A method for evaluating the generation quality of wind power low-output process samples of a generation model

Abstract

The embodiment of the present disclosure provides a method for evaluating the sample generation quality of a generation model in a low-output process of wind power, and relates to the technical field of new energy, comprising: obtaining real samples of a wind farm in a low-output process of wind power and generated samples of a generation model in the low-output process of wind power; wherein the generation model is trained by using the real samples; calculating the data distribution similarity between the real samples and the generated samples to obtain a first index value; calculating the mapping relationship similarity between the real samples and the generated samples to obtain a second index value; weighting and summing the first index value and the second index value to obtain a comprehensive index value, and obtaining a sample generation quality evaluation result of the generation model based on the comprehensive index value. The present application can fuse the comprehensive evaluation of two dimensions of global distribution consistency and key physical feature mapping relationship, and comprehensively reflect the physical authenticity and practicability of the generated samples of the generation model.

Description

Technical Field

[0001] This disclosure relates to the field of new energy technology, and in particular to a method for evaluating the quality of wind power low-output process sample generation using a generative model. Background Technology

[0002] Wind power generation is significantly affected by meteorological conditions, resulting in highly intermittent and fluctuating power output. A period during which wind power output remains below a certain threshold of rated capacity is termed a low-output wind power process. Low-output wind power processes are a critical scenario threatening the safe and stable operation of the power system. Due to the low probability and scarcity of low-output wind power events, generative methods are often used to expand the sample size. However, the quality of samples generated by generative models varies greatly and is prone to distortion of physical meaning. Therefore, it is necessary to evaluate the sample generation quality of generative models.

[0003] Existing methods for evaluating the quality of generated samples often rely on a single indicator, such as the mean or variance, or use KL divergence or JS divergence to measure the overall difference in distribution. While these indicators can reflect first- or second-order statistical characteristics to some extent, they are difficult to comprehensively assess the consistency between generated samples and real samples. This makes it impossible to accurately select physically reasonable and statistically reliable generated models and samples, which in turn cannot support subsequent power prediction modeling and affects the credibility of power system reliability analysis and risk warning. Summary of the Invention

[0004] To address, or at least partially address, the aforementioned technical problems, this disclosure provides a method for evaluating the quality of wind power low-output process sample generation using generative models.

[0005] This disclosure provides a method for evaluating the quality of sample generation during low-output wind power processes using a generative model. The method includes: acquiring real samples from a wind farm during a low-output wind power process and generated samples from a generative model during the same process; wherein the generative model is trained using real samples; calculating the data distribution similarity between real samples and generated samples to obtain a first index value; calculating the mapping relationship similarity between real samples and generated samples to obtain a second index value; weighted summing of the first and second index values ​​to obtain a comprehensive index value; and obtaining the sample generation quality evaluation result of the generative model based on the comprehensive index value.

[0006] In one possible implementation, calculating the data distribution similarity between real samples and generated samples to obtain a first index value includes: extracting multiple real feature sequences from real samples and extracting multiple corresponding generated feature sequences from generated samples; for each feature, calculating the Vascosian distance between the real feature sequence and the generated feature sequence to obtain a feature distance value; and weighted summing of multiple feature distance values ​​to obtain the first index value.

[0007] In one possible implementation, calculating the Vacancies distance between the true feature sequence and the generated feature sequence to obtain the feature distance value includes: calculating the empirical cumulative distribution function of the true feature sequence to obtain a first distribution function; calculating the empirical cumulative distribution function of the generated feature sequence to obtain a second distribution function; and calculating the Vacancies distance between the first distribution function and the second distribution function to obtain the feature distance value.

[0008] In one possible implementation, calculating the similarity of the mapping relationship between real samples and generated samples to obtain a second index value includes: extracting multiple real feature sequences from real samples and extracting corresponding multiple generated feature sequences from generated samples; calculating the linear correlation degree between real features based on multiple real feature sequences to obtain a first cross-correlation matrix; calculating the linear correlation degree between generated features based on multiple generated feature sequences to obtain a second cross-correlation matrix; and calculating the F-norm difference between the first cross-correlation matrix and the second cross-correlation matrix to obtain the second index value.

[0009] In one possible implementation, the linear correlation between real features is calculated based on multiple real feature sequences to obtain a first cross-correlation matrix, including: constructing a real feature vector and a real feature mean vector based on multiple real feature sequences; calculating the linear correlation between real features based on the real feature vector and the real feature mean vector to obtain the first cross-correlation matrix; and calculating the linear correlation between generated features based on multiple generated feature sequences to obtain a second cross-correlation matrix, including: constructing a generated feature vector and a generated feature mean vector based on multiple generated feature sequences; calculating the linear correlation between generated features based on the generated feature vector and the generated feature mean vector to obtain the second cross-correlation matrix.

[0010] In one possible implementation, the sample generation quality evaluation result of the generative model is obtained based on the comprehensive index value, including: comparing the comprehensive index value with the comprehensive index threshold; when the comprehensive index value is less than the comprehensive index threshold, the sample generation quality of the generative model is judged to be qualified; when the comprehensive index value is greater than or equal to the comprehensive index threshold, the sample generation quality of the generative model is judged to be unqualified.

[0011] In one possible implementation, the method for evaluating the quality of wind power low-output process samples generated by the generative model further includes: obtaining multiple generative models with qualified sample generation quality; comparing the comprehensive index values ​​of multiple generative models and selecting the generative model corresponding to the largest comprehensive index value as the target generative model; using the target generative model to generate target generated samples, and combining the target generated samples and real samples to train the power prediction model for wind power low-output processes.

[0012] This disclosure also provides a device for evaluating the quality of wind power low-output process samples generated by a generative model. The device includes: an acquisition module for acquiring real samples of the wind power low-output process and generated samples from the generative model; wherein the generative model is trained using real samples; a first calculation module for calculating the data distribution similarity between the real samples and the generated samples to obtain a first index value; a second calculation module for calculating the mapping relationship similarity between the real samples and the generated samples to obtain a second index value; and an evaluation module for weighted summation of the first index value and the second index value to obtain a comprehensive index value, and obtaining the sample generation quality evaluation result of the generative model based on the comprehensive index value.

[0013] This disclosure also provides a computing device, which includes: a processor; a memory for storing processor-executable instructions; the processor being configured to read the executable instructions from the memory and execute the instructions to implement the wind power low-output process sample generation quality evaluation method provided in this disclosure.

[0014] This disclosure also provides a computer-readable storage medium storing a computer program for executing the wind power low-output process sample generation quality evaluation method provided in this disclosure.

[0015] In this implementation, real samples and generated samples are first acquired, and the similarity of their data distribution and feature mapping relationships is calculated. The first and second indicator values ​​are then weighted and summed to obtain a comprehensive indicator value, which comprehensively evaluates the quality of the generated samples. This method integrates the comprehensive evaluation of two dimensions: global distribution consistency and key physical feature mapping relationships. It overcomes the limitations of a single indicator and can more comprehensively reflect the physical authenticity and practicality of the generated samples. This provides effective support for sample expansion and quality assurance in low-output wind power scenarios and further ensures the effectiveness of the expanded samples in subsequent predictive modeling. Attached Figure Description

[0016] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and the originals and elements are not necessarily drawn to scale.

[0017] Figure 1 A flowchart illustrating a method for evaluating the quality of wind power low-output process samples in a generative model, as provided in this embodiment of the disclosure. Figure 2A schematic diagram of a wind power low-output process sample generation quality evaluation device provided in an embodiment of this disclosure; Figure 3 This is a schematic diagram of the structure of a computing device provided in an embodiment of the present disclosure. Detailed Implementation

[0018] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0019] It should be understood that the steps described in the method embodiments of this disclosure may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of this disclosure is not limited in this respect.

[0020] The term "comprising" and its variations as used herein are open-ended inclusions, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Definitions of other terms will be given in the description below.

[0021] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0022] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0023] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0024] Existing sample generation evaluation methods often rely on single indicators, making it difficult to comprehensively reflect the consistency between generated samples and real samples in terms of distribution characteristics and inter-feature correlation structures. Especially for low-output wind power processes, generated samples may not be able to approximate real data in terms of the marginal distribution of power and meteorological variables, nor can they restore the mapping relationship between numerical weather prediction data and wind power output. However, existing evaluation methods cannot provide a comprehensive assessment, which will lead to physical distortion of the generated samples and make them unable to support subsequent power prediction modeling.

[0025] To address the aforementioned issues, this disclosure provides a method for evaluating the quality of wind power low-output process samples generated by a generative model. The method includes: acquiring real samples from a wind farm during a low-output process and generated samples from a generative model during the same process; wherein the generative model is trained using real samples; calculating the data distribution similarity between real and generated samples to obtain a first index value; calculating the mapping relationship similarity between real and generated samples to obtain a second index value; weighted summing of the first and second index values ​​to obtain a comprehensive index value; and obtaining the sample generation quality evaluation result based on the comprehensive index value. In this implementation, real and generated samples are first acquired, and the data distribution similarity and feature mapping relationship similarity between them are calculated respectively. The first and second index values ​​are then weighted and summed to obtain a comprehensive index value, which comprehensively evaluates the quality of the generated samples. This method integrates the comprehensive evaluation of two dimensions: global distribution consistency and key physical feature mapping relationship. It overcomes the limitations of a single indicator and can more comprehensively reflect the physical authenticity and practicality of the generated samples of the generative model. It provides effective support for sample expansion and quality assurance in low-output wind power scenarios and further ensures the effectiveness of the expanded samples in subsequent prediction modeling.

[0026] The method will be described below with reference to specific embodiments.

[0027] Figure 1 This is a flowchart illustrating a method for evaluating the quality of wind power low-output process samples generated from a generative model, provided in an embodiment of this disclosure. This method can be executed by a device for evaluating the quality of wind power low-output process samples generated from a generative model. This device can be implemented using software and / or hardware, and is generally integrated into a computing device. Figure 1 As shown, the method includes: S101. Obtain real samples of wind farms during low wind power output and generated samples of the generated model during low power output.

[0028] Among them, the low output process of wind power refers to the typical operating condition in which the overall output power of the wind farm is in the range of low rated power, the power fluctuation is slow or slightly oscillating, and the wind energy utilization efficiency is low.

[0029] Historical operating data of wind farms during periods of low wind power output are obtained to obtain real samples.

[0030] Specifically, complete historical operational data of the wind farm is collected, including both normal and low-output periods. The power output within this data is identified to determine the output data during low-output periods, and the corresponding environmental data is extracted. Real-world environmental data and real-world output data are then used as the basis for the analysis.

[0031] The environmental data includes wind speed data, wind direction data, etc.

[0032] The generative model is trained using real samples, and the generated environmental data and generated force data are output from the trained generative model to obtain the generated samples.

[0033] Among them, the generative model can be a multi-generator-single-discriminator generative adversarial network, etc.

[0034] S102. Calculate the data distribution similarity between the real sample and the generated sample to obtain the first index value.

[0035] Specifically, multiple real feature sequences are extracted from real samples, and multiple corresponding generated feature sequences are extracted from generated samples; for each feature, the Vascosian distance between the real feature sequence and the generated feature sequence is calculated to obtain the feature distance value; the multiple feature distance values ​​are weighted and summed to obtain the first index value.

[0036] In one possible implementation, the Wasserstein distance between the real sample and the generated sample is calculated, and the Wasserstein distance is used as the first index value.

[0037] Specifically, multiple real feature sequences are extracted from real samples, and corresponding multiple generated feature sequences are extracted from generated samples. For example, real wind speed sequences, real wind direction sequences, and real power output sequences are extracted, and correspondingly, generated wind speed sequences, generated wind direction sequences, and generated power output sequences are extracted.

[0038] Further, the empirical cumulative distribution function of the true feature sequence is calculated to obtain the first distribution function; the empirical cumulative distribution function of the generated feature sequence is calculated to obtain the second distribution function. For example, the empirical cumulative distribution function is calculated for both the true and generated feature sequences for each feature, assuming... To complete the separable metric space, the probability measure corresponding to the first distribution function is: The probability measure corresponding to the second distribution function is , and for The upper has limited The probability measure of the order moment.

[0039] Further, the Warburg distance between the first and second distribution functions is calculated to obtain the feature distance value. Specifically, The Vascular distance is expressed as:

[0040] In the formula, Indicated by and Let be the set of all joint measures of the marginal distribution. , For the optimal transportation plan, For sample points and The spatial distance corresponds to the minimization of the total transportation cost by the infimum.

[0041] Calculate the feature distance value for each feature separately. For example, the feature distance values ​​for wind speed, wind direction, and power output features are calculated as follows: , , .

[0042] Furthermore, the first index value is obtained by weighted summation of multiple feature distance values. Understandably, the weight corresponding to each feature distance value is set according to the importance of the corresponding feature. Specifically, the first index value is expressed as:

[0043] In the formula, , , These are the weights of the feature distance values ​​for wind speed, wind direction, and power output characteristics, respectively. In one example, , , They have the same value.

[0044] Among them, the smaller the value of the first indicator, the closer the generated sample is to the statistical distribution of the real data, and the better the feature marginal distribution fits.

[0045] In this implementation, the Wasserstein distance is used as an evaluation metric for the similarity between the generated samples and the real sample data distribution. Based on optimal transport theory, the Wasserstein distance quantifies the minimum cost required to transform one probability distribution into another. Compared to traditional metrics such as KL divergence and JS divergence, it can still provide a smooth and physically meaningful difference value even if the support sets of the two distributions do not overlap, and it has better stability and gradient friendliness.

[0046] S103. Calculate the similarity of the mapping relationship between the real sample and the generated sample to obtain the second index value.

[0047] Specifically, multiple real feature sequences are extracted from real samples, and multiple corresponding generated feature sequences are extracted from generated samples; the linear correlation between real features is calculated based on multiple real feature sequences to obtain a first cross-correlation matrix; the linear correlation between generated features is calculated based on multiple generated feature sequences to obtain a second cross-correlation matrix; the F-norm difference between the first cross-correlation matrix and the second cross-correlation matrix is ​​calculated to obtain a second index value.

[0048] In one possible implementation, the Frobenius norm of the cross-correlation matrix between the real samples and the generated samples is calculated, and the Frobenius norm of the cross-correlation matrix is ​​used as a second index value.

[0049] Specifically, multiple real feature sequences are extracted from real samples, and corresponding multiple generated feature sequences are extracted from generated samples. For example, real wind speed sequences, real wind direction sequences, and real power output sequences are extracted, and correspondingly, generated wind speed sequences, generated wind direction sequences, and generated power output sequences are extracted.

[0050] Furthermore, the linear correlation degree between real features is calculated based on multiple real feature sequences to obtain the first cross-correlation matrix.

[0051] Specifically, a true feature vector is constructed based on multiple true feature sequences. The true feature mean vector is .in, This represents the actual number of samples. The dimension of the sample features.

[0052] The linear correlation between the true features is calculated based on the true feature vector and the true feature mean vector, resulting in the first cross-correlation matrix. Specifically, the first cross-correlation matrix is ​​expressed as:

[0053] Similarly, a generated feature vector is constructed based on multiple generated feature sequences. The generated feature mean vector is .in, To generate the number of samples, The dimension of the sample features.

[0054] The linear correlation between the generated features is calculated based on the generated feature vectors and the generated feature mean vectors, resulting in the second cross-correlation matrix. Specifically, the second cross-correlation matrix is ​​expressed as:

[0055] Further, the F-norm difference between the first cross-correlation matrix and the second cross-correlation matrix is ​​calculated to obtain the second index value. Specifically, the Frobenius norm difference between the cross-correlation matrices of the real sample and the generated sample is expressed as:

[0056] In the formula, Represents the first cross-correlation matrix. Line 1 Column elements, Indicates the second mutual matrix. Line 1 The elements of the column.

[0057] The larger the value of the second indicator, the greater the deviation between the real sample and the generated sample in the linear correlation pattern between features, and the worse the fitted effect of the generated sample on the intrinsic structure of the real sample. The smaller the value of the second indicator, the smaller the deviation between the real sample and the generated sample in the linear correlation pattern between features, the more similar the feature correlation characteristics, and the stronger the structural learning ability of the generative model.

[0058] In this implementation, the Frobenius norm is used as a metric to quantify the structural differences between two types of cross-correlation matrices. Essentially, the Frobenius norm is the square root of the sum of the squares of all elements in the matrix, possessing excellent properties such as nonnegativity, homogeneity, and trigonometric inequalities, comprehensively and intuitively reflecting the overall differences at the element level. Compared to other matrix difference metrics, the Frobenius norm does not require eigenvalue decomposition of the matrix, resulting in high computational efficiency and clear physical meaning, making it suitable for methodological scenarios such as generative model performance evaluation and sample distribution consistency testing. This application introduces the Frobenius norm difference of the cross-correlation matrix, enabling a quantitative assessment of whether the generated samples accurately reproduce the intrinsic structural relationship between meteorological elements and wind power output, thus more comprehensively reflecting the physical authenticity and practicality of the generated samples.

[0059] S104. The first index value and the second index value are weighted and summed to obtain the comprehensive index value, and the sample generation quality evaluation result of the generation model is obtained based on the comprehensive index value.

[0060] Specifically, the comprehensive index value is obtained by weighted summing of the first and second index values:

[0061] In the formula, , These are the weights of the feature distance values ​​of the first and second indicator values, respectively.

[0062] In one possible implementation, a generative model is evaluated for its suitability based on a comprehensive index value.

[0063] Specifically, the comprehensive index value is compared with the comprehensive index threshold. When the comprehensive index value is less than the comprehensive index threshold, the sample generation quality of the generative model is judged to be qualified; when the comprehensive index value is greater than or equal to the comprehensive index threshold, the sample generation quality of the generative model is judged to be unqualified.

[0064] In one possible implementation, for multiple generative models, the optimal generative model is evaluated based on a comprehensive index value.

[0065] Specifically, multiple generative models with qualified sample generation quality are obtained; among them, multiple generative models can be generative models with different model structures or different hyperparameter settings.

[0066] The comprehensive index values ​​of multiple generative models are compared, and the generative model corresponding to the largest comprehensive index value is selected as the target generative model. The target generative model is used to generate target generative samples, and the target generative samples and real samples are combined to train the power prediction model for low-output wind power processes.

[0067] In this implementation, real samples and generated samples are first acquired, and the similarity of their data distribution and feature mapping relationships is calculated. The first and second index values ​​are then weighted and summed to obtain a comprehensive index value, which comprehensively evaluates the quality of the generated samples. This method integrates the comprehensive evaluation of two dimensions: global distribution consistency and key physical feature mapping relationships. It also examines marginal distributions and mapping relationships between variables, effectively preventing the problem of pursuing only distribution similarity while ignoring inherent physical constraints. It overcomes the limitations of a single index and can more comprehensively reflect the physical authenticity and practicality of the generated samples, providing a quantitative basis for performance comparison and hyperparameter optimization of generative models such as generative adversarial networks and diffusion models. This provides effective support for sample expansion and quality assurance in low-output wind power scenarios and further ensures the effectiveness of expanded samples in subsequent predictive modeling.

[0068] To achieve the above embodiments, this disclosure also proposes a sample generation quality evaluation device for low-output wind power processes using generative models.

[0069] Figure 2 This is a schematic diagram of a wind power low-output process sample generation quality evaluation device provided in an embodiment of this disclosure. The device can be implemented by software and / or hardware, and is generally integrated into a computing device. Figure 2 As shown, the sample generation quality evaluation device for low-output wind power processes in this generative model includes: The acquisition module 201 is used to acquire real samples of the low-output wind power process and generated samples of the generative model; wherein, the generative model is trained using real samples.

[0070] The first calculation module 202 is used to calculate the data distribution similarity between the real sample and the generated sample to obtain the first index value.

[0071] The second calculation module 203 is used to calculate the similarity of the mapping relationship between the real sample and the generated sample to obtain the second index value.

[0072] Evaluation module 204 is used to obtain a comprehensive index value by weighted summation of the first and second index values, and to generate a quality evaluation result for the sample generation model based on the comprehensive index value.

[0073] In one possible implementation, the first computing module 202 includes: The first extraction unit is used to extract multiple real feature sequences from real samples and multiple corresponding generated feature sequences from generated samples.

[0074] The distance calculation unit is used to calculate the Vascular distance between the true feature sequence and the generated feature sequence for each feature, and obtain the feature distance value.

[0075] The weighted calculation unit is used to sum multiple feature distance values ​​by weight to obtain the first index value.

[0076] In one possible implementation, the distance calculation unit includes: The first computational subunit is used to calculate the empirical cumulative distribution function of the true feature sequence, and obtain the first distribution function.

[0077] The second computational subunit is used to calculate the empirical cumulative distribution function of the generated feature sequence, thus obtaining the second distribution function.

[0078] The third calculation subunit is used to calculate the Warburg distance between the first distribution function and the second distribution function to obtain the characteristic distance value.

[0079] In one possible implementation, the second computing module 203 includes: The second extraction unit is used to extract multiple real feature sequences from real samples and multiple corresponding generated feature sequences from generated samples.

[0080] The correlation calculation unit is used to calculate the linear correlation between real features based on multiple real feature sequences to obtain the first cross-correlation matrix; and to calculate the linear correlation between generated features based on multiple generated feature sequences to obtain the second cross-correlation matrix.

[0081] The difference calculation unit is used to calculate the F-norm difference between the first cross-correlation matrix and the second cross-correlation matrix to obtain the second index value.

[0082] In one possible implementation, the relevant computational unit includes: The fourth computational subunit is used to construct the true feature vector and the true feature mean vector based on multiple true feature sequences.

[0083] The fifth calculation subunit is used to calculate the linear correlation degree between the true features based on the true feature vector and the true feature mean vector, and obtain the first cross-correlation matrix.

[0084] In one possible implementation, the relevant computational unit includes: The sixth computational subunit is used to construct a generated feature vector and a generated feature mean vector based on multiple generated feature sequences.

[0085] The seventh calculation subunit is used to calculate the linear correlation degree between generated features based on the generated feature vector and the generated feature mean vector, and to obtain the second cross-correlation matrix.

[0086] In one possible implementation, the evaluation module 204 includes: The comparison unit is used to compare the comprehensive indicator value with the comprehensive indicator threshold.

[0087] The judgment unit is used to determine that the sample generation quality of the generative model is qualified when the comprehensive index value is less than the comprehensive index threshold, and to determine that the sample generation quality of the generative model is unqualified when the comprehensive index value is greater than or equal to the comprehensive index threshold.

[0088] In one possible implementation, the evaluation module 204 includes: The acquisition unit is used to acquire multiple generative models whose sample generation quality is qualified.

[0089] The selection unit is used to compare the comprehensive index values ​​of multiple generative models and select the generative model with the largest comprehensive index value as the target generative model.

[0090] The application unit is used to generate target-generated samples using the target generation model, and to train a power prediction model for low-output wind power processes by combining the target-generated samples and real samples.

[0091] The wind power low-output process sample generation quality evaluation device provided in this disclosure can execute the wind power low-output process sample generation quality evaluation method provided in any embodiment of this disclosure, and has the corresponding functional modules and beneficial effects of the execution method.

[0092] To implement the above embodiments, this disclosure also proposes a computer program product, including a computer program / instruction, which, when executed by a processor, implements the wind power low-output process sample generation quality evaluation method of the generative model in the above embodiments.

[0093] Figure 3This is a schematic diagram of the structure of a computing device provided in an embodiment of the present disclosure.

[0094] The following is a detailed reference. Figure 3 The diagram illustrates a structural schematic suitable for implementing the computing device in the embodiments of this disclosure. The computing device in the embodiments of this disclosure may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 3 The computing device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0095] like Figure 3 As shown, the computing device may include a processor (e.g., a central processing unit, a graphics processing unit, etc.) 301, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 302 or a program loaded from memory 308 into random access memory (RAM) 303. The RAM 303 also stores various programs and data required for the operation of the computing device. The processor 301, ROM 302, and RAM 303 are interconnected via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0096] Typically, the following devices can be connected to I / O interface 305: input devices 306 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 307 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; memory devices 308 including, for example, magnetic tapes, hard disks, etc.; and communication devices 309. Communication device 309 allows the computing device to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 3 A computing device with various means is shown, but it should be understood that it is not required to implement or have all the means shown. More or fewer means may be implemented or have alternatively.

[0097] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication device 309, or installed from memory 308, or installed from ROM 302. When the computer program is executed by processor 301, it performs the functions defined in the wind power low-output process sample generation quality evaluation method of the generative model of embodiments of this disclosure.

[0098] It should be noted that the computer-readable medium described in this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0099] In some implementations, clients and servers can communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future-developed networks.

[0100] The aforementioned computer-readable medium may be included in the aforementioned computing device; or it may exist independently and not assembled into the computing device.

[0101] The aforementioned computer-readable medium carries one or more programs, which, when executed by the computing device, cause the computing device to perform the aforementioned method for evaluating the quality of wind power low-output process samples generated from the generative model.

[0102] The computing device can be programmed with computer program code in one or more programming languages ​​or a combination thereof to perform the operations of this disclosure. These programming languages ​​include, but are not limited to, object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0103] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0104] The units described in the embodiments of this disclosure can be implemented in software or hardware. The names of the units are not, in some cases, intended to limit the specific unit.

[0105] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoCs), complex programmable logic devices (CPLDs), and so on.

[0106] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0107] The above description is merely a preferred embodiment of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features disclosed in this disclosure that have similar functions.

[0108] Furthermore, while the operations are described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or in a sequential order. In certain environments, multitasking and parallel processing may be advantageous. Similarly, while several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of this disclosure. Certain features described in the context of individual embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented individually or in any suitable sub-combination in multiple embodiments.

[0109] Although the subject matter has been described using language specific to structural features and / or methodological logic, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. Rather, the specific features and actions described above are merely illustrative examples of implementing the claims.

Claims

1. A method for evaluating the quality of samples generated during low-output wind power processes using generative models, characterized in that, The method includes: Obtain real samples of wind farms during low wind power output processes and generated samples of the generative model during low power output processes; wherein, the generative model is trained using the real samples; Calculate the data distribution similarity between the real sample and the generated sample to obtain the first index value; Calculate the similarity of the mapping relationship between the real sample and the generated sample to obtain the second index value; The first indicator value and the second indicator value are weighted and summed to obtain a comprehensive indicator value, and the sample generation quality evaluation result of the generative model is obtained based on the comprehensive indicator value.

2. The method for evaluating the quality of wind power low-output process samples generated according to claim 1, characterized in that, The step of calculating the data distribution similarity between the real sample and the generated sample to obtain a first index value includes: Extract multiple real feature sequences from the real samples, and extract corresponding multiple generated feature sequences from the generated samples; For each feature, the Warburg distance between the true feature sequence and the generated feature sequence is calculated to obtain the feature distance value; The first index value is obtained by weighted summation of multiple feature distance values.

3. The method for evaluating the quality of wind power low-output process samples generated according to claim 2, characterized in that, The calculation of the Warburg distance between the real feature sequence and the generated feature sequence to obtain the feature distance value includes: Calculate the empirical cumulative distribution function of the real feature sequence to obtain the first distribution function; calculate the empirical cumulative distribution function of the generated feature sequence to obtain the second distribution function; The characteristic distance value is obtained by calculating the Warburg distance between the first distribution function and the second distribution function.

4. The method for evaluating the quality of wind power low-output process sample generation according to claim 1, characterized in that, The calculation of the similarity of the mapping relationship between the real sample and the generated sample to obtain the second index value includes: Extract multiple real feature sequences from the real samples, and extract corresponding multiple generated feature sequences from the generated samples; The linear correlation between real features is calculated based on the multiple real feature sequences to obtain a first cross-correlation matrix; the linear correlation between generated features is calculated based on the multiple generated feature sequences to obtain a second cross-correlation matrix. The F-norm difference between the first cross-correlation matrix and the second cross-correlation matrix is ​​calculated to obtain the second index value.

5. The method for evaluating the quality of wind power low-output process samples generated according to claim 4, characterized in that, The step of calculating the linear correlation degree between real features based on the multiple real feature sequences to obtain the first cross-correlation matrix includes: Construct a true feature vector and a true feature mean vector based on the multiple true feature sequences; The linear correlation degree between the true features is calculated based on the true feature vector and the true feature mean vector to obtain the first cross-correlation matrix; The calculation of the linear correlation degree between the generated features based on the multiple generated feature sequences to obtain the second cross-correlation matrix includes: Based on the multiple generated feature sequences, construct a generated feature vector and a generated feature mean vector; The linear correlation degree between the generated features is calculated based on the generated feature vector and the generated feature mean vector to obtain the second cross-correlation matrix.

6. The method for evaluating the quality of wind power low-output process samples generated according to claim 1, characterized in that, The process of obtaining the sample generation quality evaluation result of the generative model based on the comprehensive index value includes: Compare the comprehensive index value with the comprehensive index threshold; When the comprehensive index value is less than the comprehensive index threshold, the sample generation quality of the generated model is deemed qualified. When the comprehensive index value is greater than or equal to the comprehensive index threshold, the sample generation quality of the generated model is determined to be unqualified.

7. The method for evaluating the quality of wind power low-output process samples generated according to claim 6, characterized in that, The method further includes: Obtain multiple generative models whose sample generation quality is acceptable; The comprehensive index values ​​of multiple generative models are compared, and the generative model with the largest comprehensive index value is selected as the target generative model. The target generation model is used to generate target generation samples, and the target generation samples and the real samples are combined to train a power prediction model for low-output wind power processes.

8. A device for evaluating the quality of wind power low-output process samples generated by a generative model, characterized in that, The device includes: The acquisition module is used to acquire real samples of low-output wind power processes and generated samples of the generation model; wherein the generation model is trained using the real samples. The first calculation module is used to calculate the data distribution similarity between the real sample and the generated sample to obtain a first index value; The second calculation module is used to calculate the similarity of the mapping relationship between the real sample and the generated sample to obtain the second index value; The evaluation module is used to obtain a comprehensive index value by weighted summation of the first index value and the second index value, and to obtain the sample generation quality evaluation result of the generative model based on the comprehensive index value.

9. A computing device, characterized in that, The computing device includes: a processor; a memory for storing executable instructions of the processor; the processor is configured to read the executable instructions from the memory and execute the instructions to implement the wind power low-output process sample generation quality evaluation method of the generative model as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute the wind power low-output process sample generation quality evaluation method of the generative model according to any one of claims 1 to 7.

Images