hot encoding corresponding to each bus, the bus vector corresponding to that bus is obtained by training the model through the trained bus vector representation; The unique hot code corresponds one-to-one with the bus tag, and the bus vector is used to characterize the proportion of the target bus affected by the other buses in the power system during the voltage instability process of the power system.

2. The method for assessing the stability of power system voltage according to claim 1, characterized in that, The step of calculating the bus correlation degree between each bus and the other buses in the bus training sample based on the best matching subsequence includes: The bus correlation degree is calculated using the bus correlation degree formula, which is as follows: ； in, and The label of the bus in the bus training sample. The label of the training sample for the bus is [number]. This represents the multi-dimensional time series data corresponding to each bus in the bus training samples. , and These are the voltage data, active power data, and reactive power data for each bus in the bus training sample, respectively. and The first The busbar and the first The best matching subsequence corresponding to each busbar The formula used is the one for calculating the distance between sequences of unequal time durations.

3. The method for assessing the stability of power system voltage according to claim 1, characterized in that, The step of determining the bus label corresponding to each bus in the bus training sample based on the bus correlation degree includes: The bus correlation degree is discretized to obtain the bus label.

4. The method for assessing the stability of power system voltage according to any one of claims 1-3, characterized in that, The step of obtaining the stability assessment result of the power system voltage based on the bus vector and the distance data includes: The bus vector and the distance data are input into a pre-trained power system voltage stability assessment model to obtain the stability assessment result.

5. A device for evaluating the stability of voltage in a power system, characterized in that, include: The first acquisition module is used to acquire the bus vector corresponding to each bus in the power system based on the pre-trained bus vector representation training model. The second acquisition module is used to acquire distance data converted from the time series data corresponding to each bus in the power system; An evaluation module is used to obtain a stability evaluation result of the power system voltage based on the bus vector and the distance data; This also includes: acquiring bus training samples and labeling the bus training samples according to the bus labels; The bus vector representation training model is trained using labeled bus training samples to obtain a trained bus vector representation training model. The step of labeling the bus training samples according to the bus labels includes, prior to: Based on the multi-dimensional time series corresponding to each bus, extract the target time series subsequence corresponding to the multi-dimensional time series; Obtain the best matching subsequence in the multi-dimensional time series that corresponds to the target time series subsequence; Based on the best matching subsequence, calculate the bus correlation degree between each bus and the other buses in the bus training sample; Based on the bus correlation degree, determine the bus label corresponding to each bus in the bus training sample; Wherein, the target time series subsequence is the subsequence with the largest metric function value; The process of obtaining the bus vector corresponding to each bus in the power system includes: Based on the one-hot encoding corresponding to each bus, the bus vector corresponding to that bus is obtained by training the model through the trained bus vector representation; The unique hot code corresponds one-to-one with the bus tag, and the bus vector is used to characterize the proportion of the target bus affected by the other buses in the power system during the voltage instability process of the power system.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the power system voltage stability assessment method as described in any one of claims 1 to 4.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the power system voltage stability assessment method as described in any one of claims 1 to 4.

Images