A device and medium for generating long-term series data of photovoltaic power generation power.

Abstract

This invention discloses a method and apparatus for generating long-term series data of photovoltaic power generation from a photovoltaic power plant. The method includes: inputting pre-prepared time-series data of power generation from a target photovoltaic power plant under different weather conditions and random time-series data of the same dimension into a pre-constructed generative adversarial network model for generating typical daily power curves of the photovoltaic power plant under different weather conditions, thereby generating typical daily power time-series data of the target photovoltaic power plant under different weather conditions; constructing a daily weather type time-series generation model based on implicit Markov and Monte Carlo simulations; and sorting the typical daily power time-series data of the target photovoltaic power plant under different weather conditions using the daily weather type time-series generation model to generate long-term series data of power generation from the target photovoltaic power plant.

Description

Technical Field

[0001] This invention relates to the field of photovoltaic power generation technology, and more specifically, to a method, apparatus, and medium for generating long-term series data of photovoltaic power generation. Background Technology

[0002] The planning and operation of new power systems with a high proportion of renewable energy sources employs production simulation technology. This treats system load, renewable energy generation, and other power output as time-series variables, comprehensively considering constraints such as power balance, power reserves, unit peak shaving, grid transmission capacity, and unit ramp-up rates. Operational simulations are performed hourly (or 15-minute) intervals to obtain the optimal power balance. Production simulation requires simulating long-term output sequences for wind and solar power. With current forecasting technology, numerical weather prediction can only predict wind and solar power output for the next 72 hours, and the error in the prediction increases with the time scale. Constructing long-term output sequences for solar power plants is a current challenge for power system operators and planners.

[0003] Photovoltaic power generation methods can be categorized into indirect modeling and direct modeling based on the initial object. Indirect modeling uses solar irradiance, which affects photovoltaic power, as its initial object. It primarily establishes an irradiance model based on astronomical and meteorological information, and then obtains the final photovoltaic power based on energy conversion relationships; this can be called physical modeling. While indirect methods simplify the complexity of the actual photovoltaic conversion process, the generated photovoltaic power deviates somewhat from the actual photovoltaic power, making it difficult to simulate the production of new power systems with a high proportion of renewable energy. With the development of artificial intelligence algorithms, researchers prefer to directly utilize historical photovoltaic power data to establish photovoltaic power models, i.e., direct modeling. Currently, direct modeling often employs statistical analysis methods such as linear autoregression and Markov chain Monte Carlo methods. However, autoregression does not significantly reflect short-term and intraday fluctuations, and Markov chains are affected in their ability to handle state transitions from multiple influencing factors simultaneously.

[0004] Therefore, there is a need for a method to generate long-term series data of photovoltaic power generation that can generate time series data that are consistent with the distribution characteristics of real data, and can support the production simulation of new power systems with a high proportion of new energy sources. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method, apparatus, and medium for generating long-term series data of photovoltaic power generation.

[0006] According to one aspect of the present invention, a method for generating long-term series data of photovoltaic power generation is provided, comprising:

[0007] The pre-prepared time series data of power generation of the target photovoltaic power station under different weather conditions and random time series data of the same dimension are input into a pre-constructed generative adversarial network model for generating typical daily power curves of the photovoltaic power station under different weather conditions, thereby generating typical daily power time series data of the target photovoltaic power station under different weather conditions.

[0008] Construct a daily weather type time series generation model based on implicit Markov and Monte Carlo simulation;

[0009] By using a daily weather type time series generation model, the typical daily power time series data of the target photovoltaic power plant under different weather conditions are sorted to generate long-term power generation data of the target photovoltaic power plant.

[0010] Optionally, the generative adversarial network model is constructed through the following steps:

[0011] Constructing the embedding and recovery functions of the autoencoder network for time series generative adversarial networks;

[0012] Constructing a sequence generator network and a sequence discriminator network for a time series generative adversarial network;

[0013] Minimize the pre-set supervised loss function and reconstruction loss function to train the embedding function and recovery function, generating an autoencoder network;

[0014] The pre-set supervised and unsupervised loss functions are jointly optimized to perform adversarial training on the sequence generator network and the sequence discriminator network, thereby generating an adversarial network.

[0015] Generative Adversarial Network (GAN) models are constructed based on autoencoder networks and adversarial networks.

[0016] Optionally, a pre-defined supervised loss function and an unsupervised loss function are jointly optimized to perform adversarial training on the sequence generator network and the sequence discriminator network, generating an adversarial network, including:

[0017] Minimize both supervised and unsupervised loss functions to optimize the model parameters in the sequence generator network;

[0018] Maximize the unsupervised loss function to optimize the model parameters in the sequence discriminator network;

[0019] An adversarial network is generated based on the optimized sequence generator network and sequence discriminator network.

[0020] Optionally, the embedding function is:

[0021] h t =e p (h t-1 ,pt )

[0022] Among them, e p It is an embedding function that sets the historical power generation p of the target photovoltaic power station. t Mapped into the latent space; h t p represents the historical power generation of the target photovoltaic power plant. t The corresponding encoding in the latent space, where the subscript t represents time information;

[0023] The recovery function is:

[0024]

[0025] Where, r p It is a recovery function that restores the latent vector h. t Restore the data to the same dimensions as the historical power generation data of the target photovoltaic power plant.

[0026] Alternatively, the formula for the sequence generator network is as follows:

[0027]

[0028] Among them, g p It is the long-time series generating function of the power generation of the target photovoltaic power plant; This represents the power generation data of the target photovoltaic power plant generated by the generator at time t. The z-axis represents the power generation data of the target photovoltaic power plant generated by the generator at time t-1; t It is a random vector with the same dimension as the historical power generation data of the target photovoltaic power plant;

[0029] The formula for the sequence discriminator network is as follows:

[0030]

[0031] Where, d p It is a discriminant function that receives the historical power generation p of the target photovoltaic power station. t The corresponding encoding h in the latent space t and power generation data at time t And classify them; This represents the discrimination result of the discriminator network.

[0032] Optionally, the reconstruction loss function is characterized by:

[0033]

[0034] Among them, L R Indicates the losses incurred during reconstruction. p represents expectation t It is the historical power generation data of the target photovoltaic power station. This is the power generation data of the target photovoltaic power station after reconstruction;

[0035] The unsupervised loss function is:

[0036]

[0037] Among them, L U For unsupervised loss, c t This indicates the discrimination result of the historical power generation data of the target photovoltaic power station. This indicates the result of the generated photovoltaic power generation count;

[0038] The supervised loss function is:

[0039]

[0040] Among them, L S For the purpose of monitoring losses, h t This represents the encoding of the historical power generation data of the target photovoltaic power plant in the potential space. The encoding in the latent space represents the power generation data of the target photovoltaic power plant generated by the generator.

[0041] Optionally, a daily weather type time series generation model based on implicit Markov and Monte Carlo simulation is constructed, including:

[0042] Based on the probability distribution of the hidden Markov state at the initial time t0, the hidden state q1 at time t1 is generated by Monte Carlo sampling.

[0043] Based on the hidden state q at time t-1 t-1 The state transition matrix A generates the hidden state q at time t. t ;

[0044] The hidden state q at time t is obtained. t At time t, the observation sequence O is generated by random sampling based on the observation probability matrix B. t ;

[0045] Determine if time t is the last time T. If not, set t = t + 1 and iterate the above steps. If it is, terminate the algorithm and output the final observation sequence O. t We obtained a time series generation model for daily weather types.

[0046] According to another aspect of the present invention, a long-term series data generation apparatus for photovoltaic power generation is provided, comprising:

[0047] The first generation module is used to input the pre-prepared time series data of the power generation of the target photovoltaic power station under different weather conditions and the random time series data of the same dimension into the pre-constructed generative adversarial network model for generating typical daily power curves of the photovoltaic power station under different weather conditions, and generate typical daily power time series data of the target photovoltaic power station under different weather conditions.

[0048] The first building module is used to build a daily weather type time series generation model based on implicit Markov and Monte Carlo simulation;

[0049] The second generation module is used to sort the typical daily power time series data of the target photovoltaic power station under different weather conditions using a daily weather type time series generation model, and generate long-term power generation data of the target photovoltaic power station.

[0050] According to another aspect of the present invention, a computer-readable storage medium is provided, the storage medium storing a computer program for performing the methods described in any of the above aspects of the present invention.

[0051] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising: a processor; a memory for storing executable instructions of the processor; the processor being configured to read the executable instructions from the memory and execute the instructions to implement the method described in any of the preceding aspects of the present invention.

[0052] Therefore, this invention provides a method for generating long-term series data of photovoltaic power generation from photovoltaic power plants. It separates weather condition generation from photovoltaic power data generation, employing implicit Markov and Monte Carlo simulations to construct a daily weather type time series generation model. A time series generative adversarial network is used to construct a typical daily power curve generation model for photovoltaic power plants under different weather conditions. Combining these two methods results in a photovoltaic power generation time series that is not only more accurate than existing model results in terms of the probabilistic statistical characteristics of the data, but also better preserves the autocorrelation of historical series. Attached Figure Description

[0053] Exemplary embodiments of the present invention can be more fully understood by referring to the following figures:

[0054] Figure 1 This is a flowchart illustrating a method for generating long-term series data of photovoltaic power generation power provided in an exemplary embodiment of the present invention.

[0055] Figure 2 This is a schematic diagram of the TimeGAN structure provided in an exemplary embodiment of the present invention;

[0056] Figure 3This is a flowchart of TimeGAN generation for time series provided in an exemplary embodiment of the present invention;

[0057] Figure 4 This is a flowchart of a daily weather type time series generation model based on implicit Markov and Monte Carlo simulation provided by an exemplary embodiment of the present invention;

[0058] Figure 5 This is a schematic diagram of the structure of a photovoltaic power plant power generation long-time series data generation device provided in an exemplary embodiment of the present invention;

[0059] Figure 6 This is the structure of an electronic device provided in an exemplary embodiment of the present invention. Detailed Implementation

[0060] Hereinafter, exemplary embodiments according to the present invention will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of the present invention, and not all embodiments of the present invention. It should be understood that the present invention is not limited to the exemplary embodiments described herein.

[0061] It should be noted that, unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps described in these embodiments do not limit the scope of the invention.

[0062] Those skilled in the art will understand that the terms "first," "second," etc., in the embodiments of the present invention are only used to distinguish different steps, devices, or modules, and do not represent any specific technical meaning, nor do they indicate a necessary logical order between them.

[0063] It should also be understood that in the embodiments of the present invention, "multiple" can refer to two or more, and "at least one" can refer to one, two or more.

[0064] It should also be understood that any component, data or structure mentioned in the embodiments of the present invention can generally be understood as one or more unless explicitly defined or given contrary instructions in the context.

[0065] Furthermore, the term "and / or" in this invention is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this invention generally indicates that the preceding and following related objects have an "or" relationship.

[0066] It should also be understood that the description of the various embodiments in this invention emphasizes the differences between the various embodiments, and the similarities or similarities can be referred to each other. For the sake of brevity, they will not be described in detail.

[0067] At the same time, it should be understood that, for ease of description, the dimensions of the various parts shown in the accompanying drawings are not drawn according to actual scale.

[0068] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the invention or its application or use.

[0069] Techniques, methods, and equipment known to those skilled in the art may not be discussed in detail, but where appropriate, they should be considered part of the specification.

[0070] It should be noted that similar labels and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be discussed further in subsequent figures.

[0071] The embodiments of this invention can be applied to electronic devices such as terminal devices, computer systems, and servers, and can operate together with a wide range of other general-purpose or special-purpose computing system environments or configurations. Well-known examples of terminal devices, computing systems, environments, and / or configurations suitable for use with electronic devices such as terminal devices, computer systems, and servers include, but are not limited to: personal computer systems, server computer systems, thin clients, thick clients, handheld or laptop devices, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputer systems, mainframe computer systems, and distributed cloud computing environments including any of the above systems, etc.

[0072] Electronic devices such as terminal devices, computer systems, and servers can be described in the general context of computer system executable instructions (such as program modules) executed by a computer system. Typically, program modules can include routines, programs, object programs, components, logic, data structures, etc., which perform specific tasks or implement specific abstract data types. Computer systems / servers can be implemented in distributed cloud computing environments, where tasks are executed by remote processing devices linked through communication networks. In distributed cloud computing environments, program modules can reside on local or remote computing system storage media, including storage devices.

[0073] Exemplary methods

[0074] Figure 1 This is a schematic flowchart illustrating a method for generating long-term series data of photovoltaic power generation from an exemplary embodiment of the present invention. This embodiment can be applied to electronic devices, such as… Figure 1 As shown, the method 100 for generating long-term series data of photovoltaic power generation includes the following steps:

[0075] Step 101: Input the pre-prepared time series data of power generation of the target photovoltaic power station under different weather conditions and the random time series data of the same dimension into the pre-constructed generative adversarial network model for generating typical daily power curves of the photovoltaic power station under different weather conditions, and generate typical daily power time series data of the target photovoltaic power station under different weather conditions.

[0076] Construct a daily weather type time series generation model based on implicit Markov and Monte Carlo simulation;

[0077] By using a daily weather type time series generation model, the typical daily power time series data of the target photovoltaic power plant under different weather conditions are sorted to generate long-term power generation data of the target photovoltaic power plant.

[0078] Specifically, the specific steps of the present invention are as follows:

[0079] S1. Construct a typical daily power curve generation model for photovoltaic power plants under different weather conditions based on a time-series generative adversarial network. Step S1 specifically includes:

[0080] The TimeGAN structure is as follows: Figure 2 As shown, the TimeGAN model comprises two parts: an autoencoder network and an adversarial network. The autoencoder network consists of embedding and recovery functions, while the adversarial network consists of a sequence generator network and a sequence discriminator network. The TimeGAN model leverages the flexibility of unsupervised learning and the control over the training process by supervised learning, incorporating temporal dynamics into the data generation process to ensure that the generated data conforms to temporal characteristics.

[0081] (1) Constructing the embedding function and the recovery function

[0082] Embedding and recovery functions provide a mapping from the feature space to the latent space, enabling adversarial networks to learn the hidden temporal dynamics within the data, i.e., the autocorrelation of time series. The embedding function is:

[0083] h t =e x (h t-1 ,x t (1)

[0084] Where e x Time series features x t Mapping to the latent space; h t This represents the latent encoding corresponding to the temporal feature, where the subscript 't' represents time information. The recovery function is:

[0085]

[0086] Where r x It is a recovery function for timing embedding; This represents the historical spatial temporal features of the latent vector generated by the recovery network.

[0087] (2) Construct sequence generator network and sequence discriminator network

[0088] Random noise is generated into data via a sequence generator network. The data is first generated into the embedding space, and the corresponding formula for the sequence generator network is:

[0089]

[0090] Where g x It is a recurrent generative network with temporal features; The temporal characteristics generated by the sequence generator; Represents the temporal characteristics of the previous time series; z t These are the temporal features of random vectors. The sequence discriminator network also operates within the embedding space, discriminating the features generated by the sequence generator in the embedding space. The corresponding formula for the sequence discriminator network is:

[0091]

[0092] Where d x Receive temporal features and classify them; This represents the discrimination result of the time feature corresponding to the sequence discriminator network; and These represent the forward and reverse hidden state sequences corresponding to the temporal features, respectively. It is a loop function.

[0093] (3) Joint training of autoencoder networks and adversarial networks

[0094] The joint training process of TimeGAN involves three loss functions: reconstruction loss, supervised loss, and unsupervised loss. Reconstruction loss is the loss function of the autoencoder network, referring to the error between the data reconstructed by the embedding and recovery functions and the historical data.

[0095]

[0096] Where x represents historical time-series data. This is the reconstructed time-series data. During the forward propagation, the generator receives the generated embedded features. And generate the next embedded feature. Then to The unsupervised loss is calculated after classification.

[0097]

[0098] Where y t The result of the discrimination of the true sequence is represented. This indicates the discrimination result of the generated sequence.

[0099] To generate more realistic data, TimeGAN introduces supervised loss for supervised learning. The principle is to generate sequence data using a portion of historical data and calculate the error between the sequence data and the historical data.

[0100]

[0101] Network training involves two joint training steps. First, the embedding and recovery functions are trained by minimizing the supervised loss and reconstruction loss, as shown in Equation (8), where λ is a hyperparameter greater than 0 used to balance the two loss functions. Then, the generator and discriminator are trained adversarially by jointly optimizing the supervised loss and unsupervised loss, as shown in Equation (9), where η is a hyperparameter greater than 0 used to balance the two loss functions.

[0102]

[0103]

[0104] Where θ e θ r θ represents the parameters in the embedded network and the recovery network, respectively. g θ d Let L represent the network parameters in the generator and discriminator, respectively. For the generator, minimizing L... S and L U Two loss functions are used to optimize the model parameters in the generator; for the discriminator, maximizing L... U The loss function is used to optimize the model parameters in the discriminator. After iteratively training each network component of TimeGAN, TimeGAN can be used to generate time series.

[0105] The process of generating data using TimeGAN is as follows: Figure 3 As shown, first, prepare real time series inputs and random time series inputs; second, create the key components of the TimeGAN model (embedding network, recovery network, generator, discriminator); then define the loss function and training steps for the training phase, execute the training loop, and store the results; finally, use the trained model to generate synthetic time series data.

[0106] S2. Construct a daily weather type time series generation model based on implicit Markov and Monte Carlo simulation. Step S2 specifically includes:

[0107] The flowchart of the daily weather type time series generation model based on implicit Markov and Monte Carlo simulation is as follows: Figure 4As shown, firstly, based on the probability distribution π of the hidden state at the initial time, the hidden state q1 at time t1 is sampled and generated; then, the hidden state q at the current time is obtained. t At time t, the observation sequence O is generated by random sampling based on the observation probability matrix B. t According to the hidden state q at time t-1 t-1 The state transition matrix A generates the hidden state q at time t. t Finally, determine if time t is the last time T. If not, set t = t + 1 and repeat the above steps. If yes, end the algorithm and output the weather type time series.

[0108] S3. Combine the daily weather type time series generation results with the typical daily power curve generation results of the photovoltaic power plant to realize the long-term time series generation of photovoltaic power plant power generation. Step S4 specifically includes:

[0109] The typical daily power curves of photovoltaic power plants generated by the TimeGAN model are arranged according to the daily weather type time series model constructed by implicit Markov and Monte Carlo simulation, so as to realize the generation of long-term power series of photovoltaic power plants.

[0110] This invention separates weather condition generation from photovoltaic power generation, employing Hidden Markov and Monte Carlo simulations to construct time-series weather data. A Time-series generative adversarial network (TimeGAN) is used to construct daily photovoltaic power output curves under different weather conditions. Combining these two methods results in a photovoltaic power plant power generation time series that is not only more accurate than existing model results in terms of the probabilistic statistical properties of the data, but also better preserves the autocorrelation of historical sequences.

[0111] Therefore, this invention provides a method for generating long-term series data of photovoltaic power generation from photovoltaic power plants. It separates weather condition generation from photovoltaic power data generation, employing implicit Markov and Monte Carlo simulations to construct a daily weather type time series generation model. A time series generative adversarial network is used to construct a typical daily power curve generation model for photovoltaic power plants under different weather conditions. Combining these two methods results in a photovoltaic power generation time series that is not only more accurate than existing model results in terms of the probabilistic statistical characteristics of the data, but also better preserves the autocorrelation of historical series.

[0112] Exemplary device

[0113] Figure 5 This is a schematic diagram of the structure of a long-term series data generation device for photovoltaic power generation of a photovoltaic power plant provided in an exemplary embodiment of the present invention. Figure 5 As shown, the device 500 includes:

[0114] The first generation module 510 is used to input the pre-prepared time series data of the power generation of the target photovoltaic power station under different weather conditions and the random time series data of the same dimension into the pre-constructed generative adversarial network model for generating typical daily power curves of the photovoltaic power station under different weather conditions, and generate typical daily power time series data of the target photovoltaic power station under different weather conditions.

[0115] The first building module 520 is used to build a daily weather type time series generation model based on implicit Markov and Monte Carlo simulation;

[0116] The second generation module 530 is used to sort the typical daily power time series data of the target photovoltaic power station under different weather conditions using a daily weather type time series generation model, and generate long-term power generation data of the target photovoltaic power station.

[0117] Optionally, the generative adversarial network model is constructed through the following steps:

[0118] The first construction submodule is used to construct the embedding and recovery functions of the autoencoder network for time series generative adversarial networks;

[0119] The second construction submodule is used to construct the sequence generator network and sequence discriminator network of the adversarial network of the time series generative adversarial network;

[0120] The first generation submodule is used to minimize the pre-set supervised loss function and reconstruction loss function to train the embedding function and recovery function, thereby generating an autoencoder network;

[0121] The second generation submodule is used to jointly optimize the pre-set supervised loss function and unsupervised loss function to perform adversarial training on the sequence generator network and the sequence discriminator network to generate an adversarial network.

[0122] The third construction submodule is used to build a generative adversarial network model based on the autoencoder network and the adversarial network.

[0123] Optionally, the second generation submodule includes:

[0124] The first optimization unit is used to minimize the supervised loss function and the unsupervised loss function in order to optimize the model parameters in the sequence generator network;

[0125] The second optimization unit is used to maximize the unsupervised loss function in order to optimize the model parameters in the sequence discriminator network;

[0126] The generation unit is used to generate an adversarial network based on the optimized sequence generator network and sequence discriminator network.

[0127] Optionally, the embedding function is:

[0128] h t =e p (h t-1 ,p t )

[0129] Among them, e p It is an embedding function that sets the historical power generation p of the target photovoltaic power station. t Mapped into the latent space; h t p represents the historical power generation of the target photovoltaic power plant. t The corresponding encoding in the latent space, where the subscript t represents time information;

[0130] The recovery function is:

[0131]

[0132] Where, r p It is a recovery function that restores the latent vector h. t Restore the data to the same dimensions as the historical power generation data of the target photovoltaic power plant.

[0133] Alternatively, the formula for the sequence generator network is as follows:

[0134]

[0135] Among them, g p It is the long-time series generating function of the power generation of the target photovoltaic power plant; This represents the power generation data of the target photovoltaic power plant generated by the generator at time t. The z-axis represents the power generation data of the target photovoltaic power plant generated by the generator at time t-1; t It is a random vector with the same dimension as the historical power generation data of the target photovoltaic power plant;

[0136] The formula for the sequence discriminator network is as follows:

[0137]

[0138] Where, d p It is a discriminant function that receives the historical power generation p of the target photovoltaic power station. t The corresponding encoding h in the latent space t and power generation data at time t And classify them; This represents the discrimination result of the discriminator network.

[0139] Optionally, the reconstruction loss function is characterized by:

[0140]

[0141] Among them, L R Indicates the losses incurred during reconstruction. p represents expectation t It is the historical power generation data of the target photovoltaic power station. This is the power generation data of the target photovoltaic power station after reconstruction;

[0142] The unsupervised loss function is:

[0143]

[0144] Among them, L U For unsupervised loss, c t This indicates the discrimination result of the historical power generation data of the target photovoltaic power station. This indicates the result of the generated photovoltaic power generation count;

[0145] The supervised loss function is:

[0146]

[0147] Among them, L S For the purpose of monitoring losses, h t This represents the encoding of the historical power generation data of the target photovoltaic power plant in the potential space. The encoding in the latent space represents the power generation data of the target photovoltaic power plant generated by the generator.

[0148] Optionally, the first building module 520 includes:

[0149] The first generation unit is used to generate the hidden state q1 at time t1 by Monte Carlo sampling based on the probability distribution of the hidden Markov hidden state at the initial time t0.

[0150] The second generation unit is used to generate the hidden state q at time t-1. t-1 The state transition matrix A generates the hidden state q at time t. t ;

[0151] The third generation unit is used to obtain the hidden state q at time t. t At time t, the observation sequence O is generated by random sampling based on the observation probability matrix B. t ;

[0152] The decision unit determines whether time t is the last time t. If not, it sets t = t + 1 and iterates the above steps. If yes, the algorithm terminates and outputs the final observation sequence O. t We obtained a time series generation model for daily weather types.

[0153] Exemplary electronic devices

[0154] Figure 6 This is the structure of an electronic device provided in an exemplary embodiment of the present invention. For example... Figure 6 As shown, the electronic device 60 includes one or more processors 61 and a memory 62.

[0155] The processor 61 may be a central processing unit (CPU) or other form of processing unit with data processing and / or instruction execution capabilities, and may control other components in the electronic device to perform desired functions.

[0156] The memory 62 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 61 may execute the program instructions to implement the methods of the software programs of the various embodiments of the present invention described above, and / or other desired functions. In one example, the electronic device may also include an input device 63 and an output device 64, these components being interconnected via a bus system and / or other forms of connection mechanisms (not shown).

[0157] In addition, the input device 63 may also include, for example, a keyboard, a mouse, etc.

[0158] The output device 64 can output various information to the outside. The output device 64 may include, for example, a display, a speaker, a printer, and a communication network and its connected remote output devices, etc.

[0159] Of course, for the sake of simplicity, Figure 6 Only some of the components of this electronic device relevant to the present invention are shown, omitting components such as buses, input / output interfaces, etc. In addition, the electronic device may include any other suitable components depending on the specific application.

[0160] Exemplary computer program products and computer-readable storage media

[0161] In addition to the methods and apparatus described above, embodiments of the present invention may also be computer program products, which include computer program instructions that, when executed by a processor, cause the processor to perform the steps in the methods according to various embodiments of the present invention described in the "Exemplary Methods" section above.

[0162] The computer program product can be written in any combination of one or more programming languages ​​to perform the operations of the embodiments of the present invention. The programming languages ​​include object-oriented programming languages ​​such as Java and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0163] Furthermore, embodiments of the present invention may also be computer-readable storage media storing computer program instructions thereon, which, when executed by a processor, cause the processor to perform the steps of the methods according to various embodiments of the present invention described in the "Exemplary Methods" section above.

[0164] The computer-readable storage medium may be any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: an electrical connection having one or more wires, a portable 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.

[0165] The basic principles of the present invention have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in the present invention are merely examples and not limitations, and should not be considered as essential features of each embodiment of the present invention. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the present invention to the necessity of employing the aforementioned specific details.

[0166] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For system embodiments, since they largely correspond to method embodiments, the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0167] The block diagrams of devices, systems, devices, and systems involved in this invention are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, systems, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0168] The methods and systems of the present invention may be implemented in many ways. For example, they may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above-described order of steps for the methods is for illustrative purposes only, and the steps of the methods of the present invention are not limited to the order specifically described above unless otherwise specifically stated. Furthermore, in some embodiments, the present invention may also be implemented as a program recorded on a recording medium, the program comprising machine-readable instructions for implementing the methods according to the present invention. Thus, the present invention also covers recording media storing programs for performing the methods according to the present invention.

[0169] It should also be noted that in the systems, apparatus, and methods of the present invention, the components or steps can be disassembled and / or recombined. These disassemblies and / or recombinations should be considered equivalents of the present invention. The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the invention. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of the invention. Therefore, the invention is not intended to be limited to the aspects shown herein, but rather to be carried out within the widest scope consistent with the principles and novel features disclosed herein.

[0170] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of the invention to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations therein.

Claims

1. A method for generating long-term series data of photovoltaic power generation, characterized in that, include: The pre-prepared time series data of power generation of the target photovoltaic power station under different weather conditions and random time series data of the same dimension are input into a pre-constructed generative adversarial network model for generating typical daily power curves of the photovoltaic power station under different weather conditions, thereby generating typical daily power time series data of the target photovoltaic power station under different weather conditions. Construct a daily weather type time series generation model based on implicit Markov and Monte Carlo simulation; The daily weather type time series generation model is used to sort the typical daily power time series data of the target photovoltaic power station under different weather conditions, and generate long-term power generation data of the target photovoltaic power station. The generative adversarial network model is constructed through the following steps: Constructing the embedding and recovery functions of the autoencoder network for time series generative adversarial networks; Construct the sequence generator network and sequence discriminator network of the adversarial network of the time series generative adversarial network; The pre-set supervised loss function and reconstruction loss function are minimized to train the embedding function and the recovery function, thereby generating the autoencoder network; The pre-set supervised loss function and unsupervised loss function are jointly optimized to perform adversarial training on the sequence generator network and the sequence discriminator network, thereby generating the adversarial network; Based on the autoencoder network and the adversarial network, the generative adversarial network model is constructed; The embedding function is: in, It is an embedded function that sets the historical power generation of the target photovoltaic power station. Mapped into the latent space; This indicates the historical power generation of the target photovoltaic power station. The corresponding encoding in the latent space, index t Represents time information; The recovery function is: in, It is a recovery function that restores the latent vector. Restore the data to the same dimensions as the historical power generation data of the target photovoltaic power plant. ; The formula for the sequence generator network is as follows: in, It is the long-time series generation function of the power generation of the target photovoltaic power station; The target photovoltaic power station generated by the generator is in t Power generation data at any given time; The target photovoltaic power station generated by the generator is in t Power generation data at time -1; It is a random vector with the same dimension as the historical power generation data of the target photovoltaic power station; The formula for the sequence discriminator network is as follows: in, It is a discriminant function that receives the historical power generation of the target photovoltaic power station. Encoding in the corresponding latent space and in t Power generation data at any time And classify them; This represents the discrimination result of the discriminator network; Construct a daily weather type time series generation model based on implicit Markov and Monte Carlo simulation, including: according to t The probability distribution of the hidden states of the Hidden Markov Model at initial time 0 is generated using Monte Carlo sampling. t Hidden state at time 1 q 1; according to t Hidden state at time -1 q t-1 And the state transition matrix A is generated t Hidden state at time q t ; In obtaining t Hidden state at time q t At that time, random sampling is performed based on the observation probability matrix B. t Observation sequence at time O t ; judge t Is the time T the last time step? If not, then let... t = t +1, iterate through the above steps; if yes, end the algorithm and output the final observation sequence. O t The time series generation model for the daily weather type is obtained.

2. The method according to claim 1, characterized in that, Jointly optimizing the pre-set supervised loss function and unsupervised loss function to perform adversarial training on the sequence generator network and the sequence discriminator network, generating the adversarial network, including: Minimize the supervised loss function and the unsupervised loss function to optimize the model parameters in the sequence generator network; Maximize the unsupervised loss function to optimize the model parameters in the sequence discriminator network; The adversarial network is generated based on the optimized sequence generator network and the sequence discriminator network.

3. The method according to claim 1, characterized in that, The reconstruction loss function is: in, Indicates the losses incurred during reconstruction. Expressing expectations, These are the historical power generation data of the target photovoltaic power station. This refers to the power generation data of the reconstructed target photovoltaic power station; The unsupervised loss function is: in, L U Loss due to lack of supervision This indicates the discrimination result of the historical power generation data of the target photovoltaic power station. This indicates the result of the generated photovoltaic power generation count; The supervised loss function is: in, L S For the purpose of supervision loss, This represents the encoding of the historical power generation data of the target photovoltaic power plant in the potential space. The encoding in the latent space represents the power generation data of the target photovoltaic power station generated by the generator.

4. A device for generating long-term series data of photovoltaic power generation, used to implement the method of claim 1, characterized in that, include: The first generation module is used to input the pre-prepared time series data of the power generation of the target photovoltaic power station under different weather conditions and the random time series data of the same dimension into a pre-constructed generative adversarial network model for generating typical daily power curves of the photovoltaic power station under different weather conditions, so as to generate typical daily power time series data of the target photovoltaic power station under different weather conditions. The first building module is used to build a daily weather type time series generation model based on implicit Markov and Monte Carlo simulation; The second generation module is used to sort the typical daily power time series data of the target photovoltaic power station under different weather conditions using the daily weather type time series generation model, and generate long-term power generation data of the target photovoltaic power station.

5. The apparatus according to claim 4, characterized in that, Jointly optimizing the pre-set supervised loss function and unsupervised loss function to perform adversarial training on the sequence generator network and the sequence discriminator network, generating the adversarial network, including: Minimize the supervised loss function and the unsupervised loss function to optimize the model parameters in the sequence generator network; Maximize the unsupervised loss function to optimize the model parameters in the sequence discriminator network; The adversarial network is generated based on the optimized sequence generator network and the sequence discriminator network.

6. The apparatus according to claim 4, characterized in that, The reconstruction loss function is: in, Indicates the losses incurred during reconstruction. Expressing expectations, These are the historical power generation data of the target photovoltaic power station. This refers to the power generation data of the reconstructed target photovoltaic power station; The unsupervised loss function is: in, L U Loss due to lack of supervision This indicates the discrimination result of the historical power generation data of the target photovoltaic power station. This indicates the result of the generated photovoltaic power generation count; The supervised loss function is: in, L S For the purpose of supervision loss, This represents the encoding of the historical power generation data of the target photovoltaic power plant in the potential space. The encoding in the latent space represents the power generation data of the target photovoltaic power station generated by the generator.

7. A computer-readable storage medium, characterized in that, The storage medium stores a computer program for performing the method described in any one of claims 1-3.

8. An electronic device, characterized in that, The electronic device includes: processor; Memory used to store the processor's executable instructions; The processor is configured to read the executable instructions from the memory and execute the instructions to implement the method described in any one of claims 1-3.

Images