New energy market clearing sample set batch generation method and system

Abstract

The invention discloses a new energy market clearing sample set batch generation method and system, and belongs to the technical field of electric power data processing. The method comprises the following steps: taking the generating capacity of a near electric field as source domain data, taking the generating capacity of a new electric field as target domain data, respectively inputting the source domain data and the target domain data into a generator and a discriminator, extracting time sequence characteristics of the generating data, and respectively obtaining source domain characteristics and target domain characteristics; mapping the source domain features and the target domain features to a manifold domain through a Grassmann manifold kernel function; calculating a geodesic distance between a source domain and a target domain in the manifold domain, and taking the geodesic distance as a loss function of a sample set generation model; and performing batch generation of new energy market clearing sample sets by using a sample set generation model corresponding to the loss function. According to the method, the photovoltaic generating capacity of the whole transformer area can be predicted through the power curve data of the single user.

Description

Technical Field

[0001] The present application proposes a method and system for batch generation of new energy market clearing sample sets, which relates to the technical field of power data processing. Background Art

[0002] With the rapid development of renewable energy installed capacity, renewable energy output will affect the supply and demand relationship of the overall electricity market. However, there are quality problems such as missing, duplication, and anomalies in the collection and transmission of power data, and poor data sample quality will affect the price mechanism of market clearing. Therefore, the sample set generation method plays an important role in power market clearing, power market trading, and economic dispatch. Constructing a higher quality sample set can make the stochastic optimization problem of the power system closer to the optimal solution. However, in the early stage of operation of renewable energy power plants, the data may be very limited or there may be no historical information, which will lead to inaccurate sample set generation, thereby affecting the reliability of decision-making results. Considering that multiple power plants in neighboring areas may have similar weather conditions, it is possible to consider using other existing data-rich power plant historical information to generate samples.

[0003] The use of nearby fields for sample set generation faces the following problems: First, new energy samples may be affected by various climate phenomena. The feature spaces of power plant data with similar geographical locations may be different, but there may also be a certain degree of similarity. The existing sample set generation method generates data in the original feature space, and the power generation data of the newly built electric field and the nearby electric field are in different feature spaces. Second, because the features in the original space are often distorted, it will cause data drift between the source domain data and the target domain data, which will make the learning results underfit. Summary of the invention

[0004] In order to solve the problems existing in the prior art, this application proposes a method and system for batch generation of new energy market clearing sample sets, which can predict the photovoltaic power generation of the entire area through the power curve data of a single user.

[0005] In order to solve the above technical problems, the technical solution adopted in this application is: In a first aspect, the present application provides a method for batch generating a new energy market clearing sample set, comprising: The power generation of the adjacent electric field is used as the source domain data, and the power generation of the newly built electric field is used as the target domain data. The source domain data and the target domain data are input into the generator and the discriminator respectively, and the time series features of the power generation data are extracted to obtain the source domain features and the target domain features respectively; Through the Grassmann manifold kernel function, the source domain features and the target domain features are mapped to the manifold domain; Calculate the geodesic distance between the source domain and the target domain in the manifold domain, and use the geodesic distance as the loss function of the sample set generation model; The sample set generation model corresponding to the loss function is used to batch generate new energy market clearing sample sets.

[0006] As a further improvement of the present invention, the generator and the discriminator are time-domain convolutional networks (TCNs).

[0007] As a further improvement of the present invention, the time series features of the power generation data are extracted to obtain source domain features and target domain features, which specifically include: Electricity generation data of source domain , target domain power generation data , for the input sequence , the causal convolution is expressed as:

[0008] in, is the convolution kernel, is the convolution kernel size, yes The input sequence of moments; Then the dilated convolution of TCN is expressed as:

[0009] in, is the expansion rate; The time series features of power generation data are extracted through TCN to obtain source domain features and target domain features .

[0010] As a further improvement of the present invention, mapping the source domain features and the target domain features to the manifold domain through the Grassmann manifold kernel function includes: Map the source domain features and target domain features into the manifold domain; for the source domain features and target domain features , in the manifold domain Calculate the corresponding geodesic ; The geodesic flow kernel function GFK is used to map the features in the original space to the manifold domain; the transformed features and The inner product of defines a semi-positive definite geodesic flow kernel. The GFK kernel function is defined as follows:

[0011] The features in the original space are obtained by Transformed to the Grassmann manifold, It is a form of expression.

[0012] As a further improvement of the present invention, the geodesic distance between the source domain and the target domain in the manifold domain is calculated, and the geodesic distance is used as the loss function of the sample set generation model, including: For source domain features mapped to manifold domain and target domain features , optimize the source domain features and target domain features The geodesic distance between them is based on the loss function of the manifold geodesic distance. The KL divergence of the source domain features and the target domain features on the manifold is calculated, and the loss function is as follows:

[0013] In the formula, for The probability density of the geodesic function, for The probability density of the geodesic function is given by Compute the KL divergence between two geodesics.

[0014] As a further improvement of the present invention, the sample set generation model corresponding to the loss function is used to generate a batch of new energy market clearing sample sets, including: The target domain data is input into a sample set generation model corresponding to the loss function to generate new energy market clearing sample sets in batches.

[0015] In a second aspect, the present application provides a new energy market clearing sample set batch generation system, comprising: The extraction module is used to use the power generation of the adjacent electric field as the source domain data and the power generation of the newly built electric field as the target domain data, input the source domain data and the target domain data into the generator and the discriminator respectively, extract the time series features of the power generation data, and obtain the source domain features and the target domain features respectively; A mapping module, used to map source domain features and target domain features to the manifold domain through the Grassmann manifold kernel function; A calculation module, used for calculating the geodesic distance between the source domain and the target domain in the manifold domain, and using the geodesic distance as the loss function of the sample set generation model; A generation module is used to generate new energy market clearing sample sets in batches using a sample set generation model corresponding to the loss function.

[0016] Further preferably, the extraction module is specifically used for: Electricity generation data of source domain , target domain power generation data , for the input sequence , the causal convolution is expressed as:

[0017] in, is the convolution kernel, is the convolution kernel size, yes The input sequence of moments; Then the dilated convolution of TCN is expressed as:

[0018] in, is the expansion rate; The time series features of power generation data are extracted through TCN to obtain source domain features and target domain features .

[0019] Further preferably, the mapping module is specifically used for: Map the source domain features and target domain features into the manifold domain; for the source domain features and target domain features , in the manifold domain Calculate the corresponding geodesic ; The geodesic flow kernel function GFK is used to map the features in the original space to the manifold domain; the transformed features and The inner product of defines a semi-positive definite geodesic flow kernel. The GFK kernel function is defined as follows:

[0020] The features in the original space are obtained by Transformed to the Grassmann manifold, It is a form of expression.

[0021] Further preferably, the calculation module is specifically used for: For source domain features mapped to manifold domain and target domain features , optimize the source domain features and target domain features The geodesic distance between them is based on the loss function of the manifold geodesic distance. The KL divergence of the source domain features and the target domain features on the manifold is calculated, and the loss function is as follows:

[0022] In the formula, for The probability density of the geodesic function, for The probability density of the geodesic function is given by Compute the KL divergence between two geodesics.

[0023] Further preferably, the generating module is specifically used for: The target domain data is input into a sample set generation model corresponding to the loss function to generate new energy market clearing sample sets in batches.

[0024] In a third aspect, the present application provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method for batch generating new energy market clearing sample sets when executing the computer program.

[0025] In a fourth aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for batch generating new energy market clearing sample sets is implemented.

[0026] In a fifth aspect, the present application provides a computer program product, which includes computer instructions, and the computer instructions instruct a computer to execute the method for batch generating new energy market clearing sample sets.

[0027] The beneficial effects of this application compared to the prior art are: This application proposes a method for batch generation of sample sets based on domain migration, and adopts a transfer learning method to learn the data distribution between the newly built electric field and the adjacent electric field. A large amount of source domain data is used to train the generator model, and a small amount of target domain data is used to train the discriminator model, and finally the distribution difference between the generator and discriminator outputs is optimized. Since the power generation data is a time series, the feature extraction module of the generator and discriminator in this application adopts TCN (Temporal Convolutional Network, time domain convolutional network), and TCN uses causal convolution to learn time series features. Specifically, the adjacent electric field (source domain) data is first input into the generator model, and the newly built electric field (target domain) data is input into the discriminator model, and the KL divergence between the generator and discriminator outputs is optimized.

[0028] Specifically, this application proposes a manifold-based feature transformation method. According to the manifold hypothesis, points embedded in the manifold space and their neighbors usually have similar properties. Therefore, this application proposes a manifold-based domain transfer learning method, which performs feature transformation based on the manifold space to reduce data drift between domains. The features are mapped to the manifold space through the Grassmann kernel function, and the geodesic function between the source domain and the target domain is calculated using the geodesic flow kernel. The geodesic distance between the source domain and the target domain is optimized by the KL divergence, so that the data generated using the source domain data is close to the target domain data in distribution. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following introduction is made to the drawings of the embodiments of the present application or the related technical solutions in the prior art. It should be understood that the drawings introduced below are only for the convenience of clearly describing some embodiments of the technical solutions of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without paying any creative work.

[0030] Figure 1 A flow chart of a method for batch generation of new energy market clearing sample sets provided for this application; Figure 2 This is a flow chart of a method for batch generation of new energy market clearing sample sets based on manifold domain migration in an embodiment of the present application; Figure 3 A schematic diagram of a TCN model provided in an embodiment of the present application; Figure 4 A schematic diagram of the visualization of geodesic distance provided in an embodiment of the present application; Figure 5 A schematic diagram of visualization of manifold domain data distribution provided in an embodiment of the present application; Figure 6 A device for batch generation of new energy market clearing sample sets provided for this application; Figure 7 A schematic diagram of an electronic device provided for this application. DETAILED DESCRIPTION

[0031] Embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and are not to be construed as limitations on the present application. For the step numbers in the following embodiments, they are only provided for the convenience of explanation, and the order between the steps is not limited in any way, and the execution order of each step in the embodiment can be adaptively adjusted according to the understanding of those skilled in the art.

[0032] In the description of this application, unless otherwise clearly defined, terms such as setting, installing, connecting, etc. should be understood in a broad sense, and technicians in the relevant technical field can reasonably determine the specific meanings of the above terms in this application based on the specific content of the technical solution.

[0033] This application aims to solve the problem of insufficient samples of newly built electric fields by using transfer learning to take adjacent electric fields as source domain data and learn target domain (newly built electric field) data. In order to solve the problem of distortion of the original feature space, a new energy sample set batch generation method based on manifold domain migration is proposed, which uses the topological invariance of manifold learning to solve the problem of data drift between domains.

[0034] Specifically, Figure 1 As shown, the first purpose of this application is to provide a method for batch generating a new energy market clearing sample set, including: S101. The power generation of the adjacent electric field is used as the source domain data, and the power generation of the newly built electric field is used as the target domain data. The source domain data and the target domain data are input into the generator and the discriminator respectively to obtain the time series characteristics of the power generation data.

[0035] S102. Map source domain features and target domain features to the manifold domain through the Grassmann manifold kernel function.

[0036] S103, calculating the geodesic distance between the source domain and the target domain in the manifold domain, and using the geodesic distance as the loss function of the sample set generation model.

[0037] S104: Batch generate new energy market clearing sample sets using a sample set generation model corresponding to the loss function.

[0038] In order to solve the problem of insufficient samples of newly built electric fields, this application uses the idea of ​​transfer learning and combines the characteristics of manifold learning to propose a method for batch generation of new energy market clearing sample sets. Among them, in the transfer learning application, the source domain data selection is to select the power generation data of neighboring electric fields as the source domain data. These electric fields are similar to the newly built electric fields in terms of geographical location, climatic conditions, etc., so their power generation data has a certain reference value for the newly built electric fields. The power generation data of the newly built electric fields is used as the target domain data. Since the newly built electric fields lack historical data, it is necessary to use the source domain data to assist in the generation of target domain data.

[0039] Furthermore, the generator and discriminator are used to process the source domain and target domain data to extract the time series features of the power generation data. This step helps to capture the dynamic changes and potential patterns of the data. The source domain features and target domain features are mapped to the manifold domain through the Grassmann manifold kernel function. Grassmann manifold is a mathematical tool that describes the low-dimensional embedding structure of high-dimensional data, which can preserve the local and global structural information of the data.

[0040] Furthermore, in the manifold domain, the topological structure of the data remains unchanged, which helps to solve the problem of distortion of the original feature space. The geodesic distance between the source domain and the target domain in the manifold domain is calculated. The geodesic distance is the shortest path length between two points on the manifold, which reflects the true distance of the data on the manifold. The geodesic distance is used as the loss function of the sample set generation model. By minimizing this loss function, the generated sample set can be closer to the real target domain data on the manifold.

[0041] Finally, the sample set generation model corresponding to the loss function is used to batch generate the new energy market clearing sample set. This step is based on the features extracted in the previous step, mapped to the manifold domain, and the calculated geodesic distance to generate a new energy market clearing sample set that conforms to the data characteristics of the target domain.

[0042] This application uses transfer learning to use the data of neighboring electric fields as source domain data to assist in generating target domain data for newly built electric fields, effectively solving the problem of insufficient samples for newly built electric fields. The topological invariance of manifold learning is used to solve the problem of data drift between domains, making the generated sample set closer to the real target domain data on the manifold, thereby improving the quality of the data. This method can batch generate new energy market clearing sample sets, meet the needs of large-scale data processing, and improve data processing efficiency. The method for batch generating new energy market clearing sample sets proposed in this application has significant advantages and innovations, and provides an effective solution to the problem of insufficient samples for newly built electric fields.

[0043] The following is an explanation of the implementation of the present application by means of specific embodiments. People familiar with the art can easily understand other advantages and effects of the present application from the contents disclosed in this specification.

[0044] Please refer to Figure 2 , Figure 2 This is a flow chart of a new energy market clearing sample set batch generation technology in an embodiment of the present application. The method includes the following steps: S101. The power generation of the adjacent electric field is used as the source domain data, and the power generation of the newly built electric field is used as the target domain data. The source domain data and the target domain data are input into the generator and the discriminator respectively to obtain the time series characteristics of the power generation data.

[0045] This application uses TCN as the feature extraction module of the generator and discriminator. The core advantage of TCN (Temporal Convolutional Network) is that it can effectively capture local and long-term dependencies in time series data through convolution operations, while avoiding the gradient vanishing and exploding problems in traditional recurrent neural networks.

[0046] like Figure 3 As shown in the figure, the hidden layer is a stack of causal convolutional layers. Compared with general convolutional neural networks, TCN introduces causal convolution to ensure that the model can predict the current time point. When outputting, the information of future time points will not be used, thus maintaining the causality of the time series. Compared with recurrent neural networks RNN and LSTM, TCN introduces dilated convolution to expand the receptive field of the convolution kernel. 2 and d 4 increases the step size of the convolution, so that the model can capture longer-term dependencies without increasing the number of parameters. Therefore, this application uses TCN to extract features from power generation data.

[0047] Power generation data of source domain , target domain (target) power generation data For the input sequence , the mathematical representation of causal convolution is: (1) in, is the convolution kernel, is the convolution kernel size, is the input sequence.

[0048] In order to increase the receptive field without increasing the number of parameters, TCN uses dilated convolution. Dilated convolution, also known as hole convolution, inserts gaps between convolution kernels (i.e. skips some input units), allowing the model to capture a larger receptive field without increasing the number of parameters, thereby better understanding the contextual information in the input data. The mathematical representation of dilated convolution is: (2) in, is the expansion rate.

[0049] Furthermore, the source domain features are obtained by extracting the time series features of power generation data. and target domain features .

[0050] S102. Map source domain features and target domain features to the manifold domain through the Grassmann manifold kernel function.

[0051] To solve the problem of distortion of the original feature space, this application maps the source domain features and the target domain features to the manifold domain. In the Grassmann manifold, feature transformation and distribution adaptation usually have effective numerical forms, which can be efficiently represented and solved in the transfer learning problem. Manifold learning can maintain the local structure and geometric features of the data and maintain topological invariance in the distorted space, so the manifold domain features have good robustness.

[0052] For source domain features and target domain features , in the manifold domain Calculate their geodesics in In a manifold, a geodesic is a special curve that describes the shortest or most direct path between two points on the manifold. In a small enough neighborhood, the properties of a geodesic are similar to those of a straight line in Euclidean space. However, in a larger range, the path of a geodesic may change due to the curvature of the manifold. The schematic diagram of the geodesic between the source domain and the target domain is shown in Figure 4 As shown, Represents the location of the source domain data in the manifold space, Represents the location of the target domain data in the manifold space, the geodesic It is topologically invariant on manifolds.

[0053] This application uses the geodesic flow kernel function GFK (Geodesic Flow Kernel) to map the features in the original space to the manifold domain.

[0054] In this application, GFK is based on the kernel method, which calculates the similarity between data points by defining a kernel function without explicitly mapping the data to a high-dimensional space. This method can be seen as a arrive Incremental "walking" method. New features The transformed features and The inner product of defines a semi-positive definite geodesic flow kernel. The GFK kernel function is defined as follows: (3) Therefore, the features in the original space can be expressed by Transformed to the Grassmann manifold, note that It's just a form of expression.

[0055] S103, calculating the geodesic distance between the source domain and the target domain in the manifold domain, and using the geodesic distance as the loss function of the sample set generation model.

[0056] For source domain features mapped to manifold domain and target domain features , optimizing the geodesic distance between them, such as Figure 5 As shown, the sample set generation model can learn the data distribution of the target domain. Therefore, this application proposes a loss function based on the manifold geodesic distance to calculate the KL divergence of the source domain features and the target domain features on the manifold. The loss function is as follows: (4) In the formula, for The probability density of the geodesic function, for The probability density of the geodesic function is given by Compute the KL divergence between two geodesics.

[0057] The advantage of the present invention is that the source domain data and the target domain are mapped to the manifold space, and the geodesic distance between the source domain data and the target data is optimized on the manifold. Compared with the traditional method of generating data by transfer learning, the present invention takes into account the problem of feature space distortion and uses the topological invariance of the manifold to introduce the geodesic distance into the market clearing sample set generation method for the first time.

[0058] Using the generated sample set to participate in electricity market clearing can supplement the lack of historical data, optimize the market clearing model, assist market players in decision-making, optimize the design and management of the electricity market, and promote the effective use of renewable energy.

[0059] like Figure 6 As shown, the third purpose of the present application is to provide a new energy market clearing sample set batch generation system, the new energy market clearing sample set batch generation system of the present application is based on the new energy market clearing sample set batch generation method. The system specifically includes: an extraction module 100, a mapping module 200, a calculation module 300, and the generation module 400.

[0060] Among them, the extraction module 100 is used to use the power generation of the adjacent electric field as the source domain data and the power generation of the newly built electric field as the target domain data, input the source domain data and the target domain data into the generator and the discriminator respectively, extract the time series features of the power generation data, and obtain the source domain features and the target domain features respectively.

[0061] Furthermore, the extraction module 100 is specifically used for: Electricity generation data of source domain , target domain power generation data , for the input sequence , the causal convolution is expressed as:

[0062] in, is the convolution kernel, is the convolution kernel size, yes The input sequence of moments; Then the dilated convolution of TCN is expressed as:

[0063] in, is the expansion rate; The time series features of power generation data are extracted through TCN to obtain source domain features and target domain features .

[0064] The mapping module 200 is used to map the source domain features and the target domain features to the manifold domain through the Grassmann manifold kernel function.

[0065] Further, the mapping module is specifically used for: Map the source domain features and target domain features into the manifold domain; for the source domain features and target domain features , in the manifold domain Calculate the corresponding geodesic ; The geodesic flow kernel function GFK is used to map the features in the original space to the manifold domain; the transformed features and The inner product of defines a semi-positive definite geodesic flow kernel. The GFK kernel function is defined as follows:

[0066] The features in the original space are obtained by Transformed to the Grassmann manifold, It is a form of expression.

[0067] The calculation module 300 is used to calculate the geodesic distance between the source domain and the target domain in the manifold domain, and use the geodesic distance as the loss function of the sample set generation model.

[0068] Furthermore, the calculation module 300 is specifically used for: For source domain features mapped to manifold domain and target domain features , optimize the source domain features and target domain features The geodesic distance between them is based on the loss function of the manifold geodesic distance. The KL divergence of the source domain features and the target domain features on the manifold is calculated, and the loss function is as follows:

[0069] In the formula, for The probability density of the geodesic function, for The probability density of the geodesic function is given by Compute the KL divergence between two geodesics.

[0070] Among them, the generation module 400 is used to generate new energy market clearing sample sets in batches using the sample set generation model corresponding to the loss function.

[0071] Furthermore, the generation module 400 is specifically used to: input the target domain data into a sample set generation model corresponding to the loss function to batch generate new energy market clearing sample sets.

[0072] like Figure 7 As shown, the third purpose of the embodiment of the present application is to provide an electronic device, including a memory 701, a processor 702, and a computer program stored in the memory 701 and executable on the processor, wherein the processor implements the above-mentioned method for batch generation of new energy market clearing sample sets when executing the computer program. It also includes a communication interface 703 and a bus 704.

[0073] The fourth objective of an embodiment of the present application is to provide a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned method for batch generating new energy market clearing sample sets is implemented.

[0074] The fifth objective of an embodiment of the present application is to provide a computer program product, which includes computer instructions, and the computer instructions instruct a computer to execute the above-mentioned method for batch generating new energy market clearing sample sets.

[0075] 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.

[0076] 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 in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0077] The present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, readable storage medium, optical storage, etc.) containing computer-usable program codes.

[0078] 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.

[0079] Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work should fall within the scope of protection of the present application.

[0080] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present application rather than to limit it. Although the present application has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present application can still be modified or replaced by equivalents, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present application should be included in the scope of protection of the claims of the present application.

Claims

1. A method for batch generation of new energy market clearing sample sets, characterized in that: include: The power generation of the adjacent electric field is used as the source domain data, and the power generation of the newly built electric field is used as the target domain data. The source domain data and the target domain data are input into the generator and the discriminator respectively, and the time series features of the power generation data are extracted to obtain the source domain features and the target domain features respectively; Through the Grassmann manifold kernel function, the source domain features and the target domain features are mapped to the manifold domain; Calculate the geodesic distance between the source domain and the target domain in the manifold domain, and use the geodesic distance as the loss function of the sample set generation model; The sample set generation model corresponding to the loss function is used to batch generate new energy market clearing sample sets.

2. The method for batch generation of new energy market clearing sample sets according to claim 1 is characterized in that: The generator and discriminator are time domain convolutional networks (TCN).

3. The method for batch generating new energy market clearing sample sets according to claim 2 is characterized in that: The time series features of the power generation data are extracted to obtain source domain features and target domain features, respectively, including: Electricity generation data of source domain , target domain power generation data , for the input sequence , the causal convolution is expressed as: in, is the convolution kernel, is the convolution kernel size, yes The input sequence of moments; Then the dilated convolution of TCN is expressed as: in, is the expansion rate; The time series features of power generation data are extracted through TCN to obtain source domain features and target domain features .

4. The method for batch generating new energy market clearing sample sets according to claim 2 is characterized in that: The source domain features and the target domain features are mapped to the manifold domain through the Grassmann manifold kernel function, including: Map the source domain features and target domain features into the manifold domain; for the source domain features and target domain features , in the manifold domain Calculate the corresponding geodesic ; The geodesic flow kernel function GFK is used to map the features in the original space to the manifold domain; the transformed features and The inner product of defines a semi-positive definite geodesic flow kernel. The GFK kernel function is defined as follows: The features in the original space are obtained by Transformed to the Grassmann manifold, It is a form of expression.

5. The method for batch generating new energy market clearing sample sets according to claim 1 is characterized in that: The step of calculating the geodesic distance between the source domain and the target domain in the manifold domain and using the geodesic distance as the loss function of the sample set generation model includes: For source domain features mapped to manifold domain and target domain features , optimize the source domain features and target domain features The geodesic distance between them is based on the loss function of the manifold geodesic distance. The KL divergence of the source domain features and the target domain features on the manifold is calculated, and the loss function is as follows: In the formula, for The probability density of the geodesic function, for The probability density of the geodesic function is given by Compute the KL divergence between two geodesics.

6. The method for batch generation of new energy market clearing sample sets according to claim 1 is characterized in that: The step of performing batch generation of new energy market clearing sample sets using the sample set generation model corresponding to the loss function includes: The target domain data is input into a sample set generation model corresponding to the loss function to generate new energy market clearing sample sets in batches.

7. A new energy market clearing sample set batch generation system, characterized by: include: The extraction module is used to use the power generation of the adjacent electric field as the source domain data and the power generation of the newly built electric field as the target domain data, input the source domain data and the target domain data into the generator and the discriminator respectively, extract the time series features of the power generation data, and obtain the source domain features and the target domain features respectively; A mapping module, used to map source domain features and target domain features to the manifold domain through the Grassmann manifold kernel function; A calculation module, used for calculating the geodesic distance between the source domain and the target domain in the manifold domain, and using the geodesic distance as the loss function of the sample set generation model; A generation module is used to generate batches of new energy market clearing sample sets using a sample set generation model corresponding to the loss function.

8. A new energy market clearing sample set batch generation system according to claim 7, characterized in that: The generator and discriminator are time domain convolutional networks (TCN).

9. A new energy market clearing sample set batch generation system according to claim 8, characterized in that: The extraction module is specifically used for: Electricity generation data of source domain , target domain power generation data , for the input sequence , the causal convolution is expressed as: in, is the convolution kernel, is the convolution kernel size, yes The input sequence of moments; Then the dilated convolution of TCN is expressed as: in, is the expansion rate; The time series features of power generation data are extracted through TCN to obtain source domain features and target domain features .

10. A new energy market clearing sample set batch generation system according to claim 7, characterized in that: The mapping module is specifically used for: Map the source domain features and target domain features into the manifold domain; for the source domain features and target domain features , in the manifold domain Calculate the corresponding geodesic ; The geodesic flow kernel function GFK is used to map the features in the original space to the manifold domain; the transformed features and The inner product of defines a semi-positive definite geodesic flow kernel. The GFK kernel function is defined as follows: The features in the original space are obtained by Transformed to the Grassmann manifold, It is a form of expression.

11. A new energy market clearing sample set batch generation system according to claim 7, characterized in that: The computing module is specifically used for: For source domain features mapped to manifold domain and target domain features , optimize the source domain features and target domain features The geodesic distance between them is based on the loss function of the manifold geodesic distance. The KL divergence of the source domain features and the target domain features on the manifold is calculated, and the loss function is as follows: In the formula, for The probability density of the geodesic function, for The probability density of the geodesic function is given by Compute the KL divergence between two geodesics.

12. A new energy market clearing sample set batch generation system according to claim 7, characterized in that: The generation module is specifically used for: The target domain data is input into a sample set generation model corresponding to the loss function to generate new energy market clearing sample sets in batches.

13. An electronic device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the method for batch generating new energy market clearing sample sets as claimed in any one of claims 1 to 6 when executing the computer program.

14. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for batch generating new energy market clearing sample sets as described in any one of claims 1-6 is implemented.

15. A computer program product, comprising computer instructions, characterized in that: The computer instructions instruct the computer to execute the method for batch generating new energy market clearing sample sets as described in any one of claims 1-6.

Images