Wind-light charge extreme scene generation method, system and device and medium

Abstract

The invention discloses a wind-light charge extreme scene generation method, system and device and a medium. The method comprises the following steps: extracting key features influencing extreme wind-light-charge scene generation from historical data; constructing an improved information maximization generative adversarial neural network model based on the key features; training control parameters of the model by using a meteorological feature scene and wind-light-load output data, and generating a control parameter sequence reflecting probability distribution of different extreme scenes; extreme scene data are generated through the trained model, and the influence of the generated scene on system operation is evaluated based on the power system model. According to the method, the coupling characteristics of wind, light and load under the extreme weather condition can be reflected more accurately, a controllable scene with clear physical significance is generated, the number of errors generated by the scene is remarkably reduced, the coverage rate and accuracy of extreme scene generation are improved, and reliable extreme scene boundary conditions are provided for safe operation of a power system.

Description

Technical Field

[0001] The present invention relates to the technical field of wind, solar and load extreme scenario generation technology, and in particular to a wind, solar and load extreme scenario generation method, system, equipment and medium. Background Art

[0002] As my country advances its "dual carbon" goals, the share of renewable energy, primarily wind and photovoltaic power, in the power system continues to increase. However, the inherent intermittent and volatile nature of renewable energy, coupled with frequent extreme weather events, has led to uncertainties on both the power supply and load sides of the system. On the power supply side, wind and solar output often exhibits chronically low or drastic fluctuations under extreme weather conditions. On the load side, the rapid growth of new loads, such as electric vehicles, has further exacerbated the uncertainty of the load curve. This dual uncertainty poses significant challenges to the safe and stable operation and optimal dispatch of the power system.

[0003] Currently, research on renewable energy scenario generation falls into two main categories: scenario generation methods based on probabilistic statistical characteristics and data-driven scenario generation methods. Traditional probabilistic statistical methods typically assume that wind speed follows a Weibull distribution and light intensity follows a Beta distribution, generating scenarios through Monte Carlo sampling. However, these methods are limited by their pre-set distribution assumptions and struggle to accurately reflect the temporal and spatial variations in actual wind and solar output, with errors becoming more pronounced under extreme weather conditions. Among data-driven methods, generative adversarial networks (GANs) have garnered widespread attention due to their powerful feature learning capabilities. While traditional GANs and their improved models, such as WGAN-GP, can capture the stochastic characteristics of renewable energy, they suffer from uncontrollable scenario generation and poor interpretability. Conditional generative adversarial networks (CGANs) improve controllability to some extent by introducing label information, but are still limited by their ability to express discrete conditional information, making it difficult to achieve continuous and smooth adjustment of scenario features.

[0004] It is worth noting that existing research has mostly focused on generating typical scenarios for a single energy source, lacking exploration of methods for the coordinated generation of multiple extreme wind-solar-load scenarios. In fact, under extreme weather conditions, there is often a strong correlation between wind, solar, and load. This coupling characteristic has a significant impact on the power supply security and renewable energy consumption of the power system. Therefore, developing a method that can accurately characterize the wind-solar-load correlation characteristics and achieve controllable generation of extreme scenarios is of great significance to improving the operational safety and dispatch scientificity of new power systems. Summary of the Invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] Therefore, the present invention provides a method, system, device and medium for generating extreme wind, solar and load scenarios to solve the problem that existing extreme scenario generation methods do not consider the wind-solar-load correlation, resulting in inaccurate generated scenarios.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] In a first aspect, the present invention provides a method for generating a wind-solar-load extreme scenario, comprising:

[0009] Extract key features that influence the generation of extreme wind-light-load scenarios from historical data;

[0010] Based on the key features, an improved information maximization generative adversarial neural network model is constructed;

[0011] The control parameters of the model are trained using meteorological characteristic scenarios and wind-solar-load output data, and a control parameter sequence reflecting the probability distribution of different extreme scenarios is generated;

[0012] Extreme scenario data is generated through the trained model, and the impact of the generated scenarios on system operation is evaluated based on the power system model.

[0013] As a preferred solution of the method for generating extreme wind-solar-load scenarios described in the present invention, the key features that affect the generation of extreme wind-solar-load scenarios from historical data include:

[0014] Compressing the input data into the same interval by a first processing operation;

[0015] Calculates the cumulative distribution function of the input data and maps the data to the corresponding percentiles of the target uniform distribution;

[0016] The average index and slope index of extreme scenarios are calculated, and the quantitative data features are weighted to obtain the characteristic values ​​of extreme scenarios.

[0017] The beneficial effects of this preferred technical solution are: effectively capturing the static level and dynamic change characteristics of power output in extreme scenarios, and significantly improving the integrity and accuracy of feature extraction.

[0018] As a preferred solution of the method for generating extreme wind-solar-load scenarios described in the present invention, the control parameters of the model trained using meteorological characteristic scenarios and wind-solar-load output data include:

[0019] Establish a feature matrix based on the selected meteorological feature scenario and wind-solar-load output data;

[0020] Based on the characteristic matrix, the control parameters of the model are trained to obtain a control parameter sequence that reflects the probability distribution of wind, solar and load under different extreme scenarios.

[0021] The control parameter sequence includes discrete categories and continuously adjustable features.

[0022] As a preferred solution of the method for generating a wind, solar and load extreme scenario according to the present invention, the evaluation of the power system model includes:

[0023] Establish a power system model with the objective function of ensuring power supply and consumption;

[0024] Set energy balance constraints, wind power consumption constraints, and transmission line capacity constraints;

[0025] The power system model is used to evaluate the operation level of the power system under the generated extreme scenarios.

[0026] As a preferred solution of the method for generating extreme wind, solar and load scenarios described in the present invention, the improved information maximization generative adversarial neural network model includes a generator, a discriminator and an auxiliary network;

[0027] The generator consists of a fully connected layer, a convolutional sampling layer, and a ConvNextBlock layer, and is used to generate data samples that simulate real wind-solar-load extreme scenarios based on input control parameters and random noise;

[0028] The discriminator consists of a ConvNextBlock layer, a fully connected layer, and an output layer, and is used to distinguish between the simulated scene data generated by the generator and the real historical scene data, and improve the generation quality through adversarial training;

[0029] The auxiliary network is used to maximize the mutual information between the control parameters and the generated samples to reduce the loss of encoding information during the generation process.

[0030] The beneficial effects of this preferred technical solution are: the generator adopts a multi-level network structure to ensure the restoration of scene details, the discriminator improves the generation authenticity through adversarial training, and the auxiliary network ensures the interpretability of the generated scene by maximizing mutual information. The three work together to make the generated extreme scenes both realistic, diverse and controllable.

[0031] As a preferred solution of the method for generating extreme wind, solar and load scenarios described in the present invention, the generator further includes a time embedding module for merging time step information into the input data and adding it to the output data after convolution processing.

[0032] As a preferred solution of the method for generating extreme wind, solar and load scenarios described in the present invention, the improved information maximization generative adversarial neural network model adopts a symmetric convolution structure to retain rich contextual information of the input data.

[0033] In a second aspect, the present invention provides a system for generating extreme wind, solar and load scenarios, comprising:

[0034] Feature extraction module, used to extract key features that affect the generation of extreme wind-light-load scenarios from historical data;

[0035] A model building module, configured to build an improved information maximization generative adversarial neural network model based on the key features;

[0036] a parameter training module for training control parameters of the improved information maximization generative adversarial neural network model using meteorological characteristic scenarios and wind-solar-load output data, and generating a control parameter sequence reflecting the probability distribution of different extreme scenarios;

[0037] The generation and evaluation module is used to generate extreme scenario data through the trained model and evaluate the impact of the generated scenarios on system operation based on the power system model.

[0038] In a third aspect, the present invention provides an electronic device comprising a memory and a processor; the memory is used to store computer-executable instructions, and the processor implements the steps of a method for generating extreme wind, solar and load scenarios when executing the computer-executable instructions.

[0039] In a fourth aspect, the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of a method for generating extreme wind, solar and load scenarios.

[0040] Compared with the existing technology, the beneficial effects of the present invention are as follows: the present invention provides a method, system, equipment and medium for generating extreme scenarios of wind, light and load, which effectively solves the key problems of existing scenario generation methods in terms of accuracy, controllability and practicality by innovatively integrating feature extraction technology and deep generation models. The present invention first uses a feature extraction module to accurately capture the mathematical characteristics and correlation characteristics of wind, light and load outputs, and then constructs an improved InfoGAN model including a generator, a discriminator and an auxiliary network, and realizes the interpretable generation of extreme scenarios by maximizing the mutual information between control parameters and generated samples. At the same time, a deep convolutional network and a time embedding module are used to enhance the model stability and spatiotemporal feature extraction capabilities. Compared with traditional methods, the present invention can more accurately reflect the coupling characteristics of wind, light and load under extreme weather conditions, generate controllable scenarios with clear physical meanings, significantly reduce the number of errors in scenario generation, improve the coverage and accuracy of extreme scenario generation, provide reliable extreme scenario boundary conditions for the safe operation of the power system, and significantly enhance the ability of new power systems to cope with source and load uncertainties. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0042] Figure 1 This is a logical diagram of the overall process of the method for generating wind, solar and load extreme scenarios according to an embodiment of the present invention.

[0043] Figure 2 This is a schematic diagram of the improved information maximization generative adversarial network framework of the wind, solar and load extreme scenario generation method described in an embodiment of the present invention.

[0044] Figure 3 This is a schematic diagram of the ConvNextBlock layer framework of the wind, solar and load extreme scenario generation method described in an embodiment of the present invention.

[0045] Figure 4 Schematic diagram of the control parameter training framework of the wind-solar-load extreme scenario generation method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0046] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention.

[0047] Example 1, with reference to Figure 1-Figure 4 As an embodiment of the present invention, a method for generating extreme wind-solar-load scenarios is provided. Figure 1 The specific steps shown include:

[0048] S100: Extract key features that influence the generation of extreme wind-solar-load scenarios from historical data;

[0049] S200: Building an improved information maximization generative adversarial neural network model based on key features;

[0050] S300: Using meteorological characteristic scenarios and wind-solar-load output data to train the control parameters of the model, and generating a control parameter sequence reflecting the probability distribution of different extreme scenarios;

[0051] S400: Generate extreme scenario data through the trained model, and evaluate the impact of the generated scenario on system operation based on the power system model.

[0052] It should be noted that, given that existing extreme scenario generation methods do not consider the correlation between wind, solar power, and load, resulting in inaccurate generated scenarios, the above steps S100 to S400 effectively solve the key problems of existing scenario generation methods in terms of accuracy, controllability, and practicality by innovatively integrating feature extraction technology with deep generative models. First, a feature extraction module is used to accurately capture the mathematical characteristics and correlation characteristics of wind, solar power, and load outputs. Subsequently, an improved InfoGAN model consisting of a generator, a discriminator, and an auxiliary network is constructed. By maximizing the mutual information between control parameters and generated samples, the interpretable generation of extreme scenarios is achieved. At the same time, a deep convolutional network and a time embedding module are used to enhance the model stability and spatiotemporal feature extraction capabilities. Compared with traditional methods, the present invention can more accurately reflect the coupling characteristics of wind, solar power, and load under extreme weather conditions, generate controllable scenarios with clear physical meaning, significantly reduce the number of errors in scenario generation, improve the coverage and accuracy of extreme scenario generation, provide reliable extreme scenario boundary conditions for the safe operation of the power system, and significantly enhance the ability of new power systems to cope with source and load uncertainties.

[0053] In the embodiment of the present application, step S100 extracts key features that affect the generation of extreme wind-solar-load scenarios from historical data. Specifically, the steps include:

[0054] Compressing the input data into the same interval by a first processing operation;

[0055] Calculates the cumulative distribution function of the input data and maps the data to the corresponding percentiles of the target uniform distribution;

[0056] The average index and slope index of extreme scenarios are calculated, and the quantitative data features are weighted to obtain the characteristic values ​​of extreme scenarios.

[0057] It should be noted that in recent years, due to the impact of global climate change, extreme weather events have shown a trend of increasing frequency, increasing severity, and the simultaneous occurrence of multiple extreme weather events. Due to changes in wind and solar energy resources and climate risks, these changes have significantly constrained the stability and reliability of renewable energy generation within power systems, thereby threatening the security of power supply. Currently, research on extreme scenarios in power systems primarily defines these scenarios based on climatic conditions and their impact on production and daily life. For example, extreme heat and drought pose severe challenges to power supply security in systems heavily reliant on hydropower, while typhoon conditions have led to rapid growth in wind power generation, placing significant pressure on renewable energy grid integration. To further analyze the extreme scenarios facing power system power supply security and renewable energy grid integration, this embodiment introduces extreme meteorological factors as classification criteria. Because wind and solar power generation are significantly affected by weather, meteorological conditions, and natural disasters, prolonged high temperatures, extreme cold weather, typhoons, lack of sunshine, and no wind are considered typical extreme scenarios within power systems due to their widespread impact and high frequency of occurrence. Therefore, this invention categorizes extreme scenarios into five categories: sustained high temperatures, extreme cold waves, typhoons, lack of sunshine, and no wind. In addition, according to the "Extreme Low Temperature Cooling Detection Index" issued by my country, specific standards for extreme scenarios are defined, as shown in Table 1.

[0058] Table 1: Definition of extreme scenarios

[0059] extreme weather events Feature Selection Long-term high temperature Temperature A, duration T1 Extremely cold weather Temperature B, duration T2 Typhoon weather Wind speed C, wind turbine power growth rate T3 Long-term windless Wind speed D, duration T4 Long-term lack of sunlight Radiation intensity E, duration T5

[0060] Furthermore, we analyze extreme power system scenarios based on defined extreme weather events and extract key features that influence their generation. The wind-solar-load output of a power system is affected by various weather conditions, and this paper selects the five aforementioned extreme weather conditions. In addition to weather factors, load power demand is also influenced by time. Therefore, we select input features for extreme scenario generation, as shown in Table 2. Within the selected extreme weather events, each extreme scenario affects the wind-solar-load output.

[0061] Table 2: Potential factors affecting wind-solar-load scenario generation

[0062]

[0063]

[0064] It should be noted that under extremely hot and windless conditions, wind power output is chronically low, while solar power output is chronically high, resulting in long-term high load demand. During cold waves and typhoons, wind power output increases rapidly, while photovoltaic power output and long-term low load demand increase. When considering extreme weather conditions, changes in wind-solar-load output are concentrated on the mathematical characteristics of mean and slope. Therefore, for classification of extreme scenarios, it is necessary to extract mathematical features related to wind-solar-load output.

[0065] In an optional embodiment, the first processing operation may be a maximum-minimum normalization operation, which linearly transforms the original data into the interval [0, 1], which is most suitable for parameters with clear physical boundaries such as wind and solar load power.

[0066] In another optional embodiment, the first processing operation may also be a decimal scaling operation, which achieves normalization by moving the decimal point position, and is suitable for scenarios that require rapid processing of large-scale data.

[0067] In this embodiment of the present application, uniform distribution normalization is used to compress input features into the same interval. The input data is mapped to a specified uniform distribution by calculating the cumulative distribution function of the input data and then mapping the data to the corresponding percentiles of the target distribution. The specific implementation steps are as follows:

[0068] Calculate percentiles:

[0069]

[0070] Among them, rank(x i ) indicates that the data is sorted in ascending order. i Position in the data set, n represents the total number of data points;

[0071] Generate uniform distribution: Define a uniform distribution U on the interval [0,1] to generate values ​​corresponding to the percentiles of the input data in the uniform distribution;

[0072] Use a linear mapping to map the percentile of each data point to the corresponding position in the target uniform distribution.

[0073] Furthermore, the average index and slope index of the extreme scenarios are calculated, and the quantitative data characteristics are weighted. In order to comprehensively consider the impact of the extreme scenario on the wind-solar hybrid load under consideration, the characteristic value of the scenario is obtained, and it is determined whether the extreme scenario is a power protection scenario or a power consumption protection scenario. The formula is expressed as:

[0074]

[0075] S t =μ ave,t Ave t +μ slo,t Slo t

[0076] Among them, Ave t Represents the average index of extreme scenarios, describing x t The power level, x t represents wind power, photovoltaic output and load output, m and n represent the start and duration of extreme scenarios respectively, Slot Indicates the slope index of the extreme scenario, x t-m and x t+n They represent the power values ​​of m periods before and n periods after time t, respectively. S t Indicates the eigenvalue of the extreme scenario, μ ave,t and μ slo,t Represents the weighting coefficient of the two types of indicators.

[0077] It should be noted that the above step S100 realizes the standardized processing of data of different dimensions through uniform distribution normalization processing and cumulative distribution function mapping, so that multivariate heterogeneous data such as wind, light, and load can be feature extracted and analyzed at the same scale; by calculating the average index and slope index and weighted processing, the static level and dynamic change characteristics of power output in extreme scenarios are effectively captured, providing more representative feature input for subsequent model training, and significantly improving the integrity and accuracy of feature extraction.

[0078] In the embodiment of the present application, the above step S200 constructs an improved information maximization generative adversarial neural network model based on key features, including:

[0079] It should be noted that the core idea of ​​Generative Adversarial Networks (GANs) is a two-player zero-sum game in game theory, which is a type of unsupervised learning. In a GAN zero-sum game, the participants are composed of a generator and a discriminator. As the two parties learn the game, the accuracy of the model improves. When the game equilibrium is reached, the model training is considered complete. Figure 2 As shown, the InfoGAN generative network divides its input into two parts: incompressible random noise z and a control parameter c, which has interpretable characteristics and can control the features of the generated samples. By maximizing the mutual information between the control parameter c and the generated samples G(z,c) and minimizing the loss of encoded information during the generation process, the generator is interpretable and practical for the control parameter c. Finally, a mapping relationship is established between the control parameter c, the generated samples G(z,c), and the interpretable features. By adjusting the control parameter c, the generated samples G(z,c) are controlled.

[0080] In the InfoGAN generation network, in order to reduce the loss of the control parameter c, the objective function is as follows:

[0081]

[0082] I(c;G(z,c))=H(c)-H(c|G(z,c))

[0083] Where D(·) represents the discriminator, G(·) represents the generator, λ represents the weight of the mutual information in the objective function, I(c; G(z,c)) represents c and G(z,c), H(c) represents the information entropy of the control parameter c, that is, the quantification of the amount of information contained in the control parameter c, and H(c|G(z,c)) represents the information entropy of the control parameter c under the following conditions: the known conditions of the production sample G(z,c).

[0084] It should be noted that in unsupervised learning, the generation of extreme scenarios of wind-solar-load involves discrete categories and continuously adjustable interpretable features. By adjusting these interpretable features, the generation process of extreme scenarios can be effectively controlled, thereby enhancing the interpretability and generalization of the model.

[0085] In the embodiment of the present application, in order to enhance the quality of extreme scene generation and improve the stability of model training, the convolutional neural network (CNN) is combined with the generative adversarial network (GAN) to establish a deep convolutional adversarial generative network as the core structure of the improved InfoGAN. The model consists of three main parts: generator (G), discriminator (D) and auxiliary network (Q);

[0086] Specifically, the generator consists of a fully connected layer, a convolutional sampling layer, and a ConvNextBlock layer, which is used to generate data samples simulating real wind-solar-load extreme scenarios based on the input control parameters and random noise. The framework of the ConvNextBlock layer is as follows: Figure 3 As shown;

[0087] It should be noted that the generator also includes a time embedding module for incorporating time step information into the input data and adding it to the output data after the convolution process. This operation allows the gradient to flow directly through the network, alleviating the gradient vanishing problem.

[0088] Specifically, the discriminator consists of a ConvNextBlock layer, a fully connected layer, and an output layer. It is used to distinguish between the simulated scene data generated by the generator and the real historical scene data, and improve the generation quality through adversarial training.

[0089] Specifically, the auxiliary network is used to maximize the mutual information between the control parameters and the generated samples to reduce the loss of encoded information during the generation process.

[0090] In an embodiment of the present application, in view of the high temporal correlation of the input sample data, the improved information maximization generative adversarial neural network model adopts a symmetric convolution structure to retain the rich contextual information of the input data.

[0091] It should be noted that the above step S200 realizes the generation and optimization of high-quality extreme scenes by constructing a ternary adversarial network architecture including a generator, a discriminator and an auxiliary network; the generator adopts a multi-level network structure to ensure the restoration of scene details, the discriminator improves the generation authenticity through adversarial training, and the auxiliary network ensures the interpretability of the generated scenes by maximizing mutual information. The three work together to make the generated extreme scenes both authentic, diverse and controllable.

[0092] In the embodiment of the present application, the above step S300 includes:

[0093] Establish a feature matrix based on the selected meteorological feature scenario and wind-solar-load output data;

[0094] Based on the characteristic matrix, the control parameters of the model are trained to obtain a control parameter sequence that reflects the probability distribution of wind, solar and load under different extreme scenarios.

[0095] The control parameter sequence contains discrete categories and continuously adjustable features.

[0096] It should be noted that in the traditional InfoGAN generation network, the control parameter c is usually assumed to be a discrete sequence obeying a fixed uniform probability distribution, as shown in the following formula:

[0097]

[0098] Where S represents the discrete probability distribution followed by the control parameter c, which is usually assumed to be uniformly distributed, k represents the category of the discrete sequence, and p represents the probability of each category;

[0099] However, in actual scenarios, the probability of each extreme event occurring is not the same, and there are also large differences in wind-solar-load output. Uniform extreme events mean that each output force does not obey a uniform distribution. In the improved InfoGAN model proposed in this embodiment, the training process of the control parameter c is as follows: Figure 4 First, a feature matrix is ​​established based on the selected meteorological characteristic scenario and historical wind-solar-load output data. The control parameter c is trained based on the feature map, and finally a control parameter sequence that can reflect the probability distribution of wind load under different extreme scenarios is obtained.

[0100] It should be noted that the control parameter sequence obtained by training in the above step S300 contains both discrete categories and continuously adjustable features, which not only retains the distinguishability of typical extreme scenarios, but also realizes the fine adjustment of scenario features, greatly enhancing the controllability and flexibility of the model generation scenario, and can meet the needs of different application scenarios.

[0101] In the embodiment of the present application, the evaluation of the power system model in the above step S400 includes:

[0102] Establish a power system model with the objective function of ensuring power supply and consumption;

[0103] Set energy balance constraints, wind power consumption constraints, and transmission line capacity constraints;

[0104] The power system model is used to evaluate the operating level of the power system under the generated extreme scenarios.

[0105] Specifically, taking the guaranteed power supply and consumption of the power system as the objective function, the level of guaranteed power supply and consumption of the power system under extreme scenarios is evaluated:

[0106] minF=F L +F Re

[0107] Among them, F L represents the cost of reducing the load, F Re This is the penalty for renewable energy curtailment. The specific calculation formula is as follows:

[0108]

[0109] Among them, λ load is the unit load reduction penalty coefficient, λ RE is the unit renewable energy abandonment penalty coefficient, is the load reduction amount in time period t, node j, and scenario s, is the abandoned energy in time period t, node j, and scenario s, T is the total number of scheduling time periods, N G is the total number of conventional units, N RE is the total number of renewable energy power stations.

[0110] The constraints of the power system are as follows:

[0111]

[0112] in, is the output of conventional unit i in time period t and scenario s, is the actual output of renewable energy power station i in time period t and scenario s, B n,m is the line susceptance between nodes n and m, δ t,n,s is the voltage phase angle of node n in time period t and scenario s, is the load demand of node j in time period t and scenario s, N D is the total number of load nodes, is the predicted output of renewable energy power station i in time period t and scenario s, is the actual absorbed power in time period t, renewable energy plant i, and scenario s, is the maximum transmission capacity of line nm, δ t,1,s is the phase angle of node 1, is the start and stop status of unit i in time period t and scene s, is the minimum technical output of unit i, The maximum technical output of unit i, is the maximum ramp rate of unit i, is the maximum ramp-down rate of unit i, is the cumulative continuous operating time of unit i in time period t and scenario s, is the cumulative continuous downtime of unit i in time period t and scenario s, is the minimum continuous operating time of unit i, is the minimum downtime of unit i, is the start time of unit i in period t-1 and scenario s, is the downtime of unit i in period t-1 and scenario s.

[0113] It should be noted that the above step S400 realizes the practical verification of generating extreme scenarios by constructing a power system evaluation model that includes supply and consumption guarantee objective functions and multi-dimensional constraints; this evaluation system not only considers conventional energy balance constraints, but also pays special attention to key operating restrictions such as wind power absorption and transmission capacity, ensuring that the generated extreme scenarios can truly reflect the boundary conditions of power system operation, providing a reliable basis for system safety assessment.

[0114] Example 2: Based on the previous example, this example provides an application example of a method for generating extreme wind, solar and load scenarios, in order to verify and illustrate the technical effects adopted in this method.

[0115] This example verifies the correctness and effectiveness of the model proposed in the present invention through a case study. The relevant parameter settings are as follows: the wind energy, solar energy, and load historical data sets are taken from the publicly available measured data of the Belgian power grid, and the meteorological data are provided by the Wunderground website. The data set contains all historical data from 2018 to 2022, and the time step is set to 15 minutes; therefore, there are 96 data points in the 24-hour daily production / load curve. 80% of the data set is used as a training set, and the remaining 20% ​​is used as a test set. All simulations are performed on the Python-based TensorFlow 2.3.0 platform. The computer configuration is Intel i7-8700 3.20GHz CPU, 16.0GB RAM, and NVIDIA GTX 1660GPU.

[0116] Case 1: Traditional probability sampling model;

[0117] Case 2: Traditional Generative Adversarial Network Model;

[0118] Case 3: Propose an improved information maximization generative adversarial network model.

[0119] In Case 1, the empirical probability distribution is used for sampling, with wind speed following a Weibull distribution, solar intensity following a Beta distribution, and load following a normal distribution. Table 3 provides the relevant parameters for each distribution.

[0120] Table 3: Distribution parameter settings

[0121] Classification parameter value wind energy Scale 1 4.738 wind energy Variance β1 2.684 solar energy Scale α2 0.326 solar energy Variance β2 1.764 load Average μ(MW) 724.3701 load Variance σ(MW) 147.3181

[0122] To accurately evaluate the performance of these three models in capturing the characteristics of historical wind, solar, and load series data, we introduced an extreme scenario ensemble effectiveness index, focusing on two main indicators: extreme scenario coverage and the average width of the power interval. The calculation formulas for coverage and effectiveness are as follows:

[0123]

[0124] Among them, T′ represents the number of moments when the real scene data is included in the scene set, T represents the total number of moments in the output sequence, and C is the coverage rate. The larger its value is, the more reliable the generated scene set is. is the upper bound on the set of scenarios generated at time t, It is the lower bound of the scene set generated at time t, and W is the average width of the power interval of the generated scene. The smaller its value is, the closer the generated scene data is to the real one.

[0125] The generated scenarios are analyzed using effectiveness metrics, and comparative experiments yield effectiveness metrics for scenario data generated by the three models, as shown in Table 4. This comparison highlights the differences in the models’ ability to fit extreme scenario data and their effectiveness in capturing power interval variations.

[0126] Table 4: Overall effectiveness indicators of the scenario set under Cases 1-3

[0127]

[0128] Analysis of Table 5 shows that the data generated by the improved InfoGAN model in Case 3 for wind power, solar power generation, and power load scenarios exhibits a higher coverage index C than other cases. Furthermore, Case 3 exhibits a smaller average power interval width index W than other models. This comparison demonstrates that the InfoGAN model in Case 3 achieves higher coverage while maintaining a lower average power interval width, indicating that the scenario data generated by the proposed model is more realistic and reliably consistent with real-world conditions.

[0129] Table 5: Average Wasserstein distances for the scenario sets under Cases 1-3

[0130] Classification wind energy solar energy load Case 1 1.643 1.019 1.438 Case 2 1.283 0.924 1.361 Case 3 1.036 0.582 0.914

[0131] The above analysis demonstrates that the improved InfoGAN wind-load extreme scenario generation method proposed in this paper can significantly reduce the number of errors in scenario generation and improve the coverage of extreme scenarios. Compared with traditional statistical methods and GAN neural networks, the proposed method achieves higher accuracy and generates superior scenarios. Through simulations of real-world power grid production, the performance of the power system under extreme scenarios is evaluated from the perspectives of power supply and energy consumption.

[0132] Example 3: This embodiment provides a system for generating extreme wind, solar and load scenarios, including:

[0133] Feature extraction module, used to extract key features that affect the generation of extreme wind-light-load scenarios from historical data;

[0134] A model building module for building an improved information maximization generative adversarial neural network model based on key features;

[0135] A parameter training module is used to maximize the generation of control parameters for the adversarial neural network model using improved information from meteorological characteristic scenarios and wind-solar-load output data, and to generate a control parameter sequence that reflects the probability distribution of different extreme scenarios;

[0136] The generation and evaluation module is used to generate extreme scenario data through the trained model and evaluate the impact of the generated scenarios on system operation based on the power system model.

[0137] It should be noted that the technical solution of the wind-solar-load extreme scenario generation system and the technical solution of the above-mentioned wind-solar-load extreme scenario generation method belong to the same concept. For the details not described in detail in the technical solution of the wind-solar-load extreme scenario generation system in this embodiment, please refer to the description of the technical solution of the above-mentioned wind-solar-load extreme scenario generation method.

[0138] The above-mentioned unit modules can be embedded in or independent of the processor in the electronic device in the form of hardware, or can be stored in the memory of the electronic device in the form of software, so that the processor can call and execute the corresponding operations of the above-mentioned modules.

[0139] This embodiment also provides an electronic device, which includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the electronic device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a method for generating extreme wind, solar and load scenarios is implemented. The display screen of the electronic device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the electronic device can be a touch layer covering the display screen, or a button, trackball or touchpad provided on the housing of the electronic device, or an external keyboard, touchpad or mouse.

[0140] This embodiment further provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the method proposed in the above embodiment is implemented.

[0141] The storage medium proposed in this embodiment and the method proposed in the above embodiment belong to the same inventive concept. For technical details not fully described in this embodiment, please refer to the above embodiment, and this embodiment has the same beneficial effects as the above embodiment.

[0142] Through the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented with the help of software and necessary general-purpose hardware, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory (FLASH), hard disk or optical disk, etc., including a number of instructions for enabling an electronic device (which can be a personal computer, server, or network device, etc.) to execute the method of the embodiment of the present invention.

[0143] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

[0144] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code. The scheme in the embodiment of the present application can be implemented in various computer languages.

[0145] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0146] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0147] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0148] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present application.

[0149] Obviously, those skilled in the art may make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalents, this application is intended to include these modifications and variations.

Claims

1. A method for generating extreme wind, solar and load scenarios, characterized in that: include: Extract key features that influence the generation of extreme wind-light-load scenarios from historical data; Based on the key features, an improved information maximization generative adversarial neural network model is constructed; The control parameters of the model are trained using meteorological characteristic scenarios and wind-solar-load output data, and a control parameter sequence reflecting the probability distribution of different extreme scenarios is generated; Extreme scenario data is generated through the trained model, and the impact of the generated scenarios on system operation is evaluated based on the power system model.

2. The method for generating extreme wind, solar and load scenarios according to claim 1, characterized in that: The key features that influence the generation of extreme wind-solar-load scenarios extracted from historical data include: Compressing the input data into the same interval by a first processing operation; Calculates the cumulative distribution function of the input data and maps the data to the corresponding percentiles of the target uniform distribution; The average index and slope index of extreme scenarios are calculated, and the quantitative data features are weighted to obtain the characteristic values ​​of extreme scenarios.

3. The method for generating a wind-solar-load extreme scenario according to claim 2, characterized in that: The control parameters of the model trained using meteorological characteristic scenes and wind-solar-load output data include: Establish a feature matrix based on the selected meteorological feature scenario and wind-solar-load output data; Based on the characteristic matrix, the control parameters of the model are trained to obtain a control parameter sequence that reflects the probability distribution of wind, solar and load under different extreme scenarios. The control parameter sequence includes discrete categories and continuously adjustable features.

4. The method for generating a wind-solar-load extreme scenario according to claim 3, characterized in that: The evaluation of the power system model includes: Establish a power system model with the objective function of ensuring power supply and consumption; Set energy balance constraints, wind power consumption constraints, and transmission line capacity constraints; The power system model is used to evaluate the operation level of the power system under the generated extreme scenarios.

5. The method for generating extreme wind, solar and load scenarios according to claim 1, wherein: The improved information maximization generative adversarial neural network model includes a generator, a discriminator and an auxiliary network; The generator consists of a fully connected layer, a convolutional sampling layer, and a ConvNextBlock layer, and is used to generate data samples that simulate real wind-solar-load extreme scenarios based on input control parameters and random noise; The discriminator consists of a ConvNextBlock layer, a fully connected layer, and an output layer, and is used to distinguish between the simulated scene data generated by the generator and the real historical scene data, and improve the generation quality through adversarial training; The auxiliary network is used to maximize the mutual information between the control parameters and the generated samples to reduce the loss of encoding information during the generation process.

6. The method for generating a wind-solar-load extreme scenario according to claim 5, characterized in that: The generator also includes a time embedding module for incorporating time step information into the input data and adding it to the output data after convolution processing.

7. The method for generating extreme wind, solar and load scenarios according to claim 5, wherein: The improved information maximization generative adversarial neural network model adopts a symmetric convolutional structure to retain the rich contextual information of the input data.

8. A system for generating extreme wind, solar and load scenarios, using a method for generating extreme wind, solar and load scenarios as claimed in any one of claims 1 to 7, characterized in that: include: Feature extraction module, used to extract key features that affect the generation of extreme wind-light-load scenarios from historical data; A model building module, configured to build an improved information maximization generative adversarial neural network model based on the key features; a parameter training module for training control parameters of the improved information maximization generative adversarial neural network model using meteorological characteristic scenarios and wind-solar-load output data, and generating a control parameter sequence reflecting the probability distribution of different extreme scenarios; The generation and evaluation module is used to generate extreme scenario data through the trained model and evaluate the impact of the generated scenarios on system operation based on the power system model.

9. An electronic device comprising a memory and a processor, characterized in that: The memory is used to store computer-executable instructions, and when the processor executes the computer-executable instructions, it implements the steps of the method for generating a wind-solar-load extreme scenario as described in any one of claims 1 to 7.

10. A computer-readable storage medium having computer-executable instructions stored thereon, characterized in that: When the computer executable instructions are executed by the processor, the steps of the method for generating a wind-solar-load extreme scenario as described in any one of claims 1 to 7 are implemented.

Images