Method and system for generating renewable energy system operating scenarios

Abstract

The application discloses a renewable energy system operation scene generation method, comprising the following steps: obtaining historical data information of a target renewable energy system; establishing a diffusion process model for the operation scene of the target renewable energy system; carrying out denoising processing based on Gaussian noise; modeling a conditional diffusion model; constructing a neural network for the conditional diffusion model; using a large model to guide the generation of conditional information, and training the established conditional diffusion model to obtain a trained conditional diffusion model; and using the trained conditional diffusion model to generate the operation scene of the renewable energy system. The application also discloses a system for implementing the renewable energy system operation scene generation method. Through the large language model as a condition guide, the diffusion model can directly generate a deterministic operation scene from any Gaussian noise, so that the application can not only generate the operation scene of the renewable energy system, but also has high reliability, good accuracy and high efficiency.

Description

Methods and systems for generating operation scenarios of renewable energy systems Technical Field

[0001] This invention belongs to the field of electrical automation, and specifically relates to a method and system for generating operating scenarios for renewable energy systems. Background Technology

[0002] With economic and technological development and the improvement of people's living standards, electricity has become an indispensable secondary energy source in people's production and daily life, bringing endless convenience. Therefore, ensuring a stable and reliable supply of electricity has become one of the most important tasks of the power system.

[0003] Currently, an increasing number of renewable energy power generation systems, such as wind and solar power, are being integrated into the power grid and generating electricity. However, the volatility, intermittency, and randomness of renewable energy power generation systems bring significant uncertainty to the short-term operation of the power system. Therefore, accurate and reliable quantitative modeling of the uncertainties in power systems with large-scale renewable energy grid integration, and conducting research based on the modeling data, is of great significance to the power system.

[0004] Currently, common schemes for generating operating scenarios for renewable energy systems include sampling-based and optimization-based methods. Sampling-based methods are the most common, directly sampling the probability distribution to generate a discrete set of scenarios. However, sampling-based methods may suffer from a critical problem: the probability distribution of the renewable random variables must be known or assumed to approximately follow a certain distribution; this problem significantly affects the reliability of the final generated scenarios. Optimization-based methods extract scenarios from a given large set of scenarios through scenario reduction techniques; however, these schemes have low computational efficiency and involve serious high-dimensional solution problems. Summary of the Invention

[0005] One of the objectives of this invention is to provide a method for generating operating scenarios of renewable energy systems that is highly reliable, accurate, and efficient.

[0006] The second objective of this invention is to provide a system for implementing the method for generating operating scenarios of the renewable energy system.

[0007] The method for generating operating scenarios for renewable energy systems provided by this invention includes the following steps:

[0008] S1. Obtain historical data information of the target renewable energy system;

[0009] S2. Based on the data obtained in step S1, establish a diffusion process model for the operation scenario of the target renewable energy system;

[0010] S3. Based on the data obtained in step S1 and the diffusion process model established in step S2, perform noise reduction processing based on Gaussian noise;

[0011] S4. Model the conditional diffusion model based on the data obtained in step S3;

[0012] S5. Construct a neural network for the conditional diffusion model;

[0013] S6. Using a large model as a guide, conditional information is generated, and the established conditional diffusion model is trained to obtain the trained conditional diffusion model;

[0014] S7. Using the trained conditional diffusion model obtained in step S6, generate the operating scenario of the renewable energy system.

[0015] Step S2, which involves establishing a diffusion process model for the target renewable energy system's operating scenario based on the data obtained in step S1, specifically includes the following steps:

[0016] Let the operating scenario of the renewable energy system be x; Gaussian noise is added to operating scenario x through a diffusion process, denoted as...

[0017]

[0018] In the formula q(x 1:T |x0) represents the diffusion process; x 1:T x0 represents the sample after 1…T steps of noise diffusion; T represents the original sample sampled from the true data distribution; q(x0) represents the total number of diffusion steps; and T represents the total number of diffusion steps. t |x t-1 ) represents the t-th diffusion step; x t For samples that have undergone t-step diffusion;

[0019] For the t-th diffusion step q(x) t |x t-1 ), which accepts x t-1 And return x by adding some noise. t , represented as

[0020]

[0021] In the formula, N() is the standard Gaussian noise function; β t For noise programming tables, and β t ∈(0,1); I is the identity matrix;

[0022] After a diffusion process of T steps, x0 will be transformed into a sample x that has undergone T steps of diffusion. TIt follows an isotropic Gaussian distribution.

[0023] Step S3, which involves denoising based on Gaussian noise using the data obtained in step S1 and the diffusion process model established in step S2, specifically includes the following steps:

[0024] The denoising process is used for x T Remove noise to recover x0, so that x0 can create new data samples from the Gaussian noise distribution;

[0025] The denoising process is represented as

[0026] p θ (x t-1 |x t )=N(x t-1 μ θ (x t ,t),β t I)

[0027] In the formula p θ (x t-1 |x t ) represents the denoising process implemented through a neural network; μ θ (x t ,t) represents the neural network processing procedure with parameter θ;

[0028] Based on the denoising steps, the denoising process can be expressed as the following formula:

[0029]

[0030] In the formula p θ (x 0:T p(x) represents the denoising process; T ) is Gaussian noise, and p(x) T ) = N(0, I);

[0031] If we define the denoising process as the inverse of the diffusion process, then the objective function of the diffusion model is expressed as:

[0032] min(KL(p θ (x t-1 |x t ),q(x t-1 |x t ,x0)))

[0033] In the formula, KL() is the Kullback-Leibler divergence calculation function; q(x t-1 |x t x0) represents the posterior conditional probability of the diffusion process.

[0034] Step S4 involves modeling the conditional diffusion model on the data obtained in step S3, specifically including the following steps:

[0035] To achieve condition-guided deterministic generation, the diffusion model is extended to a conditional diffusion model, thereby learning conditional information;

[0036] The expression p θ (x t-1 |x t )=N(x t-1 μ θ (x t ,t),β t I) Expanded to

[0037] p θ (x t-1 |x t ,x c )=N(x t-1 μ θ (x t ,t|x c ),β t I)

[0038] In the formula p θ (x t-1 |x t ,x c ) for x t Change to x t-1 The noise reduction process; x c For conditional information;

[0039] Will Expand to

[0040]

[0041] In the formula p θ (x t-1 |x t ,x c ) for x t Change to x t-1 The noise reduction process;

[0042] Using the method of undetermined coefficients to Solve the following:

[0043] set up In the formula p(x t-1 |x t x0) represents the denoising process, κ t λ is the first coefficient. t The second coefficient, σ t It is the third coefficient;

[0044] Set σ t As a variable, and Where α t β is the first intermediate variable. t For noise planning table, As the second intermediate variable, It is the third intermediate variable;

[0045] Thus, the following formula is obtained.

[0046]

[0047] In the formula, γ is the fourth intermediate variable, and

[0048] If set And η∈[0,1], where η is the weight coefficient, then we can obtain that when η=0, the variance of the Gaussian distribution is 0, and the generation is deterministic.

[0049] At the same time, the neural network μ θ (x t ,t) parameterized as

[0050]

[0051] In the formula ε θ This is a diffusion model with parameter θ.

[0052] Step S5, which describes constructing the neural network for the conditional diffusion model, specifically includes the following steps:

[0053] The constructed neural network includes an embedding module, a residual module, and an output module;

[0054] The embedding module is used to project input features into a high-dimensional space; the embedding module includes a linear layer; the input is set to x. t The condition is x c The processing procedure of the embedded module is then represented as follows:

[0055] x emb =FC(x) t )+FC(x c )

[0056] In the formula x emb The output of the embedded module is FC(); FC() is the fully connected layer processing function.

[0057] The residual module connects the input to the module output, mitigating the vanishing gradient problem. The residual module comprises four residual sub-modules, each including a multi-head attention layer and a fully connected layer. The processing procedure of the i-th residual sub-module is represented as follows:

[0058] x res,i =FC(MHA(x) res,i-1 ))+x res,i-1

[0059] In the formula x res,i The output of the i-th residual submodule; MHA() is the processing function of the multi-head attention layer; if i is 1, then x res,i-1 For the output of the embedded module;

[0060] The output module is used to connect the outputs of all residual modules and map them to obtain the final output result; the output module includes a fully connected layer; the processing procedure of the output module is represented as x. out =FC(x) res,1 ,x res,2 ,x res,3 ,x res,4 ), x out This is the output of the output module.

[0061] Step S6, which involves using a large model to guide the generation of conditional information and training the established conditional diffusion model to obtain the trained conditional diffusion model, specifically includes the following steps:

[0062] For the diffusion generation task, the following prompt words are designed: The prompt word format is set as <S,P,D,C>, where S represents the season and the value range is {spring, summer, autumn, winter}, P represents the energy output intensity and the value range is {high, medium, low}, D represents daytime or nighttime, and C represents wind power or solar power.

[0063] The open-source Qwen2-7B model was adopted as the large model;

[0064] The first round of text dialogue with the large model: The large model uses semantic understanding to confirm whether the input meets the format <S,P,D,C>. If it does, it constructs conditional information to guide the training of the diffusion model. If it does not meet the format, it enters the second round of dialogue.

[0065] The second round of text dialogue with the large model: Based on the missing conditions reported by the large model, the text is completed, and text dialogue with the large model is conducted again;

[0066] The guidance process is as follows: A vocabulary is constructed to map text words; the large model performs formatted extraction on the input text, constructs <S,P,D,C> and maps it to the vocabulary as conditional information x. c Diffusion of the original sample x0 yields x T and x0, x c and x T They are input together into the conditional diffusion model obtained in step S4 for training;

[0067] Finally, the trained conditional diffusion model is obtained.

[0068] Step S7, which uses the trained conditional diffusion model obtained in step S6 to generate operating scenarios for renewable energy systems, specifically includes the following steps:

[0069] Randomly sample a Gaussian noise x from any standard Gaussian distribution. T,new The input is fed into the trained conditional diffusion model obtained in step S6 to obtain the x generated by the model. c And according to x c Complete the generation of renewable energy system operation scenarios.

[0070] This invention also provides a system for implementing the method for generating operating scenarios of the renewable energy system, comprising a data acquisition module, a diffusion modeling module, a Gaussian denoising module, a model modeling module, a network construction module, a model guidance module, and a scenario generation module; the data acquisition module, diffusion modeling module, Gaussian denoising module, model modeling module, network construction module, model guidance module, and scenario generation module are connected in series; the data acquisition module is used to acquire historical data information of the target renewable energy system and upload the data information to the diffusion modeling module; the diffusion modeling module is used to establish a diffusion process model for the operating scenario of the target renewable energy system based on the received data information and the acquired data information, and upload the data information to the Gaussian denoising module; the Gaussian denoising module is used to establish a diffusion process model for the operating scenario of the target renewable energy system based on the received data information and the acquired data information. The system first establishes a diffusion process model, denoises it using Gaussian noise, and uploads the data to the model building module. The model building module then models the conditional diffusion model based on the received data and uploads the data to the network construction module. The network construction module builds a neural network for the conditional diffusion model based on the received data and uploads the data to the model guidance module. The model guidance module generates conditional information using a large model guide based on the received data, trains the established conditional diffusion model, and uploads the trained conditional diffusion model to the scenario generation module. Finally, the scenario generation module generates a renewable energy system operation scenario based on the received data and the trained conditional diffusion model.

[0071] The renewable energy system operation scenario generation method and system provided by this invention can generate deterministic operation scenarios directly from arbitrary Gaussian noise by using a large language model as a condition guide and a diffusion model. Therefore, this invention can not only generate renewable energy system operation scenarios, but also has higher reliability, better accuracy and higher efficiency. Attached Figure Description

[0072] Figure 1 is a schematic diagram of the method flow of the present invention.

[0073] Figure 2 is a schematic diagram of the generation scenario of an embodiment of the method of the present invention.

[0074] Figure 3 is a schematic diagram of the functional modules of the system of the present invention. Detailed Implementation

[0075] Figure 1 shows a schematic diagram of the method flow of the present invention: The renewable energy system operation scenario generation method disclosed in this invention includes the following steps:

[0076] S1. Obtain historical data information of the target renewable energy system;

[0077] The purpose of this step is to mathematically model the diffusion process in order to transform the original sample into a noise sample;

[0078] S2. Based on the data obtained in step S1, establish a diffusion process model for the operation scenario of the target renewable energy system; specifically, this includes the following steps:

[0079] Let the operating scenario of the renewable energy system be x; Gaussian noise is added to operating scenario x through a diffusion process, denoted as...

[0080]

[0081] In the formula q(x 1:T |x0) represents the diffusion process; x 1:T x0 represents the sample after 1…T steps of noise diffusion; T represents the original sample sampled from the true data distribution; q(x0) represents the total number of diffusion steps; and T represents the total number of diffusion steps. t |x t-1 ) represents the t-th diffusion step; x t For samples that have undergone t-step diffusion;

[0082] In practical implementation, the operating scenario x of the renewable energy system includes the power values ​​of wind power generation and photovoltaic power generation during actual operation, and each operating scenario is divided into 24-hour periods; Gaussian noise represents a set of random data that conforms to a normal distribution;

[0083] For the t-th diffusion step q(x) t |x t-1 ), which accepts x t-1 And return x by adding some noise. t , represented as

[0084]

[0085] In the formula, N() is the standard Gaussian noise function; βt For noise programming tables, and β t ∈(0,1); I is the identity matrix;

[0086] After a T-step diffusion process, x0 will be transformed into a sample x that has undergone T-step noise diffusion. T And it follows an isotropic Gaussian distribution;

[0087] The purpose of this step is to mathematically model the denoising process in order to guide the training of the model;

[0088] S3. Based on the data obtained in step S1 and the diffusion process model established in step S2, perform noise reduction processing based on Gaussian noise; specifically including the following steps:

[0089] The denoising process is used for x T Remove noise to recover x0, so that x0 can create new data samples from the Gaussian noise distribution;

[0090] The denoising process is represented as

[0091] p θ (x t-1 |x t )=N(x t-1 μ θ (x t ,t),β t I)

[0092] In the formula p θ (x t-1 |x t ) represents the denoising process implemented through a neural network; μ θ (x t ,t) represents the neural network processing procedure with parameter θ;

[0093] Based on the denoising steps, the denoising process can be expressed as the following formula:

[0094]

[0095] In the formula p θ (x 0:T p(x) represents the denoising process; T ) is Gaussian noise, and p(x) T ) = N(0, I);

[0096] If we define the denoising process as the inverse of the diffusion process, then the objective function of the diffusion model is expressed as:

[0097] min(KL(p θ (x t-1 |x t ),q(xt-1 |x t ,x0)))

[0098] In the formula, KL() is the Kullback-Leibler divergence calculation function; q(x t-1 |x t x0) represents the posterior conditional probability of the diffusion process;

[0099] The purpose of this step is to mathematically model the conditional diffusion model in order to guide the generation of the model for renewable energy operation scenarios;

[0100] S4. Model the conditional diffusion model using the data obtained in step S3; specifically, this includes the following steps:

[0101] To achieve condition-guided deterministic generation, the diffusion model is extended to a conditional diffusion model, thereby learning conditional information;

[0102] The expression p θ (x t-1 |x t )=N(x t-1 μ θ (x t ,t),β t I) Expanded to

[0103] p θ (x t-1 |x t ,x c )=N(x t-1 μ θ (x t ,t|x c ),β t I)

[0104] In the formula p θ (x t-1 |x t ,x c ) for x t Change to x t-1 The noise reduction process; x c For conditional information;

[0105] Will Expand to

[0106]

[0107] In the formula p θ (x t-1 |x t ,x c ) for x t Change to xt-1 The noise reduction process;

[0108] Using the method of undetermined coefficients to Solve the following:

[0109] set up In the formula p(x t-1 |x t x0) represents the denoising process, κ t λ is the first coefficient. t The second coefficient, σ t It is the third coefficient;

[0110] Set σ t As a variable, and Where α t β is the first intermediate variable. t For noise planning table, As the second intermediate variable, It is the third intermediate variable;

[0111] Thus, the following formula is obtained.

[0112]

[0113] In the formula, γ is the fourth intermediate variable, and

[0114] If set And η∈[0,1], where η is the weight coefficient, then we can obtain that when η=0, the variance of the Gaussian distribution is 0, and the generation is deterministic.

[0115] At the same time, the neural network μ θ (x t ,t) parameterized as

[0116]

[0117] In the formula ε θ For a diffusion model with parameter θ;

[0118] The purpose of this step is to construct the neural network structure of the conditional diffusion model for model training;

[0119] S5. Construct a neural network for the conditional diffusion model; specifically including the following steps:

[0120] The constructed neural network includes an embedding module, a residual module, and an output module;

[0121] The embedding module is used to project input features into a high-dimensional space; the embedding module includes a linear layer; the input is set to x. tThe condition is x c The processing procedure of the embedded module is then represented as follows:

[0122] x emb =FC(x) t )+FC(x c )

[0123] In the formula x emb The output of the embedded module is FC(); FC() is the fully connected layer processing function.

[0124] The residual module connects the input to the module output, mitigating the vanishing gradient problem. The residual module comprises four residual sub-modules, each including a multi-head attention layer and a fully connected layer. The processing procedure of the i-th residual sub-module is represented as follows:

[0125] x res,i =FC(MHA(x) res,i-1 ))+x res,i-1

[0126] In the formula x res,i The output of the i-th residual submodule; MHA() is the processing function of the multi-head attention layer; if i is 1, then x res,i-1 For the output of the embedded module;

[0127] The output module is used to connect the outputs of all residual modules and map them to obtain the final output result; the output module includes a fully connected layer; the processing procedure of the output module is represented as x. out =FC(x) res,1 ,x res,2 ,x res,3 ,x res,4 ), x out This is the output of the output module;

[0128] The purpose of this step is to use a large model to identify the conditional parameters of natural language text and parse them into formatted data as conditional information to guide the training of the diffusion model.

[0129] S6. Using a large model as a guide, conditional information is generated, and the established conditional diffusion model is trained to obtain the trained conditional diffusion model; specifically, the following steps are included:

[0130] For the diffusion generation task, the following prompt words are designed: The prompt word format is set as <S,P,D,C>, where S represents the season and the value range is {spring, summer, autumn, winter}, P represents the energy output intensity and the value range is {high, medium, low}, D represents daytime or nighttime, and C represents wind power or solar power.

[0131] The open-source Qwen2-7B model was adopted as the large model;

[0132] The first round of text dialogue with the large model: The large model uses semantic understanding to confirm whether the input meets the format <S,P,D,C>. If it does, it constructs conditional information to guide the training of the diffusion model. If it does not meet the format, it enters the second round of dialogue.

[0133] The second round of text dialogue with the large model: Based on the missing conditions reported by the large model, the text is completed, and text dialogue with the large model is conducted again;

[0134] The guidance process is as follows: A vocabulary is constructed to map text words; the large model performs formatted extraction on the input text, constructs <S,P,D,C> and maps it to the vocabulary as conditional information x. c Diffusion of the original sample x0 yields x T and x0, x c and x T They are input together into the conditional diffusion model obtained in step S4 for training;

[0135] Finally, the trained conditional diffusion model is obtained;

[0136] The purpose of this step is to use the trained diffusion model to generate deterministic renewable energy operation scenarios;

[0137] S7. Using the trained conditional diffusion model obtained in step S6, generate operating scenarios for renewable energy systems; specifically including the following steps:

[0138] Randomly sample a Gaussian noise x from any standard Gaussian distribution. T,new The x T,new Corresponding to the noisy sample x obtained after T-step diffusion during the training process. T It should be noted that the generation process does not use x. T Instead of using information, new x is obtained through sampling. T,new ;

[0139] Subsequently, the user actively interacts with the large model, inputting a specific scenario to be generated, such as "the output value of a wind turbine on a windy summer night." The large model then extracts the <S,P,D,C> construction condition information x. c and with x T,new Together they are fed into the diffusion model trained in S6 to generate the operating scenario of the renewable energy system.

[0140] The method of the present invention will be further described below with reference to an embodiment:

[0141] 1. Introduce operating scenarios for renewable energy systems, including the power values ​​of wind power and photovoltaic power during actual operation. Each operating scenario is divided into 24-hour periods to construct a dataset of operating scenarios for renewable energy systems, where each sample is represented by x0.

[0142] 2. Establish a diffusion process. For each sample x0, step S2 is used to diffuse x0 to the noisy sample x. T The transformation of x makes x T It follows an isotropic Gaussian distribution. Here, T = 1000, a value that reduces the variance introduced by the diffusion process.

[0143] 3. Based on step S3, a denoising process was established, and the objective function used for model training was set.

[0144] 4. Based on step S4, perform mathematical modeling on the used conditional diffusion model to guide the generation of renewable energy operation scenarios. Wherein, the conditional information x... c This is expressed in natural language as information such as "season, day / night cycle, intensity, and energy type". σ in step S4... t The value is σ t =0.

[0145] 5. Based on step S5, construct the neural network structure of the diffusion model;

[0146] 6. Construct some natural language descriptive texts and have them interact with the large model. The text represents specific scenarios to be generated, such as "the output value of a wind turbine on a windy night in summer" or "the output value of a photovoltaic power generation on a sunny day in winter." The large model extracts <S, P, D, C> to construct conditional information x. c And compared with the noise-enhanced renewable energy scenario sample x T They are fed into the constructed diffusion model neural network and trained using the objective function set in step S3.

[0147] 7. After the diffusion model is trained, it learns the data distribution of real-world renewable energy operation scenarios. Users can communicate their generation needs with the large model, such as "Please generate the daily output of a generator in a light breeze during spring." The large model identifies user needs and extracts <S, P, D, C> to construct conditional information x. c Then, the diffusion model is invoked to randomly sample a Gaussian noise x from an arbitrary standard Gaussian distribution. T,new The corresponding scene is generated based on the denoising process established in S3. A scene diagram is shown in Figure 2; where Real represents the real scene and Reconstructed represents the generated scene. The vertical axis Power represents the power value, and the horizontal axis Time represents time.

[0148] As can be seen from Figure 2, the generated scene is very similar to the real scene, indicating that the diffusion model used can generate reliable scene samples.

[0149] Figure 3 shows a schematic diagram of the functional modules of the system of the present invention: The system for generating the operation scenario of the renewable energy system disclosed in this invention includes a data acquisition module, a diffusion modeling module, a Gaussian denoising module, a model modeling module, a network construction module, a model guidance module, and a scenario generation module; the data acquisition module, diffusion modeling module, Gaussian denoising module, model modeling module, network construction module, model guidance module, and scenario generation module are connected in series; the data acquisition module is used to acquire historical data information of the target renewable energy system and upload the data information to the diffusion modeling module; the diffusion modeling module is used to establish a diffusion process model for the operation scenario of the target renewable energy system based on the received data information and the acquired data information, and upload the data information to the Gaussian denoising module; the Gaussian denoising module is used to establish a diffusion process model for the operation scenario of the target renewable energy system based on the received data information and the acquired data information, and upload the data information to the Gaussian denoising module; the Gaussian denoising module is used to establish a diffusion process model for the operation scenario of the target renewable energy system based on the received data information and the acquired data information. The system first receives data and establishes a diffusion process model. Based on this data, it performs denoising based on Gaussian noise and uploads the data to the model building module. The model building module then models the conditional diffusion model using the received data and uploads the data to the network construction module. The network construction module builds a neural network for the conditional diffusion model based on the received data and uploads the data to the model guidance module. The model guidance module generates conditional information using a large model guide and trains the established conditional diffusion model to obtain the trained conditional diffusion model, which is then uploaded to the scenario generation module. Finally, the scenario generation module generates a renewable energy system operation scenario based on the received data and the trained conditional diffusion model.

Claims

1. A method for generating an operating scenario for a renewable energy system, comprising the following steps: S1. Obtaining historical data information of the target renewable energy system; S2. Establishing a diffusion process model for the operating scenario of the target renewable energy system based on the data information obtained in step S1; S3. Performing denoising processing based on Gaussian noise based on the data information obtained in step S1 and the diffusion process model established in step S2; specifically including the following steps: the denoising process is used to... Remove noise to restore Thus making It can create new data samples from a Gaussian noise distribution; The denoising process is represented as In the formula This represents the denoising process implemented using a neural network. The parameter is The neural network processing procedure; based on the denoising step, the denoising process is expressed as the following formula: In the formula This is the noise reduction process; It is Gaussian noise, and If we define the denoising process as the inverse of the diffusion process, then the objective function of the diffusion model is expressed as: In the formula The function for calculating the Kullback-Leibler divergence; Let be the posterior conditional probability of the diffusion process; S4. Model the conditional diffusion model using the data obtained in step S3; specifically, this includes the following steps: to achieve condition-guided deterministic generation, extend the diffusion model to a conditional diffusion model, thereby learning conditional information; [The text then abruptly shifts to a different topic:] ...the formula... Expand to In the formula To be Become The noise reduction process; For conditional information; Expand to In the formula To be Become The noise reduction process; Using the method of undetermined coefficients to To solve: Set In the formula For the noise reduction process, As the first coefficient, As the second coefficient, It is the third coefficient; set up As a variable, and ， ， ,in As the first intermediate variable, For noise planning table, As the second intermediate variable, As the third intermediate variable, the following formula is obtained. In the formula It is the fourth intermediate variable, and If set and ， If the weighting coefficient is used, then we can obtain the result when... When the variance of the Gaussian distribution is 0, the generation becomes deterministic; simultaneously, the neural network... Parameterization In the formula For parameters The diffusion model; S5. Construct a neural network for the conditional diffusion model; specifically including the following steps: the constructed neural network includes an embedding module, a residual module, and an output module; the embedding module is used to project the input features into a high-dimensional space; the embedding module includes a linear layer; the input is set as... The conditions are The processing procedure of the embedded module is then represented as follows: In the formula For the output of the embedded module; Here is the processing function for the fully connected layer; the residual module connects the input and the module output to alleviate the vanishing gradient problem; the residual module consists of 4 residual sub-modules, each of which includes a multi-head attention layer and a fully connected layer; the processing procedure of the i-th residual sub-module is expressed as follows: In the formula This is the output of the i-th residual submodule; Let i be the processing function of the multi-head attention layer; if i is 1, then For the output of the embedded module; The output module is used to connect the outputs of all residual modules and map them to obtain the final output result; the output module includes a fully connected layer; the processing procedure of the output module is represented as follows: ， This is the output of the output module; S6. Using a large model as a guide, conditional information is generated, and the established conditional diffusion model is trained to obtain the trained conditional diffusion model; specifically, the following steps are included: For the diffusion generation task, prompt words are designed: the prompt word format is set as follows... ,in Indicates the season and the range of values ​​is ， Indicates energy output intensity and its value range is ， Indicates day or night. This represents wind power or solar power; the Qwen2-7B model is used as the main model; the first round of text-based dialogue with the main model: the main model uses semantic understanding to confirm whether the input meets the requirements. The format is as follows: If the conditions are met, construct conditional information to guide the training of the diffusion model; otherwise, proceed to the second round of dialogue. The second round involves textual dialogue with the large model: based on the missing conditions reported by the large model, complete the text and engage in textual dialogue with the large model again. The guidance process is as follows: construct a vocabulary list to map text words; the large model performs formatted extraction of the input text and constructs... And mapped to the vocabulary as conditional information. ; the original sample Diffusion and will 、 and They are input together into the conditional diffusion model obtained in step S4 for training; The samples undergo T-step diffusion; finally, the trained conditional diffusion model is obtained. S7. Using the trained conditional diffusion model obtained in step S6, generate the operating scenario of the renewable energy system.

2. The method for generating operating scenarios of renewable energy systems according to claim 1, characterized in that... Step S2, which involves establishing a diffusion process model for the target renewable energy system's operating scenario based on the data obtained in step S1, specifically includes the following steps: setting the operating scenario of the renewable energy system as... ； Gaussian noise is added to the operating environment through a diffusion process. In, it is represented as In the formula Indicates the diffusion process; For samples that have undergone 1…T steps of noise diffusion; The original samples were drawn from the real data distribution; This represents the total number of diffusion steps; This represents the t-th diffusion step; For a sample that has undergone t diffusion steps; for the t-th diffusion step Its acceptance And return after adding some noise. , represented as In the formula The standard Gaussian noise function; For noise planning table, and ； It is the identity matrix; after a T-step diffusion process, This will transform into a sample that has undergone T-step diffusion. It follows an isotropic Gaussian distribution.

3. The method for generating operating scenarios of renewable energy systems according to claim 2, characterized in that... Step S7, which uses the trained conditional diffusion model obtained in step S6 to generate a renewable energy system operation scenario, specifically includes the following steps: randomly sampling a Gaussian noise from an arbitrary standard Gaussian distribution. Input the trained conditional diffusion model obtained in step S6 to obtain the model-generated... and according to Complete the generation of renewable energy system operation scenarios.

4. A system for implementing the renewable energy system operation scenario generation method according to any one of claims 1 to 3, characterized in that... It includes a data acquisition module, a diffusion modeling module, a Gaussian denoising module, a model modeling module, a network construction module, a model guidance module, and a scene generation module; the data acquisition module, diffusion modeling module, Gaussian denoising module, model modeling module, network construction module, model guidance module, and scene generation module are connected in series; the data acquisition module is used to acquire historical data information of the target renewable energy system and upload the data information to the diffusion modeling module; The diffusion modeling module is used to establish a diffusion process model for the target renewable energy system's operating scenario based on the received and acquired data, and upload the data to the Gaussian noise reduction module; the Gaussian noise reduction module is used to perform noise reduction based on Gaussian noise based on the received and acquired data and the established diffusion process model, and upload the data to the modeling module. The model building module is used to model the conditional diffusion model based on the received data information and upload the data information to the network building module; the network building module is used to build a neural network for the conditional diffusion model based on the received data information and upload the data information to the model guidance module. The model guidance module is used to generate conditional information based on the received data information using a large model, train the established conditional diffusion model to obtain the trained conditional diffusion model, and upload the data information to the scene generation module. The scenario generation module is used to generate operating scenarios for renewable energy systems based on the received data and the trained conditional diffusion model.

Images