WOA-VMD algorithm-based broadband oscillation positioning method

Abstract

The invention proposes a broadband oscillation positioning method based on a WOA-VMD algorithm, and relates to the technical field of power system optimization control, and the method comprises the steps: S1, obtaining an original oscillation signal, and decomposing the original oscillation signal sequence into a plurality of mode components; s2, each mode component is optimized through a WOA algorithm, and broadband oscillation noise in each mode component is removed through a VMD algorithm; s3, constructing a broadband oscillation positioning model based on a GAT algorithm, and performing offline training through historical sample data; s4, inputting each mode component after the broadband oscillation noise is removed into the broadband oscillation positioning model after offline training is completed so as to perform accurate positioning on an oscillation area; according to the broadband oscillation positioning method based on the WOA-VMD algorithm, the WOA-VMD algorithm is combined with the graph neural network, accurate positioning of the oscillation area is achieved, and the anti-noise capability is remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power system optimization control, and in particular to a broadband oscillation positioning method based on a WOA-VMD algorithm. Background Art

[0002] Under the guidance of the "dual carbon" goal, my country is accelerating the construction of a new power system with renewable energy as the main body. The construction of a new power system with new energy as the main body is of great significance. The reverse distribution of power generation bases and load centers poses a challenge to the full absorption of new energy. Due to long-distance transmission, the power grid presents the characteristics of a "weak grid". New broadband oscillations caused by the integration of new energy systems into weak grids have occurred frequently in recent years. The randomness of new energy output and the interaction between power electronic devices will lead to machine-grid coupling oscillations. The oscillations generated when large-scale new energy is connected to the grid show a broadband trend, which poses a major threat to the safe and stable operation of the power system and seriously affects the absorption of new energy. At present, new energy clusters are equipped with energy storage devices with fast response speed and high power density. Because they have the ability to provide virtual inertia to the system and improve voltage stability in a short time scale, it is particularly important to accurately locate their oscillation points and quickly take suppression measures.

[0003] At present, the main research methods for broadband oscillation at home and abroad are: eigenvalue analysis method, impedance modeling method, open-loop resonance method, etc. The eigenvalue analysis method determines the stability of the system by calculating the eigenvalue of the system state matrix. The impedance modeling method determines the stability of the system by analyzing the impedance characteristics of the system and can be used to design a controller to change the impedance characteristics of the system, thereby improving the stability of the system; the open-loop resonance method can identify the inherent oscillation mode and its characteristics of the system by testing and measuring the frequency response function of the system under open-loop conditions. Although this method can describe the dynamic behavior information of the oscillation in detail, it requires a long calculation time, so it is difficult to meet the needs of rapid positioning of the oscillation point. In addition, in recent years, the broadband oscillation of new power systems has shown the characteristics of multi-modal coupling. Due to the continuous random changes in the load, the randomness and intermittency of new energy generation, and the frequent switching of power electronic devices, the broadband oscillation signal is seriously polluted by noise, which may make the low-frequency oscillation hidden in the noise and undetectable. Summary of the invention

[0004] The purpose of the present invention is to provide a broadband oscillation positioning method based on the WOA-VMD algorithm, which can solve the problems mentioned in the background technology, achieve accurate positioning of the oscillation area, and significantly improve the anti-noise ability.

[0005] The technical solution of the present invention is:

[0006] In a first aspect, the present application provides a broadband oscillation positioning method based on the WOA-VMD algorithm, which comprises the following steps:

[0007] S1, obtaining an original oscillation signal, and decomposing the original oscillation signal sequence into multiple mode components;

[0008] S2, optimize each mode component by WOA algorithm, and remove broadband oscillation noise in each mode component by VMD algorithm;

[0009] S3, build a broadband oscillation positioning model based on the GAT algorithm, and perform offline training using historical sample data;

[0010] S4. Input each pattern component after removing the broadband oscillation noise into the broadband oscillation positioning model after offline training to accurately locate the oscillation area.

[0011] Furthermore, step S1 further includes: setting a constraint condition for the estimated bandwidth of each mode component, wherein the constraint change expression is:

[0012]

[0013] st∑ k u k =f

[0014] Introduce the Lagrangian operator to solve the above expression to obtain the optimal solution:

[0015]

[0016] In the formula, u k is the kth mode component after VMD decomposition, w k is the instantaneous frequency of the kth mode component, To find the partial derivative, t is the time constant, δ(t) represents the impulse function, j is the imaginary part of the imaginary number, and u k (t) is a step function, is the Laplace operator, the time domain is transformed into the complex frequency domain, f represents the original signal input, L is the Lagrangian operator, λ is the control parameter, and α is the penalty factor.

[0017] Furthermore, in step S2, the calculation process of optimizing each mode component by the WOA algorithm includes:

[0018] When p<0.5 and |A|<1:

[0019]

[0020] When p<0.5 and |A|≥1:

[0021]

[0022] When p ≥ 0.5:

[0023]

[0024] In the formula, p and l are random numbers between [0,1], and D j is the best search agent position length vector of the whale in the j-th dimension space, D is the position length vector, A and C are both control coefficients, X j* (t) is the optimal vector of the current search agent position, t is the current iteration number, t max is the maximum number of iterations, X j (t) is the current best search agent position vector, α is the penalty factor, r is a random number in [0, 1], D rand is the random individual position vector, X rand j (t) is the random individual position vector, and b is the spiral shape constant.

[0025] Furthermore, in step S2, the calculation formula for removing the broadband oscillation noise in each mode component using the VMD algorithm includes:

[0026]

[0027] Where P k is the ratio of the kth individual power spectrum value to the total power spectrum, X(k) is the discrete Fourier transform corresponding to the kth modal component x(k), k is the number of samples, N is the signal length, and H is the power spectrum entropy.

[0028] Furthermore, in step S3, the calculation process of constructing a broadband oscillation positioning model based on the GAT algorithm includes:

[0029] GAT graph attention layer:

[0030]

[0031] GAT output layer:

[0032]

[0033] Attention coefficient:

[0034]

[0035] Use the softmax function to normalize the attention coefficient:

[0036]

[0037] Merge into a complete attention mechanism using LeakyReLu as the activation function:

[0038]

[0039] The new features of each node that integrate domain information are calculated by weighted summation:

[0040]

[0041] The active power time series information P of the node is used as the input feature, and the adjacency matrix A of the power network topology structure is input into GAT as the spatial information:

[0042]

[0043]

[0044] The mapping relationship between the oscillation node position label and the feature matrix P is learned to obtain the trained broadband oscillation positioning model:

[0045]

[0046] In the formula, h is the vector set of node features, h i is the initial feature vector of node i, N is the number of system nodes, F is the number of node features, is a feature set, h′ is a vector set of new node features, is the new initial feature vector of node i, F′ is the number of new node features, is the new feature set, W is the weight matrix, e ij 、e ik is the attention coefficient, a is the learnable weight vector, || is the concatenation of the feature matrix, is the initial eigenvector of node j, j is the node adjacent to i, is the set of characteristic matrices of order F'×F, α ij is the normalized attention coefficient, K is the number of sampling points, N i is the number of the i-th node, is the vector to be learned, is the kth new node feature, is the new node feature, σ is the activation function, n represents the number of nodes, p t，n is the power of n nodes at time t, k is the number of samples, is the oscillation source position label, H GAT (·) is the mapping function between the oscillation source position label and the feature matrix Y, where Y is the feature matrix, X is the actual system quantity measurement inThe relationship function between X and the feature matrix Y in It is the actual system quantity measurement.

[0047] Furthermore, step S3 also includes: introducing a weighted cross entropy loss function to perform a balanced calculation on the historical sample data, and the calculation formula is:

[0048]

[0049] In the formula, loss(p) is the cross entropy loss function, M is the number of samples, n is the number of nodes, where n=3, refers to the label value, and y k,0 ,y k,n For Boolean values, p k,0 、p k,n The probability value of sample k belonging to each type of label is given to the discriminator, k is the number of samples, w0 is the weight of the sample when it is stable, and w n is the weight when the sample label is n.

[0050] In a second aspect, the present application provides an electronic device, including:

[0051] A memory for storing one or more programs;

[0052] processor;

[0053] When the one or more programs are executed by the processor, a broadband oscillation positioning method based on the WOA-VMD algorithm as described in any one of the first aspects is implemented.

[0054] In a third aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a broadband oscillation positioning method based on the WOA-VMD algorithm as described in any one of the first aspects above.

[0055] Compared with the prior art, the present invention has at least the following advantages or beneficial effects:

[0056] The invention discloses a broadband oscillation positioning method based on the WOA-VMD algorithm. The method decomposes the original oscillation signal sequence into multiple mode components, optimizes each mode component by using the WOA algorithm, removes broadband oscillation noise in each mode component by using the VMD algorithm, and performs online monitoring and evaluation in combination with the GAT model. The method realizes accurate positioning of the oscillation area of ​​the broadband oscillation signal, and significantly improves the anti-noise ability of the broadband oscillation signal. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0058] Figure 1 The present invention provides a broadband oscillation positioning method based on the WOA-VMD algorithm;

[0059] Figure 2 It is a schematic diagram of WOA envelope entropy iteration curve;

[0060] Figure 3 This is the time-frequency domain diagram of IMF after decomposition based on the WOA-VMD algorithm;

[0061] Figure 4 This is a schematic diagram of the comparison of broadband oscillation signals before and after denoising;

[0062] Figure 5 It is a schematic structural diagram of the broadband oscillation positioning model based on GAT;

[0063] Figure 6 The figure is a schematic structural block diagram of an electronic device according to an embodiment of the present invention.

[0064] Icon: 101, memory; 102, processor; 103, communication interface. DETAILED DESCRIPTION

[0065] Terminology explanation:

[0066] WOA: Whale Optimization Algorithm, a meta-heuristic optimization algorithm inspired by the predation behavior of humpback whales in nature;

[0067] VMD: Variational Mode Decomposition, an advanced time-frequency analysis method;

[0068] IMF: Intrinsic Mode Functions, the signal components of each layer obtained after the original signal is decomposed by EMD (Empirical Mode Decomposition).

[0069] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations.

[0070] Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the present application for which protection is sought, but merely represents selected embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0071] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0072] It should be noted that, in this article, the term "comprises" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprises..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0073] In conjunction with the accompanying drawings, some embodiments of the present application are described in detail below. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0074] Example 1

[0075] See also Figure 1 , Figure 1 The figure shows a step diagram of a broadband oscillation positioning method based on the WOA-VMD algorithm provided in an embodiment of the present application.

[0076] In a first aspect, the present application provides a broadband oscillation positioning method based on the WOA-VMD algorithm, which comprises the following steps:

[0077] S1, obtaining an original oscillation signal, and decomposing the original oscillation signal sequence into multiple mode components;

[0078] S2, optimize each mode component by WOA algorithm, and remove broadband oscillation noise in each mode component by VMD algorithm;

[0079] S3, build a broadband oscillation positioning model based on the GAT algorithm, and perform offline training using historical sample data;

[0080] S4. Input each pattern component after removing the broadband oscillation noise into the broadband oscillation positioning model after offline training to accurately locate the oscillation area.

[0081] As a preferred implementation, step S1 further includes: setting a constraint condition for the estimated bandwidth of each mode component, wherein the constraint change expression is:

[0082]

[0083] st∑ k u k =f

[0084] Introduce the Lagrangian operator to solve the expression to obtain the optimal solution:

[0085]

[0086] In the formula, u k is the kth mode component after VMD decomposition, w k is the instantaneous frequency of the kth mode component, To find the partial derivative, t is the time constant, δ(t) represents the impulse function, j is the imaginary part of the imaginary number, and u k (t) is a step function, is the Laplace operator, the time domain is transformed into the complex frequency domain, f represents the original signal input, L is the Lagrangian operator, λ is the control parameter, and α is the penalty factor.

[0087] It should be noted that the VMD algorithm redefines the restriction conditions of the bandwidth-limited intrinsic mode function, and its inner-dimensional modal components are defined as the component modal functions of AM and FM, which improves the end effect and mode component aliasing problems in the EMD algorithm, thereby effectively decomposing the non-smooth sequence; the original oscillation signal sequence is decomposed into multiple mode components, each of which has a specific center frequency and is of limited bandwidth. On the premise of ensuring that the original signal is consistent with the sum of all mode components, the estimated bandwidth of each mode component is made as small as possible, thereby creating a time constraint, and then introducing the Lagrangian operator and the quadratic penalty term to transform the constrained variational model into an unconstrained variational model.

[0088] As a preferred implementation, in step S2, the calculation process of optimizing each mode component by the WOA algorithm includes:

[0089] When p < 0.5 and |A| < 1, the algorithm is a shrinking and surrounding mechanism:

[0090]

[0091] When p < 0.5 and |A| ≥ 1, the algorithm is a random search mechanism:

[0092]

[0093] When p ≥ 0.5, the spiral contraction mechanism is selected:

[0094]

[0095] In the formula, p and l are random numbers between [0,1], and D j is the best search agent position length vector of the whale in the j-th dimension space, D is the position length vector, A and C are both control coefficients, X j* (t) is the optimal vector of the current search agent position, t is the current iteration number, t max is the maximum number of iterations, X j (t) is the current best search agent position vector, α is the penalty factor, r is a random number in [0, 1], D rand is the random individual position vector, X rand j (t) is the random individual position vector, and b is the spiral shape constant.

[0096] It should be noted that the WOA optimization algorithm can not only find the optimal solution in a complex multi-dimensional space by simulating the social behavior and predation mechanism of humpback whales, but also avoid falling into local optimality, thereby achieving better results; the process of finding the optimal solution is to determine whether the algorithm reaches the threshold. If so, the final optimal solution and the corresponding fitness value are output; otherwise, the formula value is updated and iteration continues.

[0097] As a preferred implementation, in step S2, the calculation formula for removing the broadband oscillation noise in each mode component using the VMD algorithm includes:

[0098]

[0099] Where P k is the ratio of the kth individual power spectrum value to the total power spectrum, X(k) is the discrete Fourier transform corresponding to the kth modal component x(k), k is the number of samples, N is the signal length, and H is the power spectrum entropy.

[0100] It should be noted that the present invention uses power spectrum entropy as the fitness function of WOA. In the t-times iteration process, the broadband oscillation signal is decomposed into k IMF components. The VMD denoising implementation steps using the WOA optimization algorithm are as follows:

[0101] 1) Use WOA to optimize VMD parameters. In this process, the parameter combination to be optimized in the VMD algorithm is regarded as the position of the whale individual in the search space. Based on this position, VMD decomposition is performed on the input signal, and the corresponding power spectrum entropy value is calculated accordingly. At the same time, the corresponding whale individual position when the power spectrum entropy value reaches the minimum value is recorded. As the iterative process progresses, the change trend of the fitness value will gradually become gentle. Finally, after reaching the preset termination condition, the optimal position at this time is output. The iterative curve of the envelope entropy is as follows: Figure 2 As shown, it converged after the 8th iteration, and the final envelope entropy value was 7.4341;

[0102] 2) Using the optimal parameter model obtained in the above optimization process to perform VMD decomposition on the input signal, so as to obtain the corresponding k intrinsic IMF components such as Figure 3 As shown;

[0103] 3) The Pearson correlation coefficient method is used to screen each IMF component, and the noise component is removed according to the set correlation coefficient threshold to retain the effective component;

[0104] 4) Remove the signals with correlation coefficients lower than 0.6 in the original signal, and reconstruct the signals with correlation coefficients higher than 0.6; the correlation coefficients of the IMF components are shown in Table 1:

[0105] Table 1 Correlation coefficient values ​​of various IMFs

[0106]

[0107] 5) Finally, the effective IMF component is reconstructed to obtain the denoised broadband oscillation signal; Figure 4 As shown in the figure, the waveform of the signal after noise reduction is compared with that of the original signal in the time domain. From the overall waveform, the signal after noise reduction and the original signal show a very high similarity. After amplifying the local waveform, the detailed matching between the two is more obvious. This result proves that the method adopted by the present invention not only effectively reduces the noise, but also well maintains the key oscillation characteristics of the original signal.

[0108] As a preferred implementation, in step S3, the calculation process of constructing a broadband oscillation positioning model based on the GAT algorithm includes:

[0109] GAT graph attention layer:

[0110]

[0111] GAT output layer:

[0112]

[0113] Attention coefficient:

[0114]

[0115] Use the softmax function to normalize the attention coefficient:

[0116]

[0117] Merge into a complete attention mechanism using LeakyReLu as the activation function:

[0118]

[0119] The new features of each node that integrate domain information are calculated by weighted summation:

[0120]

[0121] The active power time series information P of the node is used as the input feature, and the adjacency matrix A of the power network topology structure is input into GAT as the spatial information:

[0122]

[0123]

[0124] The mapping relationship between the oscillation node position label and the feature matrix P is learned to obtain the trained broadband oscillation positioning model:

[0125]

[0126] In the formula, h is the vector set of node features, h i is the initial feature vector of node i, N is the number of system nodes, F is the number of node features, is a feature set, h′ is a vector set of new node features, is the new initial feature vector of node i, F′ is the number of new node features, is the new feature set, W is the weight matrix, e ij 、e ik is the attention coefficient, a is the learnable weight vector, || is the concatenation of the feature matrix, is the initial eigenvector of node j, j is the node adjacent to i, is the set of characteristic matrices of order F'×F, α ij is the normalized attention coefficient, K is the number of sampling points, N i is the number of the i-th node, is the vector to be learned, is the kth new node feature, is the new node feature, σ is the activation function, n represents the number of nodes, pt，n is the power of n nodes at time t, k is the number of samples, is the oscillation source position label, H GAT (·) is the mapping function between the oscillation source position label and the feature matrix Y, where Y is the feature matrix, X is the actual system quantity measurement in The relationship function between X and the feature matrix Y in It is the actual system quantity measurement.

[0127] It should be noted that the overall structure of the broadband oscillation positioning model based on GAT is as follows: Figure 5 As shown, there are 4 layers in total. The model design of this application adopts a 4-layer dimensionality reduction structure. The first 3 layers are GAT layers, which realize the feature aggregation and extraction of the power grid topology connection relationship, and the last layer is a fully connected layer (FC) to realize the whole process from acquiring feature information to oscillation positioning. The input feature dimension is reduced after passing through each layer of the broadband oscillation positioning model, which can map the extraction process of the oscillation feature value and provide a certain reference for adjusting the model structure.

[0128] As a preferred implementation, step S3 further includes: introducing a weighted cross entropy loss function to perform a balanced calculation on the historical sample data, and the calculation formula is:

[0129]

[0130] In the formula, loss(p) is the cross entropy loss function, M is the number of samples, n is the number of nodes, where n=3, refers to the label value, and y k,0 ,y k,n For Boolean values, p k,0 、p k,n The probability value of sample k belonging to each type of label is given to the discriminator, k is the number of samples, w0 is the weight of the sample when it is stable, and w n is the weight when the sample label is n.

[0131] Example 2

[0132] See also Figure 6 , Figure 6 A schematic structural block diagram of an electronic device provided in an embodiment of the present application.

[0133] An electronic device includes a memory 101, a processor 102 and a communication interface 103, wherein the memory 101, the processor 102 and the communication interface 103 are electrically connected to each other directly or indirectly to realize data transmission or interaction. For example, these components can be electrically connected to each other through one or more communication buses or signal lines. The memory 101 can be used to store software programs and modules, and the processor 102 executes various functional applications and data processing by executing the software programs and modules stored in the memory 101. The communication interface 103 can be used to communicate signaling or data with other node devices.

[0134] Among them, the memory 101 can be but is not limited to, random access memory (Random Access Memory, RAM), read-only memory (Read Only Memory, ROM), programmable read-only memory (Programmable Read-Only Memory, PROM), erasable read-only memory (Erasable Programmable Read-Only Memory, EPROM), electrically erasable read-only memory (Electric Erasable Programmable Read-Only Memory, EEPROM), etc.

[0135] The processor 102 may be an integrated circuit chip with signal processing capability. The processor 102 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0136] It is understood that the structure shown in the figure is only for illustration, and a broadband oscillation positioning method based on the WOA-VMD algorithm may also include more or fewer components than those shown in the figure, or have a different configuration than those shown in the figure. Each component shown in the figure may be implemented by hardware, software or a combination thereof.

[0137] In the embodiments provided in the present application, it should be understood that the disclosed method can also be implemented in other ways. The embodiments described above are merely schematic, for example, the flowchart or block diagram in the accompanying drawings shows the possible implementation architecture, functions and operations of the method and computer program product according to multiple embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the module, a program segment or a part of a code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or the flowchart, and the combination of the boxes in the block diagram and / or the flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.

[0138] In addition, the functional modules in the various embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

[0139] If the functions are implemented in the form of software function modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0140] In summary, the embodiment of the present application provides a broadband oscillation positioning method based on the WOA-VMD algorithm. First, the original oscillation signal sequence is decomposed into multiple mode components, and each mode component after decomposition has a specific center frequency and a limited bandwidth. Then, a constraint change expression is set. Under the premise of ensuring that the original signal is consistent with the sum of all mode components, the estimated bandwidth of each mode component is made as small as possible. Secondly, the Lagrangian operator is introduced to solve the constraint conditions. When using the VMD method to remove broadband oscillation noise, it is necessary to preset the decomposition modal parameter K and the penalty factor α. The appropriate parameters are selected through the WOA optimization algorithm. In the t-time iteration process, the broadband oscillation signal is decomposed into k IMF components. Finally, the broadband oscillation positioning model constructed by the GAT algorithm is trained offline on historical data samples and the node information monitored in real time is evaluated online, so as to judge the stable state of each node of the system after the disturbance and realize accurate positioning of the oscillation area.

[0141] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

[0142] It will be apparent to those skilled in the art that the present application is not limited to the details of the exemplary embodiments described above, and that the present application can be implemented in other specific forms without departing from the spirit or essential features of the present application. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the present application is defined by the appended claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims be included in the present application. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A broadband oscillation positioning method based on WOA-VMD algorithm, characterized in that: The following steps are involved: S1, obtaining an original oscillation signal, and decomposing the original oscillation signal sequence into multiple mode components; S2, optimize each mode component by WOA algorithm, and remove broadband oscillation noise in each mode component by VMD algorithm; S3, build a broadband oscillation positioning model based on the GAT algorithm, and perform offline training using historical sample data; S4. Input each pattern component after removing the broadband oscillation noise into the broadband oscillation positioning model after offline training to accurately locate the oscillation area.

2. A broadband oscillation positioning method based on the WOA-VMD algorithm as claimed in claim 1, characterized in that: Step S1 also includes: The constraint condition of the estimated bandwidth of each mode component is set, and the constraint change expression is: Introduce the Lagrangian operator to solve the above expression to obtain the optimal solution: In the formula, u k is the kth mode component after VMD decomposition, w k is the instantaneous frequency of the kth mode component, To find the partial derivative, t is the time constant, δ(t) represents the impulse function, j is the imaginary part of the imaginary number, and u k (t) is a step function, is the Laplace operator, the time domain is transformed into the complex frequency domain, f represents the original signal input, L is the Lagrangian operator, λ is the control parameter, and α is the penalty factor.

3. A broadband oscillation positioning method based on WOA-VMD algorithm as claimed in claim 1, characterized in that: In step S2, the calculation process of optimizing each mode component by the WOA algorithm includes: When p<0.5 and |A|<1: When p<0.5 and |A|≥1: When p ≥ 0.5: In the formula, p and l are random numbers between [0,1], and D j is the best search agent position length vector of the whale in the j-th dimension space, D is the position length vector, A and C are both control coefficients, X j* (t) is the optimal vector of the current search agent position, t is the current iteration number, t max is the maximum number of iterations, X j (t) is the current best search agent position vector, α is the penalty factor, r is a random number in [0, 1], D rand is the random individual position vector, X rand j (t) is the random individual position vector, and b is the spiral shape constant.

4. A broadband oscillation positioning method based on WOA-VMD algorithm as claimed in claim 3, characterized in that: In step S2, the calculation formula for removing the broadband oscillation noise in each mode component using the VMD algorithm includes: Where P k is the ratio of the kth individual power spectrum value to the total power spectrum, X(k) is the discrete Fourier transform corresponding to the kth modal component x(k), k is the number of samples, N is the signal length, and H is the power spectrum entropy.

5. A broadband oscillation positioning method based on WOA-VMD algorithm as claimed in claim 1, characterized in that: In step S3, the calculation process of constructing a broadband oscillation positioning model based on the GAT algorithm includes: GAT graph attention layer: GAT output layer: Attention coefficient: Use the softmax function to normalize the attention coefficient: Merge into a complete attention mechanism using LeakyReLu as the activation function: The new features of each node that integrate domain information are calculated by weighted summation: The active power time series information P of the node is used as the input feature, and the adjacency matrix A of the power network topology structure is input into GAT as the spatial information: The mapping relationship between the oscillation node position label and the feature matrix P is learned to obtain the trained broadband oscillation positioning model: In the formula, h is the vector set of node features, h i is the initial feature vector of node i, N is the number of system nodes, F is the number of node features, is a feature set, h′ is a vector set of new node features, is the new initial feature vector of node i, F′ is the number of new node features, is the new feature set, W is the weight matrix, e ij 、e ik is the attention coefficient, a is the learnable weight vector, || is the concatenation of the feature matrix, is the initial eigenvector of node j, j is the node adjacent to i, is the set of characteristic matrices of order F'×F, α ij is the normalized attention coefficient, K is the number of sampling points, N i is the number of the i-th node, is the vector to be learned, is the kth new node feature, is the new node feature, σ is the activation function, n represents the number of nodes, p t，n is the power of n nodes at time t, k is the number of samples, is the oscillation source position label, H GAT (·) is the mapping function between the oscillation source position label and the feature matrix Y, where Y is the feature matrix, X is the actual system quantity measurement in The relationship function between X and the feature matrix Y in It is the actual system quantity measurement.

6. A broadband oscillation positioning method based on WOA-VMD algorithm as claimed in claim 5, characterized in that: Step S3 also includes: introducing a weighted cross entropy loss function to perform a balanced calculation on the historical sample data, and the calculation formula is: In the formula, loss(p) is the cross entropy loss function, M is the number of samples, n is the number of nodes, where n=3, refers to the label value, and y k,0 ,yk ,n For Boolean values, p k,0 、p k,n The probability value of sample k belonging to each type of label is given to the discriminator, k is the number of samples, w0 is the weight of the sample when it is stable, and w n is the weight when the sample label is n.

7. An electronic device, characterized in that: include: A memory for storing one or more programs; processor; When the one or more programs are executed by the processor, a broadband oscillation positioning method based on the WOA-VMD algorithm as described in any one of claims 1 to 6 is implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, a broadband oscillation positioning method based on a WOA-VMD algorithm as described in any one of claims 1 to 6 is implemented.

Images