Specific wind power scene generation method and device based on LSTM-ACGAN

Abstract

The invention discloses a specific wind power scene generation method and device based on LSTM-ACGAN, and the method comprises the steps: obtaining the wind curtailment rate of each wind power scene through solving an optimization problem based on a historical load scene and a historical wind power scene, classifying the wind power scenes according to the range of the wind curtailment rate, obtaining an original wind power scene data set, and obtaining a wind power scene data set; secondly, constructing an LSTM-ACGAN model in a mode of introducing a Bi-LSTM model into an ACGAN generator, training the LSTM-ACGAN by using the original wind power scene sample set, obtaining an LSTM-ACGAN generator capable of generating different specific types of wind power scenes after training is completed, inputting Gaussian noise and a specified wind power scene label into the generator, and finally obtaining the wind power scene. And the generator generates a large number of specific wind power scenes similar to the specified type of real wind power scene. The method can provide assistance for subsequent analysis and decision-making of the power system.

Description

Technical Field

[0001] The present invention relates to the field of new energy technologies, and in particular to a method and device for generating a specific wind power scenario based on LSTM-ACGAN. Background Art

[0002] After large-scale access to the power system, wind power usually faces absorption problems due to its intermittent and volatile nature. In order to consider the impact of wind power access, many tasks of power system dispatching decisions require a large number of wind power scenarios as basic data. Therefore, generating wind power output scenarios is of great significance. In the context of increasingly stringent requirements for wind abandonment rates and growing wind power penetration, if the absorption difficulty of wind power scenarios can be further considered when generating wind power output scenarios, and targeted generation and analysis can be carried out, it will help explore the impact of specific wind power scenarios on the power system and improve the power system's absorption capacity for different wind power scenarios.

[0003] In order to realize the generation of wind power scenarios, Chinese patent publication number CN107039981A uses Latin hypercube sampling to generate wind power scenarios, and at the same time obtains a few wind power scenarios with larger probability measures based on scenario reduction technology; Chinese patent publication number CN117767387A uses the Monte Carlo method to uniformly sample the prediction error value based on the probability distribution curve of wind power output prediction error, and calculates it through the inverse transform sampling method and the output prediction error formula to obtain the actual output scenario of the wind farm. However, due to the complex characteristics of wind power scenarios, the wind power scenarios obtained by traditional probability sampling or Monte Carlo simulation often differ greatly from the real scenarios in some features, making it difficult to ensure the similarity between the generated scenarios and the real scenarios. With the continuous improvement of computing power and algorithms, researchers have begun to pay attention to the use of artificial intelligence methods to generate wind power scenarios. Chinese patent publication number CN111008455A proposes a wind power scenario generation method based on conditional generative adversarial networks, which replaces the convolutional layers in the generator and discriminator of the generative adversarial network with time series enhancement modules to improve the model's capture of the time series characteristics of wind power scenarios. However, the time series enhancement module used in this method is based on the traditional differential model, which is difficult to capture the complex and changeable high-dimensional features in actual wind power scenarios. In addition, existing methods often classify and generate wind power output scenarios based on statistical features such as total output and change rate, which cannot reflect the difficulty of wind power scenarios and cannot meet the urgent needs of power system research on wind power scenarios with different difficulty of wind power scenarios. Summary of the invention

[0004] The present invention aims to solve one of the technical problems in the related art at least to a certain extent.

[0005] To this end, the present invention proposes a specific wind power scenario generation method based on LSTM-ACGAN, which can realize the generation of specific wind power scenarios for different absorption difficulties.

[0006] Another object of the present invention is to propose a specific wind power scenario generation device based on LSTM-ACGAN.

[0007] To achieve the above objectives, the present invention proposes a method for generating a specific wind power scenario based on LSTM-ACGAN, comprising:

[0008] Based on the historical load data and historical wind power output data of the power system, a historical load scenario data set and a historical wind power scenario data set are obtained, and the historical load scenario data set is clustered to obtain a typical load scenario data set;

[0009] Based on the typical load in the typical load scenario data set, the wind abandonment rate corresponding to each wind power scenario in the historical wind power scenario data set is obtained by solving the optimization problem, and the wind power scenarios are classified according to the range of wind abandonment rate to obtain the original wind power scenario data set;

[0010] Constructing an LSTM-ACGAN model based on Bi-LSTM and a fully connected layer neural network; wherein the LSTM-ACGAN model includes a generator, a discriminator, and a classifier;

[0011] The LSTM-ACGAN model is trained using the original wind power scenario dataset, and after the training is completed, the generator is used as a wind power scenario generator;

[0012] Based on the typical load scenario and the type of wind power scenario to be generated, a wind power scenario similar to the real wind power scenario is generated by using a corresponding wind power scenario generator.

[0013] To achieve the above object, the present invention proposes, on the other hand, a specific wind power scenario generation device based on LSTM-ACGAN, comprising:

[0014] A data processing module, used to obtain a historical load scenario data set and a historical wind power scenario data set based on the historical load data and historical wind power output data of the power system, and cluster the historical load scenario data set to obtain a typical load scenario data set;

[0015] The wind power scenario classification module is used to obtain the wind abandonment rate corresponding to each wind power scenario in the historical wind power scenario data set by solving the optimization problem based on the typical load in the typical load scenario data set, and classify the wind power scenarios according to the range of wind abandonment rate to obtain the original wind power scenario data set;

[0016] A model building module, used to build an LSTM-ACGAN model based on Bi-LSTM and a fully connected layer neural network; wherein the LSTM-ACGAN model includes a generator, a discriminator, and a classifier;

[0017] The model training module is used to train the LSTM-ACGAN model using the original wind power scenario dataset and use the generator as a wind power scenario generator after the training is completed;

[0018] The wind power scenario generation module is used to generate a wind power scenario similar to a real wind power scenario by using a corresponding wind power scenario generator based on a typical load scenario and the type of wind power scenario to be generated.

[0019] The specific wind power scenario generation method and device based on LSTM-ACGAN in the embodiment of the present invention trains LSTM-ACGAN through classified wind power scenarios, so that the model can generate specific wind power scenarios with different absorption difficulties, thereby meeting the power system's analysis requirements for wind power scenarios with different absorption difficulties.

[0020] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0022] Figure 1 is a flowchart of a method for generating a specific wind power scenario based on LSTM-ACGAN according to an embodiment of the present invention;

[0023] Figure 2 is a structural diagram of a specific wind power scenario generation device based on LSTM-ACGAN according to an embodiment of the present invention. DETAILED DESCRIPTION

[0024] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0025] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0026] The following describes a method and device for generating a specific wind power scenario based on LSTM-ACGAN according to an embodiment of the present invention with reference to the accompanying drawings.

[0027] Figure 1is a flowchart of a method for generating a specific wind power scenario based on LSTM-ACGAN according to an embodiment of the present invention. Figure 1 As shown, the method includes:

[0028] S1, based on the historical load data and historical wind power output data of the power system, a historical load scenario data set and a historical wind power scenario data set are obtained, and the historical load scenario data set is clustered to obtain a typical load scenario data set.

[0029] Specifically, based on the historical load data of the power system and the historical wind power output data, the historical wind power scenario dataset D is obtained through data processing. w0 and typical load scenario dataset D d , including the following steps:

[0030] S11, for a time length of T w The one-dimensional historical wind power output sequence is divided into n w There are historical wind power scenarios with a length of T, and the dimension is (n w ,T) historical wind power scenario dataset D w0 ,in

[0031] S12, for a time length of T d The one-dimensional historical load sequence is divided into n parts according to the target scenario time length T. d There are historical load scenarios with a length of T, and the composition dimension is (n d ,T) historical load scenario dataset D d0 ,in

[0032] S13, to D d0 n d The historical load scenarios with a length of T are divided into M clusters by K-means clustering, which are denoted as D d1 , D d2 , …, D dM . Take the cluster center P of these M clusters d1 , P d2 ,…,P dM Constructing a typical load scenario dataset D d = {P dm}, where m = 1, 2, ... M. In one embodiment of the present invention, take T w =17520,T d =8760, T=144, M=3.

[0033] S2, based on the typical load in the typical load scenario data set, obtains the wind abandonment rate corresponding to each wind power scenario in the historical wind power scenario data set by solving the optimization problem, and classifies the wind power scenarios according to the range of the wind abandonment rate to obtain the original wind power scenario data set.

[0034] Specifically, for the typical loads in the typical load scenario data set, the wind power scenarios are classified by solving the optimization problem to obtain the wind power scenario data set D w , including the following steps:

[0035] S21, construct a mathematical model for the optimization problem. The specific formula is as follows:

[0036] Objective function:

[0037]

[0038] Optimization problem constraints:

[0039] 0≤p w (t)≤P w (t)

[0040]

[0041] P gmin (i)X(i,t)≤P g (i,t)≤P gmax (i)X(i,t)

[0042] R d (i)≤P g (i,t+1)-P g (i,t)≤R u (i)

[0043]

[0044] X(i,t)-X(i,t-1)-Y(i,t)+Z(i,t)=0

[0045] -X(i,t)-X(i,t-1)+Y(i,t)≤0

[0046] X(i,t)+X(i,t-1)+Y(i,t)≤2

[0047] -X(i,t)-X(i,t-1)+Z(i,t)≤0

[0048] X(i,t)+X(i,t-1)+Z(i,t)≤2

[0049] Where t represents a time point between 1 and T; P w Represents a certain wind power scenario, Pw (t) represents the wind power scenario P w The maximum power that can be generated at time t, p w (t) represents the wind power scenario P w The actual power at time t; N gen Indicates the number of thermal power units; i indicates the number of the thermal power unit, i=1,2,...,N gen ;P g (i, t) represents the active power output of the i-th thermal power unit at time t; P d Represents a typical load scenario, P d (t) represents the load scenario P d The load size at time t; X(i,t) represents the operating status of unit i at time t, if it is in the operating state at this moment, it is 1, otherwise it is 0; Y(i,t) represents the startup process of unit i at time t, if the unit is starting at this moment, it is 1, otherwise it is 0; Z(i,t) represents the shutdown process of unit i at time t, if the unit is shutting down at this moment, it is 1, otherwise it is 0; P re is the system spinning reserve factor; P gmax (i) is the maximum active output of unit i, P gmin (i) is the minimum active output of unit i; R d (i) is the maximum downslope rate of the i-th thermal power unit, R u (i) is the maximum ramp rate of the i-th thermal power unit; T on and T off In one embodiment of the present invention, the power system considered is an IEEE 39-bus system, and the No. 30 bus generator is changed to a wind farm. Then N gen =9. Take P re =5%, R d (i) = 0.05 × P gmax (i), R u (i) = 0.05 × P gmax (i), T on =T off =2h. gmax (i) and P gmin (i) Given by the IEEE 39-node example system.

[0050] S22, for typical load scenario P dm , respectively for D w0 Each wind power scenario P wl Solve the optimization problem and obtain the wind power scenario P in time 1 to T wl The actual power p wl (t), where l = 1, 2, ..., nw Indicates the sequence number of the wind power scenario. Then calculate each wind power scenario P wl The corresponding wind curtailment rate η l , the calculation formula is as follows:

[0051]

[0052] S23, according to the range of wind abandonment rate, n w The wind power scenarios are divided into K categories, and each type of wind power scenario dataset is recorded as D wm_1 , D wm_2 , …, D wm_K , together constitute the typical load scenario P dm The corresponding original wind power scene dataset D wm ={D wm_k}, where k = 1, 2, ... K. In one embodiment of the present invention, K = 4. The relationship between the wind abandonment rate range and the classification is defined as: a wind abandonment rate of 0 is defined as Class 1, a wind abandonment rate between (0, 25%] is defined as Class 2, a wind abandonment rate between (25%, 50%] is defined as Class 3, and a wind abandonment rate greater than 50% is defined as Class 4.

[0053] S24, for all typical load scenarios P d1 , P d2 ,…,P dM , respectively carry out the operation steps in (S22) to (S23) to obtain D w1 , D w2 , …, D wM , together constitute the original wind power scene dataset D w ={D wm}, where m = 1, 2, ..., M.

[0054] S3, constructing an LSTM-ACGAN model based on Bi-LSTM and a fully connected layer neural network; wherein the LSTM-ACGAN model includes a generator, a discriminator, and a classifier.

[0055] Specifically, the LSTM-ACGAN model is constructed based on Bi-LSTM and fully connected layer neural network, including the following steps:

[0056] S31, build generator M G , the input is subject to the Z-dimensional Gaussian distribution p z The Gaussian noise z of (z) and the type label c=1,2,...,K of the wind power scenario to be generated are output as the generated wind power scenario P wG The specific calculation formula is as follows:

[0057] P wG =M G(z,c)=FC(LSTM(z+Embedding(c)))

[0058] Where Embedding(·) represents the embedding layer, which maps the integer c into a vector with the same shape as z; LSTM(·) represents N L Layer Bi-LSTM neural network; FC(·) represents N F Layer fully connected network, each layer consists of a linear mapping Linear (·) and an activation function ReLU (·). In one embodiment of the present invention, Z = 500, N L =2, N F =4.

[0059] S32, build the discriminator M D , input y is the real wind power scenario P w Or generate wind power scenario P wG , the output is the discrimination result S. S = 1 represents the discrimination as a real scene, and S = 0 represents the discrimination as a generated scene. The specific calculation formula is as follows:

[0060] S=M D (y) = Sigmoid(Linear(FC(y)))

[0061] The calculation formula of the function Sigmoid(·) is:

[0062] S33, build classifier M C , input y is the real wind power scenario P w Or generate wind power scenario P wG , the output is the classification result The specific calculation formula is as follows:

[0063]

[0064] The function Softmax(·) is a normalized exponential function, and M C With M D The FC(·) network function parameters are shared.

[0065] S34, define the loss function of the network, the specific calculation formula is as follows:

[0066] L S =E[logP(S=1|P w )]+E[logP(S=0|P wG )]

[0067]

[0068] L G =-(LC +L S )

[0069] L D =-(L C -L S )

[0070] Where E[·] represents the mean value, L S and L C Represents the accuracy of discrimination and the correctness of classification respectively. Using the loss function L G Update M G Parameters, using the loss function L D Update M D and M C Parameters.

[0071] S4, use the original wind power scenario dataset to train the LSTM-ACGAN model, and use the generator as a wind power scenario generator after training.

[0072] Specifically, we use the wind power scenario dataset D w Training LSTM-ACGAN to obtain a wind power scenario generator includes the following steps:

[0073] S41, set training parameters, including the number of training rounds Ep, batch training size B, optimizer, and learning rate lr. In one embodiment of the present invention, Ep=10000, B=64, the optimizer is Adam, and lr=0.0001.

[0074] S42, for the mth typical load scenario P dm , using its corresponding original wind power scene dataset D wm Train the corresponding LSTM-ACGAN model. After the model training is completed, take the model generator M Gm As the mth typical load scenario P dm Corresponding wind power scenario generator.

[0075] S43, for all typical load scenarios P d1 , P d2 ,…,P dM , respectively perform the operation steps in (S42), respectively obtain the wind power scenario generator corresponding to each typical load scenario, and together constitute the wind power scenario generator M for all typical loads G ={M Gm}, where m = 1, 2, ..., M.

[0076] S5, based on the typical load scenario and the type of wind power scenario to be generated, using a corresponding wind power scenario generator to generate a wind power scenario similar to the actual wind power scenario.

[0077] Specifically, for a specific typical load scenario and a specific wind power scenario type to be generated, a corresponding wind power scenario generator is used to generate a large number of specific generated wind power scenarios similar to the specific real wind power scenario, including the following steps:

[0078] S51, determine the load scenario type m, the wind power scenario type c to be generated, and the number of wind power scenarios to be generated N. For the mth type of typical load scenario, select the corresponding wind power scenario generator M Gm .

[0079] S52, towards M Gm Input N Z-dimensional Gaussian noise z and type label c, then the generator will output N typical load scenarios P dm The wind power scenario of type c is generated to complete the generation of the specific wind power scenario. In one embodiment of the present invention, N=10000.

[0080] According to the specific wind power scenario generation method based on LSTM-ACGAN of the embodiment of the present invention, the generative adversarial network ACGAN with auxiliary classifier is used as the basic model, and its deep neural network can accurately capture the characteristics of the wind power scenario in the output probability distribution. By introducing the bidirectional LSTM network Bi-LSTM into the generator of ACGAN, the generator of the model can capture the complex and changeable time series characteristics of the wind power scenario, so that the generated wind power scenario is similar to the actual wind power scenario in both output distribution and time series characteristics; this method uses the method of solving optimization problems to process the original wind power scenario, obtains the wind abandonment rate corresponding to the wind power scenario under different typical load scenarios, judges the difficulty of wind power scenario according to the range of wind abandonment rate, and classifies it. By training LSTM-ACGAN with classified wind power scenarios, the model can generate specific wind power scenarios with different absorption difficulties, meeting the power system's analysis requirements for wind power scenarios with different absorption difficulties.

[0081] In order to implement the above embodiment, Figure 2 As shown, this embodiment also provides a specific wind power scenario generation device 10 based on LSTM-ACGAN, including:

[0082] The data processing module 100 is used to obtain a historical load scenario data set and a historical wind power scenario data set based on the historical load data and the historical wind power output data of the power system, and cluster the historical load scenario data set to obtain a typical load scenario data set;

[0083] The wind power scenario classification module 200 is used to obtain the wind abandonment rate corresponding to each wind power scenario in the historical wind power scenario data set by solving the optimization problem based on the typical load in the typical load scenario data set, and classify the wind power scenarios according to the range of the wind abandonment rate to obtain the original wind power scenario data set;

[0084] A model building module 300 is used to build an LSTM-ACGAN model based on Bi-LSTM and a fully connected layer neural network; wherein the LSTM-ACGAN model includes a generator, a discriminator, and a classifier;

[0085] The model training module 400 is used to train the LSTM-ACGAN model using the original wind power scenario data set, and use the generator as a wind power scenario generator after the training is completed;

[0086] The wind power scenario generation module 500 is used to generate a wind power scenario similar to a real wind power scenario by using a corresponding wind power scenario generator based on a typical load scenario and a type of wind power scenario to be generated.

[0087] According to the specific wind power scenario generation device based on LSTM-ACGAN of the embodiment of the present invention, the generative adversarial network ACGAN with auxiliary classifier is used as the basic model, and its deep neural network can accurately capture the characteristics of the wind power scenario in the output probability distribution. By introducing the bidirectional LSTM network Bi-LSTM into the generator of ACGAN, the generator of the model can capture the complex and changeable time series characteristics of the wind power scenario, so that the generated wind power scenario is similar to the actual wind power scenario in both output distribution and time series characteristics; the original wind power scenario is processed by solving the optimization problem to obtain the wind abandonment rate corresponding to the wind power scenario under different typical load scenarios, and the wind power scenario is judged according to the range of the wind abandonment rate, and classified. By training LSTM-ACGAN with the classified wind power scenarios, the model can generate specific wind power scenarios with different absorption difficulties, meeting the power system's analysis requirements for wind power scenarios with different absorption difficulties.

[0088] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0089] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present invention, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

Claims

1. A specific wind power scenario generation method based on LSTM-ACGAN, characterized in that: include: S1, obtaining a historical load scenario data set and a historical wind power scenario data set based on the historical load data and historical wind power output data of the power system, and clustering the historical load scenario data set to obtain a typical load scenario data set; S2, based on the typical load in the typical load scenario data set, obtain the wind abandonment rate corresponding to each wind power scenario in the historical wind power scenario data set by solving the optimization problem, and classify the wind power scenarios according to the range of wind abandonment rate to obtain the original wind power scenario data set; S3, constructing an LSTM-ACGAN model based on Bi-LSTM and a fully connected layer neural network; wherein the LSTM-ACGAN model includes a generator, a discriminator, and a classifier; S4, using the original wind power scenario dataset to train the LSTM-ACGAN model, and after the training is completed, the generator is used as a wind power scenario generator; S5, based on the typical load scenario and the type of wind power scenario to be generated, using a corresponding wind power scenario generator to generate a wind power scenario similar to the actual wind power scenario.

2. The method according to claim 1, characterized in that Based on the historical load data and historical wind power output data of the power system, the historical load scenario data set and the historical wind power scenario data set are obtained, and the historical load scenario data set is clustered to obtain a typical load scenario data set, including: S11, based on the time length T w The one-dimensional historical wind power output sequence is divided into n w There are historical wind power scenarios with a length of T, and the dimension is (n w ,T) historical wind power scenario dataset D w0 ,in S12, based on the time length T d The one-dimensional historical load sequence is divided into n d There are historical load scenarios with a length of T, and the composition dimension is (n d ,T) historical load scenario dataset D d0 ,in S13, to D d0 n d The historical load scenarios with a length of T are divided into M clusters by K-means clustering, which are denoted as D d1 , D d2 , …, D dM ; Take the cluster center P of M clusters d1 , P d2 ,…,P dM Constructing a typical load scenario dataset D d = {P dm }, where m = 1, 2, ...M.

3. The method according to claim 1, characterized in that Based on the typical load in the typical load scenario data set, the wind abandonment rate corresponding to each wind power scenario in the historical wind power scenario data set is obtained by solving the optimization problem, and the wind power scenarios are classified according to the range of wind abandonment rate to obtain the original wind power scenario data set, including: S21, construct a mathematical model for the optimization problem, the formula is as follows: Objective function: Optimization problem constraints: 0≤p w (t)≤P w (t) P gmin (i)X(i,t)≤P g (i,t)≤P gmax (i)X(i,t) R d (i)≤P g (i,t+1)-P g (i,t)≤R u (i) X(i,t)-X(i,t-1)-Y(i,t)+Z(i,t)=0 -X(i,t)-X(i,t-1)+Y(i,t)≤0 X(i,t)+X(i,t-1)+Y(i,t)≤2 -X(i,t)-X(i,t-1)+Z(i,t)≤0 X(i,t)+X(i,t-1)+Z(i,t)≤2 Where t represents a time point between 1 and T; P w Represents a certain wind power scenario, P w (t) represents the wind power scenario P w The maximum power that can be generated at time t, p w (t) represents the wind power scenario P w The actual power at time t; N gen Indicates the number of thermal power units; i indicates the number of the thermal power unit, i=1,2,...,N gen ;P g (i, t) represents the active power output of the i-th thermal power unit at time t; P d Represents a typical load scenario, P d (t) represents the load scenario P d The load size at time t; X(i,t) represents the operating status of unit i at time t, if it is in the operating state at this moment, it is 1, otherwise it is 0; Y(i,t) represents the startup process of unit i at time t, if the unit is starting at this moment, it is 1, otherwise it is 0; Z(i,t) represents the shutdown process of unit i at time t, if the unit is shutting down at this moment, it is 1, otherwise it is 0; P re is the system spinning reserve factor; P gmax (i) is the maximum active output of unit i, P gmin (i) is the minimum active output of unit i; R d (i) is the maximum downslope rate of the i-th thermal power unit, R u (i) is the maximum ramp rate of the i-th thermal power unit; T on and T off They represent the minimum start-up time and minimum shutdown time of the unit respectively; S22, for typical load scenario P dm , respectively for D w0 Each wind power scenario P wl Solve the optimization problem and obtain the wind power scenario P in time 1 to T wl The actual power p wl (t), where l = 1, 2, ..., n w Indicates the sequence number of the wind power scene; then calculate each wind power scene P wl The corresponding wind curtailment rate η l , the calculation formula is as follows: S23, according to the range of wind abandonment rate, n w The wind power scenarios are divided into K categories, and each type of wind power scenario dataset is recorded as D wm_1 , D wm_2 , …, D wm_K , together constitute the typical load scenario P dm The corresponding original wind power scene dataset D wm ={D wm_k }, where k = 1, 2, ... K; S24, for all typical load scenarios P d1 , P d2 ,…,P dM , respectively carry out the operation steps in (S22) to (S23) to obtain D w1 , D w2 , …, D wM , together constitute the original wind power scene dataset D w ={D wm }, where m = 1, 2, ..., M.

4. The method according to claim 1, characterized in that: The LSTM-ACGAN model is built based on Bi-LSTM and fully connected layer neural network, including: S31, build generator M G , the input is subject to the Z-dimensional Gaussian distribution p z The Gaussian noise z of (z) and the type label c=1,2,...,K of the wind power scenario to be generated are output as the generated wind power scenario P wG ; The calculation formula is as follows: P wG =M G (z,c)=FC(LSTM(z+Embedding(c))) Among them, Embedding(·) represents the embedding layer, which maps the integer c to a vector with the same shape as z; LSTM(·) represents N L Layer Bi-LSTM neural network; FC(·) represents N F Layer fully connected network, each layer consists of a linear mapping Linear(·) and an activation function ReLU(·); S32, build the discriminator M D , input y is the real wind power scenario P w Or generate wind power scenario P wG , the output is the discrimination result S; S = 1 represents the discrimination as a real scene, S = 0 represents the discrimination as a generated scene; the calculation formula is as follows: S=M D (y)=Sigmoid(Linear(FC(y))) Among them, the calculation formula of the function Sigmoid(·) is: S33, build classifier M C , input y is the real wind power scenario P w Or generate wind power scenario P wG , the output is the classification result The calculation formula is as follows: Among them, the function Softmax(·) is a normalized exponential function, and M C With M D FC(·) network function parameter sharing; S34, define the loss function of the network, the specific calculation formula is as follows: L S =E[logP(S=1|P w )]+E[logP(S=0|P wG )] L G =-(L C +L S ) L D =-(L C -L S ) Among them, E[·] represents the mean value, L S and L C Represents the discrimination accuracy and classification correctness respectively; using the loss function L G Update M G Parameters, using the loss function L D Update M D and M C Parameters.

5. The method according to claim 1, characterized in that The LSTM-ACGAN model is trained using the original wind power scenario dataset, and after training, the generator is used as a wind power scenario generator, including: S41, set training parameters, including the number of training rounds Ep, batch training size B, optimizer, and learning rate lr; S42, for the mth typical load scenario P dm , using its corresponding original wind power scene dataset D wm Train the corresponding LSTM-ACGAN model. After the model training is completed, take the model generator M Gm As the mth typical load scenario P dm Corresponding wind power scenario generator; S43, for all typical load scenarios P d1 , P d2 ,…,P dM , respectively perform the operation steps in (S42), respectively obtain the wind power scenario generator corresponding to each typical load scenario, and together constitute the wind power scenario generator M for all typical loads G ={M Gm }, where m = 1, 2, ..., M.

6. The method according to claim 1, characterized in that Based on the typical load scenario and the type of wind power scenario to be generated, a wind power scenario similar to the real wind power scenario is generated by using the corresponding wind power scenario generator, including: S51, determine the load scenario type m, the wind power scenario type c to be generated and the number N of wind power scenarios to be generated, and for the mth type of typical load scenario, select the corresponding wind power scenario generator M Gm ; S52, towards M Gm Input N Z-dimensional Gaussian noise z and type label c, then the generator will output N typical load scenarios P dm The type of wind power scenario is c, and the specific wind power scenario generation is completed.

7. A specific wind power scenario generation device based on LSTM-ACGAN, characterized in that: include: A data processing module, used to obtain a historical load scenario data set and a historical wind power scenario data set based on the historical load data and historical wind power output data of the power system, and cluster the historical load scenario data set to obtain a typical load scenario data set; The wind power scenario classification module is used to obtain the wind abandonment rate corresponding to each wind power scenario in the historical wind power scenario data set by solving the optimization problem based on the typical load in the typical load scenario data set, and classify the wind power scenarios according to the range of wind abandonment rate to obtain the original wind power scenario data set; A model building module, used to build an LSTM-ACGAN model based on Bi-LSTM and a fully connected layer neural network; wherein the LSTM-ACGAN model includes a generator, a discriminator, and a classifier; The model training module is used to train the LSTM-ACGAN model using the original wind power scenario dataset and use the generator as a wind power scenario generator after the training is completed; The wind power scenario generation module is used to generate a wind power scenario similar to a real wind power scenario by using a corresponding wind power scenario generator based on a typical load scenario and the type of wind power scenario to be generated.

Images