Distributed photovoltaic clustering equivalent modeling method of LSO-GRU-ECM

Through the improved K-means algorithm and lion group optimization algorithm, the target clustering equivalent model of distributed photovoltaic clusters is constructed, which solves the problem of traditional modeling methods ignoring the difference in operating status and parameter, and realizes efficient and accurate distributed photovoltaics Cluster dynamic feature fitting.

CN118965047APending Publication Date: 2024-11-15GUANGDONG POWER GRID CO LTD +1

Patent Information

Application Number
CN202411066446.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-05
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

The traditional distributed photovoltaic cluster modeling method ignores the operating status and parameter differences of the cluster itself, is difficult to generalize, and cannot fit the dynamic characteristics of the cluster well.

Method used

The improved K-means algorithm is used for clustering and clustering, combined with the lion group optimization algorithm, the gated recurrent neural network hyperparameters are optimized, and the error correction model is constructed, and the dynamic response error is compensated to generate the target clustering equivalent model.

Benefits of technology

It significantly improves the generalization ability and accuracy of the distributed photovoltaic clustering model, reduces simulation time, and corrects model errors with high accuracy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118965047A_ABST
    Figure CN118965047A_ABST
Patent Text Reader

Abstract

The invention provides a distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM, and the method comprises the steps: carrying out the clustering of a distributed photovoltaic system through employing an improved K-means algorithm, and constructing an initial clustering equivalent model; determining a dynamic response error between a distributed photovoltaic original model and the initial clustering equivalent model, wherein the original model is obtained by parameters and topology of a photovoltaic system; optimizing hyper-parameters of the gated recurrent neural network by using a lion group optimization algorithm, and inputting the dynamic response error into the gated recurrent neural network for training to obtain an error correction model; and inputting the dynamic response error into the error correction model to obtain a compensation value, and compensating the compensation value to the initial clustering equivalent model to obtain a target clustering equivalent model for analyzing operation parameters of the distributed photovoltaic system, and the target clustering equivalent model obtained by the method can simulate actual distributed photovoltaic characteristics with high precision and low time consumption.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of distributed photovoltaic equivalent modeling, and in particular to a distributed photovoltaic cluster equivalent modeling method based on LSO-GRU-ECM, a distributed photovoltaic cluster equivalent modeling device based on LSO-GRU-ECM, a computer-readable storage medium and an electronic device. Background Art

[0002] At present, there are a lot of large-scale studies on photovoltaic system equivalent modeling in the existing technology, which provides a theoretical basis for the construction of distributed photovoltaic grid-connected equivalent models. Generally speaking, photovoltaic system equivalent modeling includes single-machine cluster equivalent and multi-machine cluster equivalent. In actual simulation analysis, the photovoltaic model constructed by multi-machine equivalent modeling can better fit the actual situation, reflect the actual operating status of the photovoltaic cluster, and has the characteristics of short simulation time and high accuracy.

[0003] Among the existing clustering methods, existing research mainly uses cluster analysis algorithms, complex network community discovery algorithms, etc. Existing research constructs comprehensive performance indicators, uses genetic algorithms for clustering; uses dynamic nearest neighbor algorithms, which are suitable for clustering in big data situations; and uses K-means clustering algorithms to perform spatially related clustering analysis on distributed photovoltaic clusters.

[0004] The distributed photovoltaic clustering equivalent model is constructed based on the clustering equivalent method, which improves the accuracy of the model and reduces the simulation time to a certain extent. However, it ignores the operating status and parameter differences of the distributed photovoltaic cluster itself, and the equivalent clustering model constructed based on the inherent characteristics of the clustering algorithm is difficult to generalize. In addition, the distributed photovoltaic cluster is a complex and strongly nonlinear system, and the traditional modeling method cannot fit the dynamic characteristics of the distributed photovoltaic cluster well. Summary of the invention

[0005] The main purpose of this application is to provide a distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM, a distributed photovoltaic clustering equivalent modeling device based on LSO-GRU-ECM, a computer-readable storage medium and an electronic device, so as to at least solve the problem that the traditional modeling method ignores the operating status and parameter differences of the distributed photovoltaic cluster itself, is difficult to generalize, and cannot well fit the dynamic characteristics of the distributed photovoltaic cluster.

[0006] In order to achieve the above-mentioned purpose, according to one aspect of the present application, a distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM is provided, including: clustering and grouping the distributed photovoltaic system using an improved K-means algorithm to construct an initial clustering equivalent model of the distributed photovoltaic system; determining the dynamic response error between the distributed photovoltaic original model and the initial clustering equivalent model, wherein the distributed photovoltaic original model is obtained according to the parameters and topology of the distributed photovoltaic system; optimizing the hyperparameters of the gated recurrent neural network using a lion group optimization algorithm, and inputting the dynamic response error into the gated recurrent neural network for training to obtain an error correction model; inputting the dynamic response error into the error correction model to obtain a compensation value, and compensating the compensation value to the initial clustering equivalent model to obtain a target clustering equivalent model, wherein the target clustering equivalent model is used to analyze the operating parameters of the distributed photovoltaic system.

[0007] Optionally, an improved K-means algorithm is used to cluster the distributed photovoltaic system and construct an initial clustering equivalence model, including: selecting distributed photovoltaic clustering indicators; combining the firefly algorithm with the K-means algorithm to obtain the improved K-means algorithm, and using the improved K-means algorithm to equate the entire distributed photovoltaic system to one or more photovoltaics; performing parameter calculations on each equivalent photovoltaic to construct the initial clustering equivalence model.

[0008] Optionally, selecting a distributed photovoltaic clustering index includes: determining the distributed photovoltaic clustering index based on the distributed photovoltaic original model and according to a zero-pole expression of a system closed-loop transfer function, wherein the distributed photovoltaic clustering index vector is a three-dimensional vector X=[z, p1, p2], and the zero-pole expression of the system closed-loop transfer function is: Among them, L f is the filter inductance; k p is the proportional control constant; k i is the integral control constant.

[0009] Optionally, the distributed photovoltaic system is clustered using an improved K-means algorithm, including: normalizing the distributed photovoltaic related data as the input data set X of the algorithm, and randomly selecting K different photovoltaic data from the data set X as a group of initial centers of the firefly algorithm, i.e., the first generation individuals P, wherein, x nd represents the data of the dth period of the nth photovoltaic, p j∈X(j=1,2,…,K); repeatedly randomly select until a preset number of firefly individuals are obtained, and all the firefly individuals and the first-generation individuals form a first-generation firefly population, and the preset number is the difference between the preset number of individuals in the population and 1; the DBI index is used to determine the luminous brightness of the firefly population, and after obtaining the population brightness, the brightest individual is recorded, and the individual position is updated according to the position update formula to achieve population iteration, so as to use the improved K-means algorithm to cluster the distributed photovoltaic system, wherein, d(c i ,c j ) are two different cluster centers c i With c j The distance between i , S j Indicates c i 、c j The average intra-cluster distance of the corresponding cluster.

[0010] Optionally, after using the DBI indicator to determine the luminous brightness of the firefly population and obtaining the population brightness, the brightest individual is recorded, and the individual position is updated according to the position update formula. After the population iteration is realized, the method further includes: using the recorded brightest individual as the optimal initial center, inputting it into the K-means algorithm, and using the brightest individual obtained by the firefly algorithm as the clustering center of the first cluster, and calculating the distance from each sample to each cluster center according to the Euclidean distance method, wherein, D(a,b) is the distance from a sample to the cluster center. is the dimension of the cluster space; a k and b k The cluster center is updated by the distance measurement method to the mean of all samples belonging to the cluster. For each cluster, the sum of the square distances between all sample points in the cluster and the global average value is calculated as the square sum of the discreteness of the cluster. According to the formula Calculate the Calinski-Harabasz index and set the maximum number of samples k max As the iteration termination condition, the maximum value of CH is calculated, CH is the Calinski-Harabasz index, Tr(B k ) is the trace of the inter-cluster scatter matrix, Tr(W k ) is the trace of the intra-cluster scatter matrix, is the total number of samples.

[0011] Optionally, determining the dynamic response error between the distributed photovoltaic original model and the initial cluster equivalent model includes: performing transient simulation on the initial cluster equivalent model to obtain an active power transient response expression P of the initial cluster equivalent model. v(t) = [p v1 (t),p v2 (t),p v3 (t),...,p vn (t)] and the reactive power transient response expression Q of the initial cluster equivalent model v (t) = [q v1 (t),q v2 (t),q v3 (t),...,q vn (t)], where P v (t) is the active power of the initial clustering equivalent model, Q v (t) is the reactive power of the initial clustering equivalent model, p vn (t), q vn (t) are respectively the transient response expressions of the nth group of output active power and reactive power of the initial clustering equivalent model, p vi (t) = [p vi (1),p vi (2),p vi (3),...,p vi (τ)],q vi (t) = [q vi (1),q vi (2),q vi (3),...,q vi (τ)],p vi (τ) and q vi (τ) represents the output active power value and reactive power value of the i-th group of data of the initial clustering equivalent model at the τth moment; where i = 1, 2, ..., n; transient simulation is performed on the distributed photovoltaic original model to obtain the active power transient response expression P of the distributed photovoltaic original model c (t) = [p c1 (t),p c2 (t),p c3 (t),...,p cn (t)], the reactive power transient response expression Q of the distributed photovoltaic original model c (t) = [q c1 (t),q c2 (t),q c3 (t),...,q cn (t)], where P c (t) is the active power of the distributed photovoltaic original model, Q c (t) is the reactive power of the distributed photovoltaic original model, p ci (t), q ci(t) are the transient response expressions of the nth group of output active power and reactive power of the distributed photovoltaic primitive model, respectively. ci (t) = [p ci (1),p ci (2),p ci (3),...,p ci (τ)],q ci (t) = [q ci (1),q ci (2),q ci (3),...,q ci (τ)],p ci (τ) and q ci (τ) represents the output active power value and reactive power value of the i-th group of data of the distributed photovoltaic original model at the τth time; where i = 1, 2, ..., n; the difference between the active power of the distributed photovoltaic original model and the active power of the initial clustering equivalent model is determined as the active power error P e (t), and the difference between the reactive power of the distributed photovoltaic original model and the reactive power of the initial clustering equivalent model is determined as the reactive power error Q e (t), and obtain the dynamic response error between the distributed photovoltaic original model and the initial clustering equivalent model.

[0012] Optionally, the hyperparameters of the gated recurrent neural network are optimized using a lion group optimization algorithm, including: step S1, taking the hyperparameters of the gated recurrent neural network as the optimization object of the lion group optimization algorithm, establishing a three-dimensional lion group optimization search algorithm dimension, and initializing the lion group, wherein the hyperparameters include: an initial learning rate, a maximum number of iterations, and a number of neurons in the hidden layer; step S2, setting the individual historical optimal position as the current position of each lion, and the initial group optimal position as the lion king position; step S3, updating the positions of the lion king, lionesses, and cubs according to the position update formula of the lion group optimization algorithm, calculating the fitness, and updating the historical optimal position of each lion; step S4, judging whether the lion group needs to be reordered, and re-updating the lion king position, the lionesses position, and the cub position if the lion group needs to be reordered, returning to step S2, and executing step S5 if the lion group does not need to be reordered; step S5: judging whether the maximum number of iterations has been reached, and if the maximum number of iterations has not been reached, returning to step S3 to continue calculating, and outputting the lion king position and the hyperparameter value of the gated recurrent neural network if the maximum number of iterations has been reached.

[0013] According to another aspect of the present application, a distributed photovoltaic clustering equivalent modeling device based on LSO-GRU-ECM is provided, including: a construction unit, used to cluster and group the distributed photovoltaic system using an improved K-means algorithm, and construct an initial clustering equivalent model of the distributed photovoltaic system; a first determination unit, used to determine the dynamic response error between the distributed photovoltaic original model and the initial clustering equivalent model, wherein the distributed photovoltaic original model is obtained according to the parameters and topology of the distributed photovoltaic system; a training unit, used to optimize the hyperparameters of the gated recurrent neural network using a lion group optimization algorithm, and input the dynamic response error into the gated recurrent neural network for training to obtain an error correction model; a compensation unit, used to input the dynamic response error into the error correction model to obtain a compensation value, and compensate the initial clustering equivalent model with the compensation value to obtain a target clustering equivalent model, wherein the target clustering equivalent model is used to analyze the operating parameters of the distributed photovoltaic system.

[0014] According to another aspect of the present application, a computer-readable storage medium is provided, the computer-readable storage medium including a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the distributed photovoltaic clustering equivalent modeling methods based on LSO-GRU-ECM.

[0015] According to another aspect of the present application, an electronic device is provided, comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include a method for executing any one of the above-mentioned distributed photovoltaic clustering equivalent modeling methods based on LSO-GRU-ECM.

[0016] Applying the technical solution of the present application, the above-mentioned distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM first clusters and groups the distributed photovoltaic system using the improved K-means algorithm to construct the initial clustering equivalent model of distributed photovoltaic; then determine the dynamic response error between the distributed photovoltaic original model and the initial clustering equivalent model, the distributed photovoltaic original model is obtained according to the parameters and topology of the distributed photovoltaic system; then use the lion group optimization algorithm to optimize the hyperparameters of the gated recurrent neural network, and input the dynamic response error into the gated recurrent neural network for training to obtain the error correction model; finally, input the dynamic response error into the error correction model to obtain the compensation value, and compensate the compensation value to the initial clustering equivalent model to obtain the target clustering equivalent model, which is used to analyze the operating parameters of the distributed photovoltaic system. This method not only completes the model simulation with extremely low time consumption, but also corrects the error of the distributed photovoltaic clustering equivalent model with high precision, significantly improves the generalization ability and accuracy of the clustering model, and solves the problem that the traditional modeling method ignores the operating status and parameter differences of the distributed photovoltaic cluster itself, is difficult to generalize, and cannot fit the dynamic characteristics of the distributed photovoltaic cluster well. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings constituting part of the present application are used to provide a further understanding of the present application. The exemplary embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0018] Figure 1 A hardware structure block diagram of a mobile terminal for executing a distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM provided in an embodiment of the present application is shown;

[0019] Figure 2 A schematic flow chart of a distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM provided according to an embodiment of the present application is shown;

[0020] Figure 3 A distributed photovoltaic detailed model provided according to an embodiment of the present application is shown;

[0021] Figure 4 An active power effect diagram of a distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM provided in accordance with an embodiment of the present application is shown;

[0022] Figure 5 A reactive power effect diagram of a distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM provided in an embodiment of the present application is shown;

[0023] Figure 6A schematic diagram of a process of another distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM provided according to an embodiment of the present application is shown;

[0024] Figure 7 A structural block diagram of a distributed photovoltaic clustering equivalent modeling device based on LSO-GRU-ECM provided according to an embodiment of the present application is shown.

[0025] The above drawings include the following reference numerals:

[0026] 102, processor; 104, memory; 106, transmission device; 108, input and output devices. DETAILED DESCRIPTION

[0027] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0028] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0029] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0030] For the convenience of description, some nouns or terms involved in the embodiments of the present application are explained below:

[0031] Loin Swarm Optimization, or LSO for short;

[0032] Gated Recurrent Neural Network: Gated Recurrent Neural Network, referred to as GRU;

[0033] Error Correction Model: Error Correction Model, referred to as ECM.

[0034] As introduced in the background technology, the existing technology builds a distributed photovoltaic cluster equivalent model based on the cluster equivalent method, which improves the accuracy of the model and reduces the simulation time to a certain extent, but ignores the operating status and parameter differences of the distributed photovoltaic cluster itself, and based on the inherent characteristics of the clustering algorithm, the constructed equivalent cluster model has the problem of difficulty in generalization. In addition, the distributed photovoltaic cluster is a complex and strongly nonlinear system, and the traditional modeling method cannot well fit the dynamic characteristics of the distributed photovoltaic cluster.

[0035] In order to solve the problems that traditional modeling methods ignore the operating status and parameter differences of distributed photovoltaic clusters themselves, are difficult to generalize, and cannot well fit the dynamic characteristics of distributed photovoltaic clusters, the embodiments of the present application provide a distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM, a distributed photovoltaic clustering equivalent modeling device based on LSO-GRU-ECM, a computer-readable storage medium and an electronic device.

[0036] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0037] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Taking running on a mobile terminal as an example, Figure 1 1 is a hardware structure block diagram of a mobile terminal of a distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM according to an embodiment of the present invention. Figure 1 As shown, the mobile terminal may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data, wherein the mobile terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the mobile terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations shown.

[0038] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM in the embodiment of the present invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, the above method is implemented. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories can be connected to the mobile terminal through a network. Examples of the above-mentioned network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof. The transmission device 106 is used to receive or send data via a network. The above-mentioned specific examples of the network may include a wireless network provided by a communication provider of the mobile terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0039] In this embodiment, a distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM that runs on a mobile terminal, a computer terminal or a similar computing device is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in an order different from that shown here.

[0040] Figure 2 FIG. 1 is a flow chart of a distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM according to an embodiment of the present application. Figure 2 As shown, the method comprises the following steps:

[0041] Step S201, clustering the distributed photovoltaic system using an improved K-means algorithm to construct an initial clustering equivalent model of the distributed photovoltaic system;

[0042] Specifically, the distributed photovoltaic clustering index is first selected to make the photovoltaics with similar photovoltaic characteristics in the distributed photovoltaics aggregated into equal values; then the entire distributed photovoltaic system is equivalent to one or several photovoltaics through the clustering algorithm; after clustering, the parameters of the equivalent photovoltaics are calculated and an equivalent model is constructed, which improves the accuracy of the model and reduces the simulation time to a certain extent.

[0043] Among them, the improved K-means algorithm is used to cluster the distributed photovoltaic system and construct an initial clustering equivalent model, which includes the following steps:

[0044] Step S301, selecting distributed photovoltaic clustering indicators;

[0045] Step S302, combining the firefly algorithm with the K-means algorithm to obtain the above-mentioned improved K-means algorithm, and using the above-mentioned improved K-means algorithm to equate the entire above-mentioned distributed photovoltaic system to one or more photovoltaics;

[0046] Step S303, calculating parameters of each equivalent photovoltaic, and constructing the above-mentioned initial clustering equivalent model.

[0047] Specifically, the K-means clustering algorithm is a simple and efficient unsupervised learning method. The above steps use the firefly algorithm for parameter optimization of the K-means algorithm, and combine the firefly algorithm with the K-means algorithm to achieve the optimal distribution network load clustering result. The optimal firefly individual of the firefly algorithm is used as the initial center of the K-means algorithm to achieve a better distributed photovoltaic clustering effect.

[0048] Among them, selecting distributed photovoltaic clustering indicators includes the following steps:

[0049] Step S401, based on the above distributed photovoltaic original model, the above distributed photovoltaic clustering index is determined according to the zero-pole expression of the system closed-loop transfer function, wherein the above distributed photovoltaic clustering index vector is a three-dimensional vector X=[z, p1, p2], and the zero-pole expression of the above system closed-loop transfer function is: Among them, L f is the filter inductance; k p is the proportional control constant; k i is the integral control constant.

[0050] Specifically, the effect of clustering equivalent modeling depends to a large extent on the selection of clustering indicators. At present, there is no standard clustering indicator for equivalent modeling of photovoltaic power generation systems. Based on the distributed photovoltaic system model, the zero-pole expression of the system closed-loop transfer function is used as the clustering indicator to directly characterize the directly related parameters of the dynamic performance of the photovoltaic system, which can significantly improve the accuracy of the model.

[0051] Among them, the improved K-means algorithm is used to cluster the distributed photovoltaic system, including the following steps:

[0052] Step S501: normalize the distributed photovoltaic related data and use it as the input data set X of the algorithm, and randomly select K different photovoltaic data from the data set X as a group of initial centers of the firefly algorithm, i.e., the first generation individuals P, where: x nd represents the data of the dth period of the nth photovoltaic, p j ∈X(j=1,2,…,K);

[0053] Step S502, repeatedly randomly selecting until a preset number of firefly individuals are obtained, and all the firefly individuals and the first-generation individuals are combined into a first-generation firefly population, where the preset number is the difference between the preset number of individuals in the population and 1;

[0054] Step S503, using the DBI index to determine the luminous brightness of the firefly population, after obtaining the population brightness, record the brightest individual, and update the individual position according to the position update formula to achieve population iteration, so as to cluster the distributed photovoltaic system using the improved K-means algorithm, wherein, d(c i ,c j ) are two different cluster centers c i With c j The distance between i , S j Indicates c i 、c j The average intra-cluster distance of the corresponding cluster.

[0055] Specifically, the DBI indicator is used to determine the brightness of fireflies. The smaller the indicator value, the better the clustering effect, which is manifested as a stronger brightness. After obtaining the population brightness, the brightest individual is recorded, and then moved according to the probability attraction formula. The individual position is updated according to the position update formula to achieve population iteration, overcoming the problems of poor generalization ability and high randomness caused by traditional hyperparameter selection methods.

[0056] Among them, after using the DBI indicator to determine the luminous brightness of the firefly population, obtaining the population brightness, recording the brightest individual, and updating the individual position according to the position update formula, after realizing population iteration, the above method also includes the following steps:

[0057] Step S601, the brightest individual recorded is used as the optimal initial center and input into the K-means algorithm, and the brightest individual obtained by the firefly algorithm is used as the cluster center of the first cluster, and the distance from each sample to each cluster center is calculated according to the Euclidean distance method, where: D(a,b) is the distance from a sample to the cluster center. is the dimension of the cluster space; a k and b k It is divided into a k-dimensional vector, which is used to represent the k-th attribute;

[0058] Step S602, updating the cluster center by a distance measurement method to the mean of all samples belonging to the cluster, and for each cluster, calculating the sum of the square distances between all sample points in the cluster and the global mean value as the square sum of the discreteness of the cluster;

[0059] Step S603, according to the formula Calculate the Calinski-Harabasz index and set the maximum number of samples k max As the iteration termination condition, the maximum value of CH is calculated, CH is the Calinski-Harabasz index, Tr(B k ) is the trace of the inter-cluster dispersion matrix, Tr(W k ) is the trace of the intra-cluster scatter matrix, is the total number of samples.

[0060] Specifically, the cluster center is updated by the distance measurement method to the mean of all samples belonging to the cluster. For each cluster, the sum of the square distances between all sample points in the cluster and the global mean is calculated as the square sum of the discreteness of the cluster. The Calinski-Harabasz index is calculated, and the larger the CH, the better the clustering effect.

[0061] Step S202, determining a dynamic response error between a distributed photovoltaic original model and the initial clustering equivalent model, wherein the distributed photovoltaic original model is obtained according to the parameters and topology of the distributed photovoltaic system;

[0062] Specifically, the distributed photovoltaic clustering equivalent model reduces the accuracy of distributed photovoltaic modeling to a certain extent, but reduces the simulation time of the model to a certain extent. Modeling using the clustering method cannot overcome the inherent defects of distributed photovoltaics themselves, and the accuracy will still be significantly reduced in transient response simulation. Therefore, it is necessary to combine the distributed photovoltaic original model to construct an error correction model.

[0063] The determination of the dynamic response error between the distributed photovoltaic original model and the above-mentioned initial clustering equivalent model includes the following steps:

[0064] Step S701, performing transient simulation on the initial clustering equivalent model to obtain the active power transient response expression P of the initial clustering equivalent model v (t) = [p v1 (t),p v2 (t),pv3 (t),...,p vn (t)] and the reactive power transient response expression Q of the above initial cluster equivalent model v (t) = [q v1 (t),q v2 (t),q v3 (t),...,q vn (t)], where P v (t) is the active power of the above initial clustering equivalent model, Q v (t) is the reactive power of the above initial clustering equivalent model, p vn (t), q vn (t) are the transient response expressions of the nth group of output active power and reactive power of the above initial clustering equivalent model, respectively. vi (t) = [p vi (1),p vi (2),p vi (3),...,p vi (τ)],q vi (t) = [q vi (1),q vi (2),q vi (3),...,q vi (τ)],p vi (τ) and q vi (τ) represents the output active power value and reactive power value of the i-th group of data of the above initial clustering equivalent model at the τ-th time; where i=1,2,...,n;

[0065] Step S702, performing transient simulation on the distributed photovoltaic primitive model to obtain the active power transient response expression P of the distributed photovoltaic primitive model. c (t) = [p c1 (t),p c2 (t),p c3 (t),...,p cn (t)], the reactive power transient response expression Q of the above distributed photovoltaic original model c (t) = [q c1 (t),q c2 (t),q c3 (t),...,q cn (t)], where P c (t) is the active power of the above distributed photovoltaic original model, Q c (t) is the reactive power of the above distributed photovoltaic original model, p ci (t), q ci(t) are the transient response expressions of the nth group of output active power and reactive power of the above distributed photovoltaic original model, respectively. ci (t) = [p ci (1),p ci (2),p ci (3),...,p ci (τ)],q ci (t) = [q ci (1),q ci (2),q ci (3),...,q ci (τ)],p ci (τ) and q ci (τ) represents the output active power value and reactive power value of the i-th group of data of the above distributed photovoltaic original model at the τth moment; where i=1,2,...,n;

[0066] Step S703: the difference between the active power of the distributed photovoltaic original model and the active power of the initial clustering equivalent model is determined as the active power error P e (t), and the difference between the reactive power of the above distributed photovoltaic original model and the reactive power of the above initial clustering equivalent model is determined as the reactive power error Q e (t), and obtain the dynamic response error between the above distributed photovoltaic original model and the above initial clustering equivalent model.

[0067] Specifically, the training result of ECM is the mapping from low-precision equivalent model data to its error. For any transient time t, the output of CEM is close to the error value P of a specific scenario. e (t) and Q e (t), the distributed photovoltaic simulation model is combined with deep learning technology to build an error correction model framework, demonstrating the superior performance of the model.

[0068] In some instances, during the transient response of distributed photovoltaics, the clustering equivalent model outputs active power P v (t) and reactive power Q v (t), distributed photovoltaic original model active power P c (t) and reactive power Q c (t), the active power and reactive power of the distributed photovoltaic original model are subtracted from the active power and reactive power of the cluster equivalent model to obtain the active power error P between the distributed photovoltaic detailed model and the cluster equivalent model in the transient process. e (t) and reactive power error Q e (t).

[0069] P e (t) = Pc (t)-P v (t);

[0070] Q e (t) = Q c (t)-Q v (t).

[0071] Let the input of ECM be the low-precision transient response data P of the clustering equivalent model v (t) and Q v (t), the output active power error P of the distributed photovoltaic original model and the cluster equivalent model e (t) and reactive power error Q e (t) is the fitting target of GRU in ECM, that is, the training result of ECM is the mapping from low-precision equivalent model data to its error. For any transient time t, the output result of CEM is close to the error value P of a specific scene. e (t) and Q e (t).

[0072] Step S203, using a lion group optimization algorithm to optimize the hyperparameters of the gated recurrent neural network, and inputting the above dynamic response error into the above gated recurrent neural network for training to obtain an error correction model;

[0073] Specifically, the GRU network is a special type of recurrent neural network that has the advantage of maintaining high computational accuracy while maintaining fast computational speed, and preventing gradient vanishing and explosion problems by resetting gates and updating gates. When faced with large-scale, high-dimensional network traffic data feature mining, the GRU network has the advantage of fast fitting.

[0074] In some instances, the calculation process of the entire network is:

[0075] 1) Reset gate: The gate for forgetting information. Its size is determined by the current input data and the hidden gate at the previous moment. The expression is as follows:

[0076] r t =σ(W r x t +U r h t-1 +b r ), where r t is the reset gate at time t; σ is the Sigmoid activation function; W r Input data parameter matrix for reset gate; x t is the input at time t; U r is the parameter matrix of the hidden door at time t-1; h t-1 is the hidden door at time t-1; b r To reset the overall gate offset size.

[0077] 2) Hidden door: The size is determined by the hidden door at the previous moment and the reset door at the current moment. The expression is as follows:

[0078] In the formula, h t Hide the door for time t; is the dot multiplication operation; U h is the parameter matrix of the hidden door at time t-1; W h b is the input data parameter matrix for the hidden gate; h The overall offset size of the hidden door.

[0079] 3) Update gate: The gate where information is retained. Its size is determined by the hidden gate at the previous moment and the input at the current moment. The expression is as follows:

[0080] u t =σ(W u x t +U u h t-1 +b u ), where u t is the update gate; W u Input state parameter matrix at the current moment; U u h is the input data parameter matrix for the update gate; t-1 is the hidden door at time t-1; b u To update the overall gate offset size.

[0081] Among them, the hyperparameters of the gated recurrent neural network are optimized using the lion group optimization algorithm, including the following steps:

[0082] Step S801, taking the hyperparameters of the gated recurrent neural network as the optimization object of the lion group optimization algorithm, establishing a three-dimensional lion group optimization search algorithm dimension, and initializing the lion group. The hyperparameters include: initial learning rate, maximum number of iterations, and number of neurons in the hidden layer;

[0083] Step S802, setting the individual historical optimal position as the current position of each lion, and the initial group optimal position is the position of the lion king;

[0084] Step S803, according to the position update formula of the lion group optimization algorithm, update the positions of the lion king, lionesses, and lion cubs, calculate the fitness, and update the historical optimal position of each lion;

[0085] Step S804, determining whether the lion group needs to be reordered, and if the lion group needs to be reordered, re-updating the position of the lion king, the position of the lioness, and the position of the cubs, and returning to step S802; if the lion group does not need to be reordered, executing step S805;

[0086] Step S805, determine whether the maximum number of iterations has been reached. If the maximum number of iterations has not been reached, return to step S803 to continue the calculation. If the maximum number of iterations has been reached, output the lion king position and the hyperparameter value of the gated recurrent neural network.

[0087] Specifically, this method not only completes the model simulation in a very low time, but also corrects the errors of the distributed photovoltaic clustering equivalence model with high precision, significantly improving the generalization ability and accuracy of the clustering model.

[0088] In some instances, before training based on the GRU network, the network hyperparameters are the key to training the GRU network, including the initial learning rate, the maximum number of iterations, the number of neurons in the hidden layer, etc. In order to solve the hyperparameter problem of the GRU model in this paper, the Lion Group Optimization (LSO) algorithm is introduced to optimize the GRU network hyperparameters.

[0089] The specific process of LSO optimizing GRU network hyperparameters is as follows:

[0090] Step 1: Take the GRU network hyperparameters including the initial learning rate, the maximum number of iterations, and the number of hidden layer neurons as the optimization objects of the LSO algorithm, establish the 3D LSO search algorithm dimension, and initialize the lion group;

[0091] Step 2: Set the individual historical optimal position as the current position of each lion, and the initial group optimal position is the position of the lion king;

[0092] Step 3: Update the positions of the lion king, lionesses, and lion cubs according to the position update formula of the LSO algorithm, calculate the fitness, and update the historical optimal position of each lion;

[0093] Step 4: Determine whether the lion group needs to be reordered. If so, re-update the positions of the lion king, lionesses, and cubs and return to step 2. Otherwise, continue with the next step.

[0094] Step 5: Determine whether the maximum number of iterations has been reached. If not, return to step 3 to continue the calculation. If yes, output the lion king's position and the GRU network hyperparameter value, and end the process.

[0095] Step S204, inputting the above dynamic response error into the above error correction model to obtain a compensation value, and compensating the above compensation value to the above initial clustering equivalent model to obtain a target clustering equivalent model, and the above target clustering equivalent model is used to analyze the operating parameters of the above distributed photovoltaic system.

[0096] Specifically, the GRU network optimized by LSO is used, and the input of ECM is the low-precision transient response data P of the clustering equivalent model. v (t) and Q v(t), the output active power error P of the distributed photovoltaic original model and the cluster equivalent model e (t) and reactive power error Q e (t) is used as the fitting target of GRU in ECM; the first 4 / 5 of the data are taken as the training set, and the last 1 / 5 of the data are taken as the validation set. The output result of CEM is close to the error value P of a specific scene. e (t) and Q e (t) Finally, the output result is added and compensated to the clustering equivalent model based on the improved K-means algorithm to obtain a clustering equivalent model close to the original distributed photovoltaic model. This method not only completes the model simulation with extremely low time consumption, but also corrects the error of the distributed photovoltaic clustering equivalent model with high precision, significantly improving the generalization ability and accuracy of the clustering model.

[0097] The above-mentioned distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM of the present application first uses an improved K-means algorithm to cluster the distributed photovoltaic system and construct an initial clustering equivalent model of distributed photovoltaics; then the dynamic response error between the distributed photovoltaic original model and the initial clustering equivalent model is determined, and the distributed photovoltaic original model is obtained according to the parameters and topology of the distributed photovoltaic system; then the hyperparameters of the gated recurrent neural network are optimized using the lion group optimization algorithm, and the dynamic response error is input into the gated recurrent neural network for training to obtain an error correction model; finally, the dynamic response error is input into the error correction model to obtain a compensation value, and the compensation value is compensated to the initial clustering equivalent model to obtain a target clustering equivalent model, which is used to analyze the operating parameters of the distributed photovoltaic system. This method not only completes the model simulation with extremely low time consumption, but also corrects the error of the distributed photovoltaic clustering equivalent model with high precision, significantly improves the generalization ability and accuracy of the clustering model, and solves the problem that the traditional modeling method ignores the operating status and parameter differences of the distributed photovoltaic cluster itself, is difficult to generalize, and cannot well fit the dynamic characteristics of the distributed photovoltaic cluster.

[0098] Among them, the above embodiment adopts the improved K-means algorithm and capacity weighted method to cluster the distributed photovoltaic clusters, constructs a distributed photovoltaic clustering equivalent model, uses LSO to optimize the GRU network hyperparameters, takes the dynamic response errors of the distributed photovoltaic detailed model and the clustering equivalent model as input, constructs an error correction model, adds the output of the network to the clustering equivalent model, and constructs a distributed photovoltaic clustering equivalent model based on LSO-GRU-ECM.

[0099] The beneficial effects of the above embodiment are: using the firefly algorithm to optimize the K-means algorithm parameters to obtain a more accurate clustering equivalence model, then using the LSO algorithm to optimize the GRU network hyperparameters, training and constructing the error correction model, overcoming the problems of poor generalization ability and large randomness caused by the traditional hyperparameter selection method, and using the additive compensation clustering equivalence model to finally construct a distributed photovoltaic clustering equivalence model based on LSO-GRU-ECM, which has the characteristics of high accuracy and short time consumption.

[0100] Example

[0101] The simulation scenario used for model verification in this embodiment is as follows Figure 3 shown.

[0102] The photovoltaic power generation system consists of 15 photovoltaic power stations, among which the installed capacity of each power station is: 1-8 is 50kW, 9-11 is 240kW, 12-15 is 180kW, and the total photovoltaic installed capacity is 1840kW. The parameters of each photovoltaic power station are different, and the dynamic parameters are shown in the table.

[0103] Table 1. Distributed photovoltaic dynamic parameters

[0104]

[0105]

[0106] The photovoltaic inverter control parameter k p , k i and filter parameter L f Taking the index as the index, the improved K-means clustering is performed on 15 photovoltaic power stations, and the number of clusters is set to 2. Then the photovoltaic power stations in each cluster are equivalent to one photovoltaic power station, and the capacity calculation method is used to calculate the equivalent parameters. The clustering results are shown below.

[0107] Table 2. Improved K-means clustering results

[0108] category Unit No. 1 2,4,5,6,9,11,12,13,14 2 1,3,7,8,10,15

[0109] According to the distributed photovoltaic equivalent modeling method mentioned above in this paper, after detailed modeling and cluster equivalent modeling of distributed photovoltaic power stations, simulation runs were performed under different working conditions to obtain the transient response characteristics of distributed photovoltaics under the two methods. Based on the above analysis, a distributed photovoltaic detailed model and a cluster equivalent model were built in the PSCAD / EMTDC platform, and the transient response data of the distributed photovoltaic original model and the cluster equivalent model were analyzed and obtained. Then the collected data was processed by Z-score standardization, and LSO-GRU network training was performed after the processing was completed.

[0110] The GRU network used in this paper consists of an input layer, a hidden layer, and an output layer. To prevent overfitting, the hidden layer contains two connected GRU layers, a fully connected Dense layer, and two Dropout layers. The hidden layer mainly randomly discards and restores some neurons in each iteration, so that the network structure of each training is different, so that the model does not depend on the local features of the input layer, and solves the overfitting problem. Enhance the generalization ability of the neural network. In order to prevent gradient explosion, this paper sets the gradient threshold to 1. Due to the large data set, in order to reduce the time spent on training the network, this paper packages the data set into batches for training, with a small batch size of 20.

[0111] The specific parameters of LSO-GRU are shown in the table.

[0112] Table 3. LSO parameter table

[0113] parameter Numeric Adult lion proportion factor β 0.2 <![CDATA[Random numbers α1α2]]> 0.1 Fitness calculation RMSE

[0114] Table 4. GRU parameter table

[0115] parameter Numeric Input sequence dimension 1 Output sequence dimension 1 Dropout probability 0.5 Gradient Threshold 1 Training set and validation set division 80% and 20% Learning Rate LSO Optimization Maximum number of iterations LSO Optimization Number of neurons in the hidden layer LSO Optimization

[0116] The distributed photovoltaic equivalent modeling method based on LSO-GRU-ECM proposed in this paper combines the distributed photovoltaic simulation model with deep learning technology to build an error correction model framework. In order to verify the superior performance of the model, the model proposed in this paper is compared with the distributed photovoltaic detailed model, the K-means clustering equivalent model, and the error correction model without LSO hyperparameter optimization under the same working conditions. The neural network parameters without the LSO algorithm are set as follows: the number of hidden neural units is 200, the learning rate is 0.01, and the maximum number of iterations is set to 20 considering that the network iteration takes a long time. The model performance is evaluated by the root mean square error RMSE.

[0117] In the formula, is the transient response of the distributed photovoltaic original model of the distributed photovoltaic power station, and λX(t) is the final output of each model.

[0118] In the simulation software, the irradiance of the distributed photovoltaic power station changes at 0.2s, and the dynamic response process curves of the distributed photovoltaic original model, K-means clustering equivalent model, GRU error correction model without LSO hyperparameter optimization, LSO-GRU error correction model and detailed modeling are analyzed. The simulation results are as follows: Figure 4 and Figure 5 shown.

[0119] Figure 4An active power effect diagram of a distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM provided according to an embodiment of the present application is shown; in the figure, A is an active power curve based on the K-means equivalent model, B is an active power curve based on the LSO-GRU-ECM equivalent model, C is an active power curve based on the detailed model, and D is an active power curve based on the LSO-GRU equivalent model.

[0120] Figure 5 A reactive power effect diagram of a distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM provided in an embodiment of the present application is shown; in the figure, A is based on a detailed model reactive power curve, B is based on an LSO-GRU-ECM equivalent model reactive power curve, C is based on an LSO-GRU equivalent model reactive power curve, and D is based on a K-means equivalent model reactive power curve.

[0121] The performance comparison and simulation time of different models are shown in Tables 4 and 5.

[0122] Table 5. Comparison of simulation performance of different models

[0123] Model Active power error Reactive power error Distributed photovoltaic original model / / K-means clustering model 13% 12.8% GRU-ECM equivalent model 8.6% 7.3% The model proposed in this paper 3.8% 4.5%

[0124] Table 6. Comparison of simulation time of different models

[0125] Model Simulation time Distributed photovoltaic original model 679s Cluster Equivalence Model 26s The model proposed in this paper 29.5s

[0126] Figure 4 , Figure 5 The high accuracy of the model proposed in this paper is verified. In the process of irradiation intensity change, the compensation correction of the LSO-GRU network can basically accurately fit the errors of the distributed photovoltaic original model and the cluster equivalent model. At the same time, as shown in the table, the fitting accuracy of the LSO-GRU error correction model is significantly improved compared with the cluster equivalent model and the GRU error correction model, which proves the effectiveness of the Lion Group Optimization Algorithm (LSO). In summary, the distributed photovoltaic equivalent modeling method based on LSO-GRU-ECM proposed in this paper not only completes the model simulation with extremely low time consumption, but also corrects the errors of the distributed photovoltaic cluster equivalent model with high accuracy, significantly improving the generalization ability and accuracy of the cluster model.

[0127] In order to enable those skilled in the art to more clearly understand the technical solution of the present application, the implementation process of the distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM of the present application will be described in detail below in combination with specific embodiments.

[0128] This embodiment relates to a specific distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM, such as Figure 6As shown, the following steps are included:

[0129] Step S1: Obtain distributed photovoltaic clustering indicators and data, normalize the sampled data, perform clustering based on the improved K-means algorithm, determine the optimal clustering in combination with the CH indicator, and construct a distributed photovoltaic clustering equivalent model based on the clustering results;

[0130] Step S2: construct a distributed photovoltaic detailed model, and construct an ECM based on the dynamic response error of the distributed photovoltaic detailed model and the cluster equivalent model;

[0131] Step S3: Use LSO to optimize the GRU network hyperparameters, verify the network, and calculate the fitness function; update the positions of the lion king, lioness, and cubs, and determine whether the conditions for stopping iteration are met. If so, output the three optimal hyperparameters, otherwise return to the verification network step and train to obtain the error correction model;

[0132] Step S4: Add the output result to the clustering equivalent model based on the improved K-means algorithm to obtain a clustering equivalent model close to the original model of distributed photovoltaics.

[0133] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0134] The embodiment of the present application also provides a distributed photovoltaic clustering equivalent modeling device based on LSO-GRU-ECM. It should be noted that the distributed photovoltaic clustering equivalent modeling device based on LSO-GRU-ECM in the embodiment of the present application can be used to execute the distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM provided in the embodiment of the present application. The device is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware for a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceived.

[0135] The following is an introduction to the distributed photovoltaic clustering equivalent modeling device based on LSO-GRU-ECM provided in an embodiment of the present application.

[0136] Figure 7 Schematic diagram of a distributed photovoltaic clustering equivalent modeling device based on LSO-GRU-ECM according to an embodiment of the present application. Figure 7As shown, the device includes: a construction unit 10, a first determination unit 20, a training unit 30 and a compensation unit 40, the construction unit 10 is used to cluster the distributed photovoltaic system using an improved K-means algorithm to construct an initial clustering equivalent model of the distributed photovoltaic system; the first determination unit 20 is used to determine the dynamic response error between the distributed photovoltaic original model and the above-mentioned initial clustering equivalent model, and the above-mentioned distributed photovoltaic original model is obtained according to the parameters and topology of the above-mentioned distributed photovoltaic system; the training unit 30 is used to optimize the hyperparameters of the gated recurrent neural network using the lion group optimization algorithm, and input the above-mentioned dynamic response error into the above-mentioned gated recurrent neural network for training to obtain an error correction model; the compensation unit 40 is used to input the above-mentioned dynamic response error into the above-mentioned error correction model to obtain a compensation value, and compensate the above-mentioned compensation value to the above-mentioned initial clustering equivalent model to obtain a target clustering equivalent model, and the above-mentioned target clustering equivalent model is used to analyze the operating parameters of the above-mentioned distributed photovoltaic system.

[0137] The above-mentioned distributed photovoltaic clustering equivalent modeling device based on LSO-GRU-ECM of the present application includes: a construction unit, a first determination unit, a training unit and a compensation unit, the construction unit is used to cluster and group the distributed photovoltaic system using an improved K-means algorithm to construct an initial clustering equivalent model of distributed photovoltaics; the first determination unit is used to determine the dynamic response error between the distributed photovoltaic original model and the initial clustering equivalent model, and the distributed photovoltaic original model is obtained according to the parameters and topology of the distributed photovoltaic system; the training unit is used to optimize the hyperparameters of the gated recurrent neural network using the lion group optimization algorithm, and input the dynamic response error into the gated recurrent neural network. The training is carried out in the network to obtain the error correction model; the compensation unit is used to input the dynamic response error into the error correction model to obtain the compensation value, and the compensation value is compensated to the initial clustering equivalent model to obtain the target clustering equivalent model. The target clustering equivalent model is used to analyze the operating parameters of the distributed photovoltaic system. The device not only completes the model simulation with extremely low time consumption, but also corrects the error of the distributed photovoltaic clustering equivalent model with high precision, significantly improves the generalization ability and accuracy of the clustering model, and solves the problems that the traditional modeling method ignores the operating status and parameter differences of the distributed photovoltaic cluster itself, is difficult to generalize, and cannot fit the dynamic characteristics of the distributed photovoltaic cluster well.

[0138] In some examples, the construction unit includes a first acquisition module, a first processing module and a first calculation module. The first acquisition module is used to select distributed photovoltaic clustering indicators; the first processing module is used to combine the firefly algorithm with the K-means algorithm to obtain the above-mentioned improved K-means algorithm, and the above-mentioned improved K-means algorithm is used to equate the entire distributed photovoltaic system to one or more photovoltaics; the first calculation module is used to calculate the parameters of each equivalent photovoltaic and construct the above-mentioned initial clustering equivalent model. The firefly algorithm is used to optimize the parameters of the K-means algorithm. The firefly algorithm is combined with the K-means algorithm to achieve the optimal distribution network load clustering result. The optimal firefly individual of the firefly algorithm is used as the initial center of the K-means algorithm to achieve a better distributed photovoltaic clustering effect.

[0139] In some examples, the first acquisition module includes a first determination module, which is used to determine the distributed photovoltaic clustering index based on the distributed photovoltaic original model and the zero-pole expression of the system closed-loop transfer function, wherein the distributed photovoltaic clustering index vector is a three-dimensional vector X=[z, p1, p2], and the zero-pole expression of the system closed-loop transfer function is: Among them, L f is the filter inductance; k p is the proportional control constant; k i is the integral control constant. At present, there is no standard clustering index for equivalent modeling of photovoltaic power generation system. Based on the distributed photovoltaic system model, the zero-pole expression of the system closed-loop transfer function is used as the clustering index, which can significantly improve the accuracy of the model.

[0140] In some examples, the construction unit includes a second processing module, a third processing module, and a fourth processing module. The second processing module is used to normalize the distributed photovoltaic related data as the input data set X of the algorithm, and randomly select K different photovoltaic data from the data set X as a group of initial centers of the firefly algorithm, i.e., the initial generation individuals P, wherein, x nd represents the data of the dth period of the nth photovoltaic, p j ∈X(j=1,2,…,K); the third processing module is used to repeatedly randomly select until a preset number of firefly individuals are obtained, and all the above firefly individuals and the above first-generation individuals form a first-generation firefly population, and the above preset number is the difference between the preset number of individuals in the population and 1; the fourth processing module is used to use the DBI indicator to determine the luminous brightness of the firefly population, after obtaining the population brightness, record the brightest individual, and update the individual position according to the position update formula to achieve population iteration, so as to use the improved K-means algorithm to cluster the distributed photovoltaic system, wherein, d(c i ,c j) are two different cluster centers c i With c j The distance between i , S j Indicates c i 、c j The average intra-class distance of the corresponding cluster. This method overcomes the problems of poor generalization ability and high randomness caused by traditional hyperparameter selection methods.

[0141] In some examples, the fourth processing module includes a second calculation module, a third calculation module and a fourth calculation module, wherein the second calculation module is used to input the brightest individual recorded as the optimal initial center into the K-means algorithm, and use the brightest individual obtained by the firefly algorithm as the cluster center of the first cluster, and calculate the distance from each sample to each cluster center according to the Euclidean distance method, wherein, D(a,b) is the distance from a sample to the cluster center. is the dimension of the cluster space; a k and b k The k-dimensional vector is used to represent the k-th attribute; the third calculation module is used to update the cluster center by the distance measurement method, and update it to the mean of all samples belonging to the cluster. For each cluster, the sum of the square distances of all sample points in the cluster and the global average value is calculated as the square sum of the discreteness of the cluster; the fourth calculation module is used to calculate the sum of the square distances of all sample points in the cluster and the global average value according to the formula Calculate the Calinski-Harabasz index and set the maximum number of samples k max As the iteration termination condition, the maximum value of CH is calculated, CH is the Calinski-Harabasz index, Tr(B k ) is the trace of the inter-cluster dispersion matrix, Tr(W k ) is the trace of the intra-cluster scatter matrix, is the total number of samples. The cluster center is updated by the distance measurement method to the mean of all samples belonging to the cluster. For each cluster, the sum of the square distances between all sample points in the cluster and the global average is calculated as the square sum of the discreteness of the cluster. The Calinski-Harabasz index is calculated, and the larger the CH, the better the clustering effect.

[0142] In some examples, the first determination unit includes a first simulation module, a second simulation module and a second determination module, the first simulation module is used to perform transient simulation on the initial cluster equivalent model to obtain an active power transient response expression P of the initial cluster equivalent model. v (t) = [p v1 (t),p v2 (t),p v3 (t),...,p vn(t)] and the reactive power transient response expression Q of the above initial cluster equivalent model v (t) = [q v1 (t),q v2 (t),q v3 (t),...,q vn (t)], where P v (t) is the active power of the above initial clustering equivalent model, Q v (t) is the reactive power of the above initial clustering equivalent model, p vn (t), q vn (t) are the transient response expressions of the nth group of output active power and reactive power of the above initial clustering equivalent model, respectively. vi (t) = [p vi (1),p vi (2),p vi (3),...,p vi (τ)],q vi (t) = [q vi (1),q vi (2),q vi (3),...,q vi (τ)],p vi (τ) and q vi (τ) represents the output active power value and reactive power value of the i-th group of data of the above-mentioned initial clustering equivalent model at the τth moment; where i = 1, 2, ..., n; the second simulation module is used to perform transient simulation on the above-mentioned distributed photovoltaic original model to obtain the active power transient response expression P of the above-mentioned distributed photovoltaic original model c (t) = [p c1 (t),p c2 (t),p c3 (t),...,p cn (t)], the reactive power transient response expression Q of the above distributed photovoltaic original model c (t) = [q c1 (t),q c2 (t),q c3 (t),...,q cn (t)], where P c (t) is the active power of the above distributed photovoltaic original model, Q c (t) is the reactive power of the above distributed photovoltaic original model, p ci (t), q ci (t) are the transient response expressions of the nth group of output active power and reactive power of the above distributed photovoltaic original model, respectively. ci (t) = [p ci (1),pci (2),p ci (3),...,p ci (τ)],q ci (t) = [q ci (1),q ci (2),q ci (3),...,q ci (τ)],p ci (τ) and q ci (τ) represents the output active power value and reactive power value of the i-th group of data of the above-mentioned distributed photovoltaic original model at the τth moment; wherein i=1,2,...,n; the second determination module is used to determine the difference between the active power of the above-mentioned distributed photovoltaic original model and the active power of the above-mentioned initial clustering equivalent model as the active power error P e (t), and the difference between the reactive power of the above distributed photovoltaic original model and the reactive power of the above initial clustering equivalent model is determined as the reactive power error Q e (t), and the dynamic response error between the above distributed photovoltaic original model and the above initial clustering equivalent model is obtained. The distributed photovoltaic simulation model is combined with deep learning technology to build an error correction model framework, showing the superior performance of the model.

[0143] In some examples, the training unit includes a first construction module, a first setting module, a first updating module, a first judgment module and a second judgment module. In step S1, the first construction module is used to use the hyperparameters of the gated recurrent neural network as the optimization object of the lion group optimization algorithm, establish the dimension of the three-dimensional lion group optimization search algorithm, and initialize the lion group. The hyperparameters include: initial learning rate, maximum number of iterations, and number of neurons in the hidden layer; in step S2, the first setting module is used to set the individual historical optimal position as the current position of each lion, and the initial group optimal position is the lion king position; in step S3, the first updating module is used to update the public according to the position of the lion group optimization algorithm. Formula, update the positions of the lion king, lionesses, and lion cubs, calculate fitness, and update the historical optimal position of each lion; step S4, the first judgment module is used to determine whether the lion group needs to be reordered, and re-update the lion king position, lionesses position, and lion cub position if the lion group needs to be reordered, return to step S2, and execute step S5 if the lion group does not need to be reordered; step S5: the second judgment module is used to determine whether the maximum number of iterations has been reached. If the maximum number of iterations has not been reached, return to step S3 to continue calculating, and output the position of the lion king and the hyperparameter value of the gated recurrent neural network if the maximum number of iterations has been reached. This method not only completes the model simulation with extremely low time consumption, but also corrects the error of the distributed photovoltaic clustering equivalent model with high precision, significantly improving the generalization ability and accuracy of the clustering model.

[0144] The above-mentioned distributed photovoltaic clustering equivalent modeling device based on LSO-GRU-ECM includes a processor and a memory, and the above-mentioned construction units are stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to realize the corresponding functions. The above-mentioned modules are all located in the same processor; or, the above-mentioned modules are located in different processors in the form of any combination.

[0145] The processor contains a kernel, which calls the corresponding program unit from the memory. One or more kernels can be set, and the kernel parameters can be adjusted to solve the problem that the traditional modeling method ignores the operating status and parameter differences of the distributed photovoltaic cluster itself, is difficult to generalize, and cannot fit the dynamic characteristics of the distributed photovoltaic cluster well.

[0146] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0147] An embodiment of the present invention provides a computer-readable storage medium, the computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM. An embodiment of the present invention provides a processor, the processor is used to run the program, wherein when the program is executed, the distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM is executed.

[0148] An embodiment of the present invention provides a device, comprising a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, at least the steps of the above-mentioned distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM are implemented.

[0149] The devices in this article can be servers, PCs, PADs, mobile phones, etc.

[0150] The present application also provides a computer program product, which, when executed on a data processing device, is suitable for executing a program that initializes at least the steps of the above-mentioned distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM.

[0151] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0152] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0153] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0154] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0155] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0156] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0157] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0158] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0159] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0160] From the above description, it can be seen that the above embodiments of the present application achieve the following technical effects:

[0161] 1) The distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM of the present application first clusters the distributed photovoltaic system using an improved K-means algorithm to construct an initial clustering equivalent model of distributed photovoltaics; then the dynamic response error between the distributed photovoltaic original model and the initial clustering equivalent model is determined, and the distributed photovoltaic original model is obtained according to the parameters and topology of the distributed photovoltaic system; then the hyperparameters of the gated recurrent neural network are optimized using the lion group optimization algorithm, and the dynamic response error is input into the gated recurrent neural network for training to obtain an error correction model; finally, the dynamic response error is input into the error correction model to obtain a compensation value, and the compensation value is compensated to the initial clustering equivalent model to obtain a target clustering equivalent model, which is used to analyze the operating parameters of the distributed photovoltaic system. This method not only completes the model simulation with extremely low time consumption, but also corrects the error of the distributed photovoltaic clustering equivalent model with high precision, significantly improves the generalization ability and accuracy of the clustering model, and solves the problem that the traditional modeling method ignores the operating status and parameter differences of the distributed photovoltaic cluster itself, is difficult to generalize, and cannot fit the dynamic characteristics of the distributed photovoltaic cluster well.

[0162] 2) The above-mentioned distributed photovoltaic clustering equivalent modeling device based on LSO-GRU-ECM of the present application includes: a construction unit, a first determination unit, a training unit and a compensation unit. The construction unit is used to cluster and group the distributed photovoltaic system using an improved K-means algorithm to construct an initial clustering equivalent model of distributed photovoltaics; the first determination unit is used to determine the dynamic response error between the distributed photovoltaic original model and the initial clustering equivalent model, and the distributed photovoltaic original model is obtained according to the parameters and topology of the distributed photovoltaic system; the training unit is used to optimize the hyperparameters of the gated recurrent neural network using the lion group optimization algorithm, and input the dynamic response error into the gated recurrent neural network. The training is carried out in the neural network to obtain the error correction model; the compensation unit is used to input the dynamic response error into the error correction model to obtain the compensation value, and the compensation value is compensated to the initial clustering equivalent model to obtain the target clustering equivalent model, and the target clustering equivalent model is used to analyze the operating parameters of the distributed photovoltaic system. The device not only completes the model simulation with extremely low time consumption, but also corrects the error of the distributed photovoltaic clustering equivalent model with high precision, significantly improves the generalization ability and accuracy of the clustering model, and solves the problems that the traditional modeling method ignores the operating status and parameter differences of the distributed photovoltaic cluster itself, is difficult to generalize, and cannot well fit the dynamic characteristics of the distributed photovoltaic cluster.

[0163] 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.

Claims

1. A distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM, characterized in that: include: The improved K-means algorithm is used to cluster the distributed photovoltaic system and construct the initial clustering equivalent model of distributed photovoltaic. Determining a dynamic response error between a distributed photovoltaic original model and the initial clustering equivalent model, wherein the distributed photovoltaic original model is obtained according to parameters and topology of the distributed photovoltaic system; Optimizing the hyperparameters of the gated recurrent neural network using a lion group optimization algorithm, and inputting the dynamic response error into the gated recurrent neural network for training to obtain an error correction model; The dynamic response error is input into the error correction model to obtain a compensation value, and the compensation value is compensated to the initial clustering equivalent model to obtain a target clustering equivalent model, and the target clustering equivalent model is used to analyze the operating parameters of the distributed photovoltaic system.

2. The method according to claim 1, characterized in that The improved K-means algorithm is used to cluster the distributed photovoltaic system and construct an initial clustering equivalent model, including: Select distributed photovoltaic clustering indicators; The firefly algorithm is combined with the K-means algorithm to obtain the improved K-means algorithm, and the improved K-means algorithm is used to equate the entire distributed photovoltaic system to one or more photovoltaics; Parameters of each equivalent photovoltaic are calculated to construct the initial clustering equivalent model.

3. The method according to claim 2, characterized in that Select distributed photovoltaic clustering indicators, including: Based on the distributed photovoltaic primitive model, the distributed photovoltaic clustering index is determined according to the zero-pole expression of the system closed-loop transfer function, wherein the distributed photovoltaic clustering index vector is a three-dimensional vector X=[z, p1, p2], and the zero-pole expression of the system closed-loop transfer function is: Among them, L f is the filter inductance; k p is the proportional control constant; k i is the integral control constant.

4. The method according to claim 1, characterized in that: The improved K-means algorithm is used to cluster the distributed photovoltaic system, including: The distributed photovoltaic related data is normalized and used as the input data set X of the algorithm. K different photovoltaic data are randomly selected from the data set X as a set of initial centers of the firefly algorithm, i.e., the first generation individuals P. x nd represents the data of the dth period of the nth photovoltaic, p j ∈X(j=1,2,…,K); Repeatedly randomly select until a preset number of firefly individuals are obtained, and all the firefly individuals and the first-generation individuals form a first-generation firefly population, where the preset number is the difference between the preset number of individuals in the population and 1; The DBI index is used to determine the luminous brightness of the firefly population. After obtaining the population brightness, the brightest individual is recorded, and the individual position is updated according to the position update formula to achieve population iteration, so as to cluster the distributed photovoltaic system using the improved K-means algorithm. d(c i ,c j ) are two different cluster centers c i With c j The distance between i , S j Indicates c i 、c j The average intra-cluster distance of the corresponding cluster.

5. The method according to claim 4, characterized in that After the brightness of the firefly population is determined by using the DBI indicator and the population brightness is obtained, the brightest individual is recorded, and the individual position is updated according to the position update formula to implement population iteration, the method further includes: The brightest individual recorded is used as the optimal initial center and input into the K-means algorithm. The brightest individual obtained by the firefly algorithm is used as the cluster center of the first cluster, and the distance from each sample to each cluster center is calculated according to the Euclidean distance method, where: D(a,b) is the distance from a sample to the cluster center. is the dimension of the cluster space; a k and b k It is divided into a k-dimensional vector, which is used to represent the k-th attribute; The cluster center is updated by the distance measurement method to the mean of all samples belonging to the cluster. For each cluster, the sum of the square distances between all sample points in the cluster and the global mean is calculated as the square sum of the discreteness of the cluster. According to the formula Calculate the Calinski-Harabasz index and set the maximum number of samples k max As the iteration termination condition, the maximum value of CH is calculated. CH is the Calinski-Harabasz index. Tr(B k ) is the trace of the inter-cluster dispersion matrix, Tr(W k ) is the trace of the intra-cluster scatter matrix, is the total number of samples.

6. The method according to claim 1, characterized in that Determining the dynamic response error between the distributed photovoltaic original model and the initial clustering equivalent model includes: The transient simulation is performed on the initial clustering equivalent model to obtain the active power transient response expression P of the initial clustering equivalent model. v (t) = [p v1 (t),p v2 (t),p v3 (t),...,p vn (t)] and the reactive power transient response expression Q of the initial cluster equivalent model v (t) = [q v1 (t),q v2 (t),q v3 (t),...,q vn (t)], where P v (t) is the active power of the initial clustering equivalent model, Q v (t) is the reactive power of the initial clustering equivalent model, p vn (t), q vn (t) are respectively the transient response expressions of the nth group of output active power and reactive power of the initial clustering equivalent model, p vi (t) = [p vi (1),p vi (2),p vi (3),...,p vi (τ)],q vi (t) = [q vi (1),q vi (2),q vi (3),...,q vi (τ)],p vi (τ) and q vi (τ) represents the output active power value and reactive power value of the i-th group of data of the initial clustering equivalent model at the τ-th time; wherein i=1, 2, ..., n; The distributed photovoltaic primitive model is subjected to transient simulation to obtain the active power transient response expression P of the distributed photovoltaic primitive model. c (t) = [p c1 (t),p c2 (t),p c3 (t),...,p cn (t)], the reactive power transient response expression Q of the distributed photovoltaic original model c (t) = [q c1 (t),q c2 (t),q c3 (t),...,q cn (t)], where P c (t) is the active power of the distributed photovoltaic original model, Q c (t) is the reactive power of the distributed photovoltaic original model, p ci (t), q ci (t) are the transient response expressions of the nth group of output active power and reactive power of the distributed photovoltaic primitive model, respectively. ci (t) = [p ci (1),p ci (2),p ci (3),...,p ci (τ)],q ci (t) = [q ci (1),q ci (2),q ci (3),...,q ci (τ)],p ci (τ) and q ci (τ) represents the output active power value and reactive power value of the i-th group of data of the distributed photovoltaic original model at the τ-th time; wherein i=1,2,...,n; The difference between the active power of the distributed photovoltaic original model and the active power of the initial clustering equivalent model is determined as the active power error P e (t), and the difference between the reactive power of the distributed photovoltaic original model and the reactive power of the initial clustering equivalent model is determined as the reactive power error Q e (t), and obtain the dynamic response error between the distributed photovoltaic original model and the initial clustering equivalent model.

7. The method according to claim 1, characterized in that The lion group optimization algorithm is used to optimize the hyperparameters of the gated recurrent neural network, including: Step S1, taking the hyperparameters of the gated recurrent neural network as the optimization object of the lion group optimization algorithm, establishing a three-dimensional lion group optimization search algorithm dimension, and initializing the lion group, wherein the hyperparameters include: initial learning rate, maximum number of iterations, and number of neurons in the hidden layer; Step S2, setting the individual historical optimal position as the current position of each lion, and the initial group optimal position is the lion king position; Step S3, according to the position update formula of the lion group optimization algorithm, update the positions of the lion king, lionesses, and lion cubs, calculate the fitness, and update the historical optimal position of each lion; Step S4, determining whether the lion group needs to be reordered, and if the lion group needs to be reordered, re-updating the position of the lion king, the position of the lioness, and the position of the cubs, returning to step S2, and if the lion group does not need to be reordered, executing step S5; Step S5: Determine whether the maximum number of iterations has been reached. If the maximum number of iterations has not been reached, return to step S3 to continue the calculation. If the maximum number of iterations has been reached, output the lion king position and the hyperparameter value of the gated recurrent neural network.

8. A distributed photovoltaic clustering equivalent modeling device based on LSO-GRU-ECM, characterized in that: include: A construction unit is used to cluster the distributed photovoltaic system using an improved K-means algorithm and to construct an initial clustering equivalent model of the distributed photovoltaic system; A first determination unit is used to determine a dynamic response error between a distributed photovoltaic original model and the initial clustering equivalent model, wherein the distributed photovoltaic original model is obtained according to parameters and topology of the distributed photovoltaic system; A training unit, used for optimizing the hyperparameters of the gated recurrent neural network using a lion group optimization algorithm, and inputting the dynamic response error into the gated recurrent neural network for training to obtain an error correction model; A compensation unit is used to input the dynamic response error into the error correction model to obtain a compensation value, and to compensate the initial clustering equivalent model with the compensation value to obtain a target clustering equivalent model, wherein the target clustering equivalent model is used to analyze the operating parameters of the distributed photovoltaic system.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM according to any one of claims 1 to 7.

10. An electronic device, characterized in that: include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include a method for executing the distributed photovoltaic clustering equivalent modeling method based on LSO-GRU-ECM as described in any one of claims 1 to 7.

Citation Information

Patent Citations

  • Photovoltaic system dynamic process hybrid equivalent modeling method based on data driving

    CN112036010A

  • Photovoltaic power station equivalent modeling method and system

    CN115718881A

  • Wind power plant equivalent modeling method based on PSO-LSTM-ECM

    CN116738807A

  • Photovoltaic power station equivalent modeling method, medium and system

    CN117540638A

Cited By

  • High voltage ride through control method, device and equipment and storage medium

    CN120033719A

  • A control method, device and equipment for high voltage ride through and storage medium

    CN120033719B

  • Distributed intelligent power grid digital simulation method, device, equipment, medium and product

    CN121457086A