Power system panoramic scene generation method integrating conventional scene and extreme scene

Abstract

The invention discloses a power system panoramic scene generation method integrating a conventional scene and an extreme scene, and belongs to the technical field of power system dispatching optimization. Calculating a probability value and a safety margin index value of each operation mode to construct a high-probability operation mode subset and a low-margin operation mode subset, and generating an expanded fault scene set; calculating a fault occurrence probability value and a risk index value to construct a high-probability fault subset and a high-risk fault subset; and combining the operation modes in the high-probability operation mode subset with the fault scenes in the high-probability fault subset to construct a conventional scene set, combining the operation modes in the low-margin operation mode subset with the fault scenes in the high-risk fault subset to construct an extreme scene set, and forming a power system panoramic scene set. According to the method, the problems of insufficient coverage of extreme scenes and insufficient fault sample feature mining in an existing power system panoramic scene generation technology are solved.

Description

Technical Field

[0001] This invention relates to a method for generating panoramic power system scenes that integrates conventional and extreme scenarios, belonging to the field of power system dispatch optimization technology. Background Technology

[0002] Existing panoramic scene generation technologies that reflect the possible future operating states of a system are typically based on historical data, using probabilistic modeling or scene generation to account for uncertainties such as wind power, solar power output, and load. Regarding operating modes, methods such as cluster analysis are often used to analyze historical wind, solar, and load data to extract representative typical operating modes and calculate their probability of occurrence. For fault analysis, historical fault statistics are primarily relied upon, using probabilistic models to filter high-probability faults. Finally, by combining high-probability operating modes with high-probability faults, a set of regular scenarios for analysis is formed. This approach focuses on the system's routine operation most of the time, aiming to cover frequently occurring scenarios in power system operation and provide a reference for daily dispatching.

[0003] However, the existing methods have significant shortcomings. First, the panoramic scene sets they generate are heavily biased towards high-probability events, neglecting extreme scenarios with low probability of occurrence but severe consequences. These extreme scenarios are usually triggered by external factors such as severe weather. In these situations, the system's operation is already under considerable stress, and the probability of equipment failure is significantly increased. The coupling of these two factors can easily lead to serious system risks, but existing methods fail to effectively capture the correlation between this operational vulnerability and the high risk of failure during scene construction. Second, in constructing the fault set, relying solely on limited historical fault records may result in insufficient representation of low-probability, high-risk faults. The lack of sufficient mining and sample expansion of fault characteristics under extreme conditions makes it difficult for the generated scene set to comprehensively cover extreme events and scenarios. Summary of the Invention

[0004] The purpose of this invention is to provide a method for generating panoramic power system scenes that integrates conventional and extreme scenarios. By ensuring the accuracy of conventional scenarios while expanding the coverage of extreme scenarios in the panoramic power system scene, this method addresses the problems of insufficient coverage of extreme scenarios and inadequate feature mining of fault samples in existing panoramic power system scene generation technologies.

[0005] To solve the above-mentioned technical problems, the present invention is implemented using the following technical solution.

[0006] This embodiment provides a method for generating panoramic power system scenes that integrates conventional and extreme scenarios, including:

[0007] Acquire wind power data, photovoltaic data, load data, and fault data;

[0008] Based on wind power data, photovoltaic data, and load data, a clustered set of operating modes is extracted using a trained variational autoencoder.

[0009] Calculate the probability value and safety margin index value of each operating mode in the set of operating modes, and select a target number of operating modes to construct a high-probability operating mode subset and a low-margin operating mode subset respectively;

[0010] Based on fault data, an expanded set of fault scenarios is generated using a trained generative adversarial network.

[0011] Calculate the failure probability value and risk index value of each failure scenario in the failure scenario set, and select a target number of failure scenarios to construct a high-probability failure subset and a high-risk failure subset respectively;

[0012] By using the power flow matching method, each operating mode in the high-probability operating mode subset is combined with each fault scenario in the high-probability fault subset to construct a set of normal scenarios, and each operating mode in the low-margin operating mode subset is combined with each fault scenario in the high-risk fault subset to construct a set of extreme scenarios.

[0013] The power system panoramic scene set is composed of the set of conventional scenes and the set of extreme scenes.

[0014] Furthermore, based on wind power data, photovoltaic data, and load data, a clustered set of operating modes is extracted using a trained variational autoencoder, including:

[0015] Historical wind power data, historical photovoltaic data, and historical load data are acquired and preprocessed to obtain historical wind, solar, and load samples.

[0016] Input historical landscape samples into the variational autoencoder for training. Repeat the following steps until the variational autoencoder training converges to obtain a trained variational autoencoder:

[0017] The latent spatial representation is obtained by extracting features from historical landscape samples using an encoder;

[0018] The latent spatial representation is sampled using a decoder to generate wind and solar load power samples;

[0019] Among them, the parameters of the encoder and decoder are updated based on the difference between the wind and solar load power samples and historical wind and solar load data, so that the variational autoencoder learns the statistical characteristics of historical wind and solar load data and reduces the difference between the wind and solar load power samples and historical wind and solar load data.

[0020] Using a trained variational autoencoder model, generate the target number of wind and solar load power samples;

[0021] Clustering algorithms are used to divide the mixed sample set containing historical wind and solar load samples and wind and solar load power samples into several clusters. The center of each cluster is extracted as the typical operating mode, forming a set of clustered operating modes.

[0022] Further, the probability value and safety margin index value of each operating mode in the set of operating modes are calculated, and a target number of operating modes are selected to construct a high-probability operating mode subset and a low-margin operating mode subset, including:

[0023] The probability values ​​of each operating mode are calculated based on the nonparametric kernel density estimation method;

[0024] Based on power flow calculation and line power overload, the safety margin index value of each operating mode is calculated by the overload reverse analysis method.

[0025] Based on the probability values ​​of each operating mode, a target number of operating modes are selected from high to low to construct a subset of high-probability operating modes;

[0026] Based on the safety margin index values ​​of each operating mode, a target number of operating modes are selected from low to high to construct a subset of low-margin operating modes.

[0027] Furthermore, the probability values ​​for each operating mode are calculated based on the nonparametric kernel density estimation method, including:

[0028] Substitute the Gaussian kernel as the kernel function into the kernel density estimation formula to obtain the probability density function for each operating mode;

[0029] The probability value of each operating mode is obtained by integrating the probability density function of each operating mode.

[0030] The probability values ​​of each operating mode are used to characterize the probability of each operating mode occurring in the set of operating modes.

[0031] Furthermore, based on power flow calculations and line power overload, the safety margin index values ​​for each operating mode are calculated using the overload inverse analysis method, including:

[0032] By performing power flow calculations under various operating modes, the voltage and phase angle of each bus in the power grid are determined, and the active power of each line under each operating mode is calculated.

[0033] Based on the active power of each line under each operating mode and the maximum active power that it can withstand, the power overload of each line under each operating mode is determined.

[0034] The power overload of each line is expressed as follows:

[0035] ;

[0036] In the formula, For the first Under the first operating mode, the first Line power overload, For the first Under the first operating mode, the first The active power of the line, For the first Under the first operating mode, the first The maximum active power that the line can withstand;

[0037] Calculate the safety margin index value for each operating mode based on the power overload of each line under each operating mode;

[0038] The safety margin index values ​​for each operating mode are expressed as follows:

[0039] ;

[0040] In the formula, Indicates the first Safety margin index values ​​for each operating mode This represents the set of lines with a power overload greater than zero. Represents the set of all lines. The logical judgment symbol "exists" This indicates taking the maximum value. A larger safety margin value indicates a more stable current system operation. The logical judgment symbol is "arbitrary".

[0041] Furthermore, based on the fault data, an expanded set of fault scenarios is generated using a trained generative adversarial network, including:

[0042] Historical fault data is acquired and preprocessed to obtain historical fault samples;

[0043] Random noise vectors are obtained by random sampling from a standard normal distribution;

[0044] The random noise vector is used as the input to the generator in the generative adversarial network. The following steps are repeated until the generator converges, and a trained generator is obtained.

[0045] Use a generator to generate candidate fault samples;

[0046] Historical fault samples and candidate fault samples are used together as input to the discriminator, and the discriminator outputs the discrimination result.

[0047] The parameters of the discriminator are updated based on the discrimination results, which enhances the discriminator's ability to distinguish between candidate fault samples and historical fault samples.

[0048] The parameters of the generator are updated based on the discrimination results, so that the candidate fault samples generated by the generator can approximate the statistical characteristics of historical fault samples, and reduce the differences between candidate fault samples and historical fault samples in terms of temporal correlation and probability distribution.

[0049] Based on the fault data, a target number of generated fault samples are generated using a trained generator model.

[0050] Fault scenarios are extracted using a hybrid fault sample that includes historical fault samples and generated fault samples, forming an expanded fault scenario set.

[0051] Further, the probability value of failure occurrence and the risk index value of each failure scenario in the failure scenario set are calculated, and a target number of failure scenarios are selected to construct a high-probability failure subset and a high-risk failure subset, including:

[0052] The probability values ​​of fault occurrence for each fault scenario are calculated based on the line load rate and regression analysis.

[0053] The severity of each fault scenario is calculated based on the line power overload, and the risk index value of each fault scenario is calculated based on the fault occurrence probability value and the fault severity value.

[0054] A high-probability fault subset is constructed by selecting a target number of fault scenarios based on their probability of occurrence from high to low.

[0055] A high-risk fault subset is constructed by selecting a target number of fault scenarios based on risk index values ​​from high to low.

[0056] Furthermore, based on line load rate and regression analysis, the probability values ​​of fault occurrence for each fault scenario are calculated, including:

[0057] The load rate of each line under each fault scenario is calculated based on the actual power flow and the rated allowable power.

[0058] The load rate of each line under each fault scenario is expressed as follows:

[0059] ;

[0060] In the formula, Indicates the first In the fault scenario, the first Load rate of the line Indicates the first In the fault scenario, the first The active power of the line, For the first The maximum active power that each line can withstand; based on the load rate of each line under each fault scenario, the probability of each fault scenario occurring is calculated by regression analysis; the regression analysis method includes: determining adjustment parameters by minimizing a preset objective function, so that the error between the probability of each fault scenario occurring calculated based on the adjustment parameters and the actual probability of each fault scenario occurring is minimized; the fault probability of each line is calculated based on the load rate of each line under each fault scenario and the adjustment parameters.

[0061] The preset objective function is expressed as follows:

[0062] ;

[0063] In the formula, This represents the preset objective function. Indicates the first The actual probability of a fault scenario occurring Indicates based on adjustment parameters The calculated probability of each failure scenario occurring, Indicates the total number of failure scenarios;

[0064] The failure probability of each line is expressed as follows:

[0065] ;

[0066] In the formula, Indicates the first In the fault scenario, the first The probability of failure on each line. Indicates the adjustment parameter. This represents an exponential function; for all fault scenarios, the fault probabilities of each line under each fault scenario are multiplied together to obtain the fault occurrence probability value of each fault scenario.

[0067] The probability values ​​of each fault scenario are expressed as follows:

[0068] ;

[0069] In the formula, Indicates the first The probability of failure occurrence for each failure scenario. Indicates the first In a fault scenario, the set of lines that are in a fault state at the initial stage of the system fault state. Indicates the first Under each fault scenario The failure probabilities of each line in the line set are multiplied together.

[0070] Furthermore, the severity of each fault scenario is calculated based on the line power overload, and the risk index value for each fault scenario is calculated based on the fault occurrence probability value and the fault severity value, including:

[0071] For each failure scenario, perform the following steps to obtain the risk indicator value for each failure scenario:

[0072] Calculate the power overload of each line; if the power overload of all lines is not greater than zero, then the maximum value of the power overload of each line is taken as the fault severity of the current fault scenario; if the power overload of any line is greater than zero, then the sum of the power overloads of all lines with power overloads greater than zero is taken as the fault severity of the current fault scenario; multiply the fault occurrence probability value of the current fault scenario by the fault severity value of the current fault scenario to obtain the risk index value of the current fault scenario.

[0073] The severity of each fault scenario is expressed as follows:

[0074] ;

[0075] In the formula, Indicates the first The severity of the fault in each fault scenario;

[0076] The risk index values ​​for each failure scenario are expressed as follows:

[0077] ;

[0078] In the formula, Indicates the first Risk indicator values ​​for each failure scenario.

[0079] Furthermore, using the power flow matching method, each operating mode in the high-probability operating mode subset is combined with each fault scenario in the high-probability fault subset to construct a set of normal scenarios; each operating mode in the low-margin operating mode subset is combined with each fault scenario in the high-risk fault subset to construct a set of extreme scenarios; and based on the set of normal scenarios and the set of extreme scenarios, a panoramic power system scenario set is formed, including:

[0080] Based on the set of normal scenarios and the set of extreme scenarios, the power difference of each line under each operating mode and under each fault scenario is calculated using the power flow matching method.

[0081] Based on the principle of minimizing power differences, fault scenarios are matched for each operating mode;

[0082] Based on the matched operating mode and the corresponding fault scenario, the following two types of combinations are formed:

[0083] The operation mode with a probability value exceeding the first preset threshold is matched with the fault scenario with a fault occurrence probability value exceeding the second preset threshold and defined as a typical scenario.

[0084] The operation mode with a safety margin index value lower than the third preset threshold and the failure scenario with a risk index value exceeding the fourth preset threshold are matched and defined as extreme scenarios.

[0085] According to the set operation rules, the typical scenario set and the extreme scenario set are combined to form a panoramic scenario set of the power system.

[0086] The formula for calculating the power difference is as follows:

[0087] ;

[0088] In the formula, Indicates the first The operating mode and the first Power difference between different fault scenarios Indicates the first The set of lines in the power grid that remain operational under a fault scenario.

[0089] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

[0090] 1. This invention generates a clustered set of operating modes through a variational autoencoder and filters high-probability and low-margin operating mode subsets. It then combines this with an expanded fault scenario set generated by a generative adversarial network to filter high-probability and high-risk fault subsets. Finally, it uses a power flow matching method to construct a set of normal scenarios and a set of extreme scenarios, ultimately forming a panoramic set of power system scenarios. This achieves the collaborative generation of normal and extreme scenarios, ensuring coverage of normal scenarios for high-probability operating modes and high-probability fault scenarios while highlighting the extreme scenario characterization of low-margin operating modes and high-risk fault scenarios. This effectively improves the comprehensiveness and relevance of power system scenario generation, providing more accurate scenario support for system reliability assessment, risk warning, and optimized scheduling.

[0091] 2. This invention generates a clustered set of operating modes based on a variational autoencoder, calculates the probability value of each operating mode using a nonparametric kernel density estimation method, and calculates the safety margin index value using an overload reverse analysis method. It then selects and constructs subsets of high-probability operating modes and subsets of low-margin operating modes, achieving efficient quantification and accurate screening of operating mode characteristics. This provides reliable basic data support for the collaborative generation of normal and extreme scenarios, effectively improving the relevance and reliability of power system scenario generation.

[0092] 3. This invention utilizes generative adversarial networks to generate an expanded set of fault scenarios. It combines line load rate and regression analysis to calculate the probability of fault occurrence, and line power overload to calculate the severity of faults and synthesize risk index values. It then selects and constructs subsets of high-probability faults and subsets of high-risk faults, thereby realizing intelligent expansion of fault scenarios and quantitative risk assessment. This ensures both regular coverage of high-probability fault scenarios and highlights the extreme characterization of high-risk fault scenarios, significantly enhancing the comprehensiveness of the power system's panoramic scenarios and its risk early warning capabilities.

[0093] 4. This invention uses a power flow matching method to combine a subset of high-probability operating modes with a subset of high-probability faults to construct a set of normal scenarios, and a subset of low-margin operating modes with a subset of high-risk faults to construct a set of extreme scenarios. Based on the principle of minimizing power difference, the optimal operating mode and fault scenario are matched, ultimately forming a panoramic set of power system scenarios. This achieves the collaborative optimization generation of normal and extreme scenarios, ensuring accurate matching of typical scenarios and strengthening risk identification of extreme scenarios. It provides more accurate and comprehensive scenario support for system reliability assessment, optimized scheduling, and risk warning. Attached Figure Description

[0094] Figure 1 This is a flowchart illustrating a method for generating a panoramic power system scene that integrates conventional and extreme scenarios, as provided in an embodiment of the present invention. Detailed Implementation

[0095] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present invention and the specific features in the embodiments are detailed descriptions of the technical solution of the present invention, rather than limitations thereof. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.

[0096] Example 1

[0097] like Figure 1 As shown in the figure, this embodiment introduces a method for generating a panoramic power system scene that integrates conventional and extreme scenarios, including:

[0098] Step 1: Obtain wind power data, photovoltaic data, load data, and fault data.

[0099] This invention acquires wind power data, photovoltaic data, load data, and fault data, providing a comprehensive and reliable data foundation for subsequent operation mode extraction and fault scenario generation. It ensures the integrity and accuracy of data support during the analysis process and effectively supports the feasibility verification of subsequent model training and scenario construction.

[0100] Step 2: Based on wind power data, photovoltaic data, and load data, use the trained variational autoencoder to extract the clustered set of operating modes.

[0101] This invention is based on wind power data, photovoltaic data, and load data. It uses a trained variational autoencoder to extract a clustered set of operating modes. Through the latent spatial mapping capability of the variational autoencoder, it achieves efficient compression of data features and pattern recognition, forming a structured operating mode classification system, thereby improving the clustering accuracy and interpretability of operating mode representation.

[0102] Step 3: Calculate the probability value and safety margin index value of each operating mode in the set of operating modes, and select the target number of operating modes to construct a high-probability operating mode subset and a low-margin operating mode subset respectively.

[0103] This invention calculates the probability value and safety margin index value of each operating mode in the set of operating modes, and selects a target number of operating modes to construct a subset of high-probability operating modes and a subset of low-margin operating modes. By quantifying the probability distribution and safety boundary index, it accurately identifies key operating scenarios, provides data-driven decision-making basis for the formulation of system safety operation strategies, and optimizes resource allocation efficiency.

[0104] Step 4: Based on the fault data, use the trained generative adversarial network to generate an expanded set of fault scenarios.

[0105] This invention uses fault data and a trained generative adversarial network to generate an expanded set of fault scenarios. Through the adversarial training mechanism of the generator and discriminator of the generative adversarial network, the invention achieves diversified expansion and high-fidelity simulation of fault scenarios, enhances the breadth and depth of fault scenario coverage, and improves the system's robust prediction ability for unknown fault modes.

[0106] Step 5: Calculate the failure probability value and risk index value of each failure scenario in the failure scenario set, and select a target number of failure scenarios to construct a high-probability failure subset and a high-risk failure subset respectively.

[0107] This invention calculates the probability of fault occurrence based on line load rate and regression analysis, calculates the severity of fault by combining line power overload, and synthesizes risk index values. It selects a target number of fault scenarios to construct a high-probability fault subset and a high-risk fault subset, realizing a dual-indicator quantitative assessment of the probability of fault occurrence and risk level. It accurately identifies high-probability common fault scenarios and high-risk extreme fault scenarios, enhancing the targeting of fault scenario selection and the accuracy of risk warning.

[0108] Step Six: Using the power flow matching method, combine each operating mode in the high-probability operating mode subset with each fault scenario in the high-probability fault subset to construct a set of normal scenarios, and combine each operating mode in the low-margin operating mode subset with each fault scenario in the high-risk fault subset to construct a set of extreme scenarios. Based on the set of normal scenarios and the set of extreme scenarios, a panoramic scene set of the power system is formed.

[0109] This invention uses a power flow matching method to combine a subset of high-probability operating modes with a subset of high-probability faults to construct a set of normal scenarios, and a subset of low-margin operating modes with a subset of high-risk faults to construct a set of extreme scenarios. Based on the principle of minimizing power difference, it matches the optimal operating mode with the fault scenario, and finally forms a set of panoramic scenarios for the power system. This achieves the collaborative optimization and generation of normal and extreme scenarios, which not only ensures the accurate matching of typical operating scenarios, but also highlights the ability to characterize extreme risk scenarios, providing comprehensive and accurate panoramic scenario support for system reliability assessment, optimized scheduling and risk warning.

[0110] Example 2

[0111] Based on the same inventive concept as Embodiment 1, this embodiment introduces the implementation steps of a method for generating panoramic power system scenes that integrates conventional and extreme scenarios, including:

[0112] Step 1: Obtain wind power data, photovoltaic data, load data, and fault data.

[0113] Step 2: Based on wind power data, photovoltaic data, and load data, extract the clustered set of operating modes using a trained variational autoencoder.

[0114] Step 2.1: Obtain historical wind power data, historical photovoltaic data, and historical load data, and preprocess them to obtain historical wind, solar, and load samples.

[0115] Step 2.2: Input the historical wind and light load samples into the variational autoencoder for training. Repeat the following steps until the variational autoencoder training converges to obtain the trained variational autoencoder.

[0116] Step 2.2.1: Use the encoder to extract features from historical wind and light load samples to obtain a latent spatial representation.

[0117] Step 2.2.2: Use the decoder to sample the latent spatial representation to generate wind and solar load power samples.

[0118] Specifically, the parameters of the encoder and decoder are updated based on the difference between the wind and solar load power samples and historical wind and solar load data, so that the variational autoencoder can learn the statistical characteristics of historical wind and solar load data and reduce the difference between the wind and solar load power samples and historical wind and solar load data.

[0119] Step 2.3: Use the trained variational autoencoder model to generate the target number of wind and solar load power samples.

[0120] Step 2.4: Use a clustering algorithm to divide the mixed sample set containing historical wind and solar load samples and wind and solar load power samples into several clusters, extract the center of each cluster as the typical operating mode, and form a set of clustered operating modes.

[0121] Step 3: Calculate the probability value and safety margin index value of each operating mode in the set of operating modes, and select the target number of operating modes to construct a high-probability operating mode subset and a low-margin operating mode subset respectively.

[0122] Step 3.1: Calculate the probability values ​​of each operating mode based on the nonparametric kernel density estimation method.

[0123] Step 3.1.1: Substitute the Gaussian kernel as the kernel function into the kernel density estimation formula to obtain the probability density function of each operating mode.

[0124] Step 3.1.2: Integrate the probability density function of each operating mode to obtain the probability value of each operating mode.

[0125] In this embodiment, the probability value of each operating mode is used to characterize the probability of each operating mode occurring in the set of operating modes.

[0126] Step 3.2: Based on power flow calculation and line power overload, the safety margin index value of each operating mode is calculated by the overload reverse analysis method.

[0127] Step 3.2.1: Determine the voltage and phase angle of each bus in the power grid by performing power flow calculations under each operating mode, and calculate the active power of each line under each operating mode.

[0128] Step 3.2.2: Based on the active power of each line under each operating mode and the maximum active power limit that it can withstand, determine the power overload of each line under each operating mode.

[0129] In this embodiment, the power overload of each line is expressed as follows:

[0130] ;

[0131] In the formula, For the first Under the first operating mode, the first Line power overload, For the first Under the first operating mode, the first The active power of the line, For the first Under the first operating mode, the first The maximum active power that a line can withstand.

[0132] Step 3.2.3: Calculate the safety margin index value of each operating mode based on the power overload of each line under each operating mode.

[0133] In this embodiment, the safety margin index values ​​for each operating mode are expressed as follows:

[0134] ;

[0135] In the formula, Indicates the first Safety margin index values ​​for each operating mode This represents the set of lines with a power overload greater than zero. Represents the set of all lines. The logical judgment symbol "exists" This indicates taking the maximum value. A larger safety margin value indicates a more stable current system operation. The logical judgment symbol is "arbitrary".

[0136] Step 3.3: Select the target number of operating modes from the probability values ​​of each operating mode in descending order to construct a subset of high-probability operating modes.

[0137] Step 3.4: Select the target number of operating modes from the safety margin index values ​​of each operating mode in ascending order to construct a subset of low-margin operating modes.

[0138] Step 4: Based on the fault data, use the trained generative adversarial network to generate an expanded set of fault scenarios.

[0139] Step 4.1: Obtain historical fault data and preprocess it to obtain historical fault samples;

[0140] Step 4.2: Obtain a random noise vector by randomly sampling from the standard normal distribution;

[0141] Step 4.3: Use the random noise vector as input to the generator in the generative adversarial network, and repeat the following steps until the generator converges to obtain a trained generator;

[0142] Step 4.3.1: Generate candidate fault samples using a generator.

[0143] Step 4.3.2: Use historical fault samples and candidate fault samples as input to the discriminator, and use the discriminator to output the discrimination result.

[0144] Step 4.3.3: Update the parameters of the discriminator based on the discrimination results to enhance the discriminator's ability to distinguish between candidate fault samples and historical fault samples.

[0145] Step 4.3.4: Update the parameters of the generator based on the discrimination results, so that the candidate fault samples generated by the generator approximate the statistical characteristics of the historical fault samples, and reduce the differences between the candidate fault samples and the historical fault samples in terms of temporal correlation and probability distribution.

[0146] Step 4.3.5: Based on the fault data, use the trained generator model to generate the target number of generated fault samples.

[0147] Step 4.3.6: Extract fault scenarios using a hybrid fault sample containing historical fault samples and generated fault samples to form an expanded fault scenario set.

[0148] Step 5: Calculate the failure probability value and risk index value of each failure scenario in the failure scenario set, and select a target number of failure scenarios to construct a high-probability failure subset and a high-risk failure subset respectively.

[0149] Step 5.1: Calculate the probability of failure for each fault scenario based on the line load rate and regression analysis.

[0150] Step 5.1.1: Calculate the load rate of each line under each fault scenario based on the actual power flow and the rated allowable power.

[0151] In this embodiment, the load rate of each line under each fault scenario is expressed as follows:

[0152] ;

[0153] In the formula, Indicates the first In the fault scenario, the first Load rate of the line Indicates the first In the fault scenario, the first The active power of the line, For the first Step 5.1.2: Based on the load rate of each line under each fault scenario, calculate the probability of each fault scenario occurring using regression analysis. In this embodiment, the regression analysis method includes: determining adjustment parameters by minimizing a preset objective function, such that the error between the probability of each fault scenario occurring calculated based on the adjustment parameters and the actual probability of each fault scenario occurring is minimized. Step 5.1.3: Based on the load rate of each line under each fault scenario and the adjustment parameters, calculate the fault probability of each line.

[0154] In this embodiment, the preset objective function is expressed as:

[0155] ;

[0156] In the formula, This represents the preset objective function. Indicates the first The actual probability of a fault scenario occurring Indicates based on adjustment parameters The calculated probability of each failure scenario occurring, This indicates the total number of failure scenarios.

[0157] In this embodiment, the failure probability of each line is expressed as follows:

[0158] ;

[0159] In the formula, Indicates the first In the fault scenario, the first The probability of failure on each line. Indicates the adjustment parameter. This represents an exponential function. Step 5.1.4: For all fault scenarios, multiply the fault probabilities of each line under each fault scenario to obtain the fault occurrence probability value of each fault scenario.

[0160] In this embodiment, the probability value of each fault scenario is expressed as follows:

[0161] ;

[0162] In the formula, Indicates the first The probability of failure occurrence for each failure scenario. Indicates the first In a fault scenario, the set of lines that are in a fault state at the initial stage of the system fault state. Indicates the first Under each fault scenario The failure probabilities of each line in the line set are multiplied together.

[0163] Step 5.2: Calculate the severity of each fault scenario based on the line power overload, and calculate the risk index value of each fault scenario based on the fault occurrence probability value and the fault severity value.

[0164] Step 5.2.1: For each failure scenario, perform the following steps to obtain the risk indicator value for each failure scenario:

[0165] Calculate the power overload of each line.

[0166] If the power overload of all lines is not greater than zero, the maximum power overload of each line is taken as the fault severity of the current fault scenario. If the power overload of any line is greater than zero, the sum of the power overloads of all lines with a power overload greater than zero is taken as the fault severity of the current fault scenario. Multiplying the fault occurrence probability value of the current fault scenario by the fault severity value of the current fault scenario yields the risk index value of the current fault scenario.

[0167] In this embodiment, the severity of each fault scenario is expressed as follows:

[0168] ;

[0169] In the formula, Indicates the first The severity of the fault in each fault scenario;

[0170] In this embodiment, the risk index values ​​for each fault scenario are expressed as follows:

[0171] ;

[0172] In the formula, Indicates the first Risk indicator values ​​for each failure scenario.

[0173] Step 5.3: Select the target number of fault scenarios from high to low based on the probability of failure and construct a high-probability fault subset.

[0174] Step 5.4: Select the target number of fault scenarios based on risk index values ​​from high to low to construct a high-risk fault subset.

[0175] Step 6: Using the power flow matching method, combine each operating mode in the high-probability operating mode subset with each fault scenario in the high-probability fault subset to construct a set of normal scenarios, combine each operating mode in the low-margin operating mode subset with each fault scenario in the high-risk fault subset to construct a set of extreme scenarios, and form a set of panoramic power system scenarios based on the set of normal scenarios and the set of extreme scenarios.

[0176] Step 6.1: Based on the set of normal scenarios and the set of extreme scenarios, use the power flow matching method to calculate the power difference of each line under each operating mode and under each fault scenario.

[0177] Step 6.2: Match fault scenarios for each operating mode based on the principle of minimizing power difference.

[0178] Step 6.3: Based on the matched operating mode and the corresponding fault scenario, the following two combinations are formed.

[0179] Step 6.4: Match the operating mode with the probability value exceeding the first preset threshold with the fault scenario with the fault occurrence probability value exceeding the second preset threshold and define them as typical scenarios.

[0180] Step 6.5: Match the operating mode with the safety margin index value below the preset third preset threshold with the failure scenario with the risk index value exceeding the preset fourth preset threshold and define it as an extreme scenario.

[0181] Step 6.6: Perform a union operation on the typical scenario set and the extreme scenario set according to the set operation rules to form a panoramic scenario set of the power system.

[0182] In this embodiment, the formula for calculating the power difference is expressed as follows:

[0183] ;

[0184] In the formula, Indicates the first The operating mode and the first Power difference between different fault scenarios Indicates the first The set of lines in the power grid that remain operational under a fault scenario.

[0185] Example 3

[0186] Based on the same inventive concept as other embodiments, this embodiment describes a computer-readable storage medium having computer instructions stored thereon, which, when executed by a processor, implement the steps of the methods of Embodiment 1 or 2 described above.

[0187] Example 4

[0188] Based on the same inventive concept as other embodiments, this embodiment introduces a computer program product, including computer instructions that, when executed by a processor, implement the steps of the methods described in Embodiment 1 or 2 above.

[0189] In summary, this invention generates a clustered set of operating modes through a variational autoencoder and filters high-probability and low-margin operating mode subsets. It then combines this with an expanded fault scenario set generated by a generative adversarial network to filter high-probability and high-risk fault subsets. Finally, it uses a power flow matching method to construct a set of conventional and extreme scenarios, ultimately forming a panoramic set of power system scenarios. This achieves the collaborative generation of conventional and extreme scenarios, ensuring coverage of high-probability operating modes and high-probability fault scenarios while highlighting the extreme scenarios of low-margin and high-risk fault scenarios. This effectively improves the comprehensiveness and relevance of power system scenario generation, providing more accurate scenario support for system reliability assessment, risk warning, and optimized scheduling.

[0190] This invention generates a clustered set of operating modes based on a variational autoencoder, calculates the probability value of each operating mode using a nonparametric kernel density estimation method, and calculates the safety margin index value using an overload reverse analysis method. It then selects and constructs subsets of high-probability operating modes and subsets of low-margin operating modes, achieving efficient quantification and accurate screening of operating mode characteristics. This provides reliable basic data support for the collaborative generation of normal and extreme scenarios, effectively improving the relevance and reliability of power system scenario generation.

[0191] This invention utilizes generative adversarial networks to generate an expanded set of fault scenarios. It combines line load rate and regression analysis to calculate the probability of fault occurrence, and line power overload to calculate the severity of faults and synthesize risk index values. It then selects and constructs subsets of high-probability faults and subsets of high-risk faults, thereby realizing intelligent expansion of fault scenarios and quantitative risk assessment. This ensures both regular coverage of high-probability fault scenarios and highlights the extreme characterization of high-risk fault scenarios, significantly enhancing the comprehensiveness of the power system's panoramic scenario and its risk early warning capabilities.

[0192] This invention uses a power flow matching method to combine a subset of high-probability operating modes with a subset of high-probability faults to construct a set of normal scenarios, and a subset of low-margin operating modes with a subset of high-risk faults to construct a set of extreme scenarios. Based on the principle of minimizing power difference, it matches the optimal operating mode with the fault scenario, ultimately forming a panoramic set of power system scenarios. This achieves the collaborative optimization and generation of normal and extreme scenarios, ensuring accurate matching of typical scenarios and strengthening risk identification of extreme scenarios. It provides more accurate and comprehensive scenario support for system reliability assessment, optimized scheduling, and risk warning.

[0193] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0194] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0195] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0196] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0197] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims. All of these forms are within the protection scope of the present invention.

Claims

1. A method for generating panoramic power system scenes that integrates conventional and extreme scenarios, characterized in that, include: Acquire wind power data, photovoltaic data, load data, and fault data; Based on wind power data, photovoltaic data, and load data, a clustered set of operating modes is extracted using a trained variational autoencoder. Calculate the probability value and safety margin index value of each operating mode in the set of operating modes, and select a target number of operating modes to construct a high-probability operating mode subset and a low-margin operating mode subset respectively; Based on fault data, an expanded set of fault scenarios is generated using a trained generative adversarial network. Calculate the failure probability value and risk index value of each failure scenario in the failure scenario set, and select a target number of failure scenarios to construct a high-probability failure subset and a high-risk failure subset respectively; By using the power flow matching method, each operating mode in the high-probability operating mode subset is combined with each fault scenario in the high-probability fault subset to construct a set of normal scenarios, and each operating mode in the low-margin operating mode subset is combined with each fault scenario in the high-risk fault subset to construct a set of extreme scenarios. The power system panoramic scene set is composed of the set of conventional scenes and the set of extreme scenes.

2. The method for generating a panoramic power system scene integrating conventional and extreme scenarios according to claim 1, characterized in that, Based on wind power data, photovoltaic data, and load data, a clustered set of operating modes is extracted using a trained variational autoencoder, including: Historical wind power data, historical photovoltaic data, and historical load data are acquired and preprocessed to obtain historical wind, solar, and load samples. Input historical landscape samples into the variational autoencoder for training. Repeat the following steps until the variational autoencoder training converges to obtain a trained variational autoencoder: The latent spatial representation is obtained by extracting features from historical landscape samples using an encoder; The latent spatial representation is sampled using a decoder to generate wind and solar load power samples; Among them, the parameters of the encoder and decoder are updated based on the difference between the wind and solar load power samples and historical wind and solar load data, so that the variational autoencoder learns the statistical characteristics of historical wind and solar load data and reduces the difference between the wind and solar load power samples and historical wind and solar load data. Using a trained variational autoencoder model, generate the target number of wind and solar load power samples; Clustering algorithms are used to divide the mixed sample set containing historical wind and solar load samples and wind and solar load power samples into several clusters. The center of each cluster is extracted as the typical operating mode, forming a set of clustered operating modes.

3. The method for generating a panoramic power system scene integrating conventional and extreme scenarios according to claim 1, characterized in that, Calculate the probability value and safety margin index value of each operating mode in the set of operating modes, and select a target number of operating modes to construct a high-probability operating mode subset and a low-margin operating mode subset, including: The probability values ​​of each operating mode are calculated based on the nonparametric kernel density estimation method; Based on power flow calculation and line power overload, the safety margin index value of each operating mode is calculated by the overload reverse analysis method. Based on the probability values ​​of each operating mode, a target number of operating modes are selected from high to low to construct a subset of high-probability operating modes; Based on the safety margin index values ​​of each operating mode, a target number of operating modes are selected from low to high to construct a subset of low-margin operating modes.

4. The method for generating a panoramic power system scene integrating conventional and extreme scenarios according to claim 3, characterized in that, The probability values ​​for each operating mode are calculated based on the nonparametric kernel density estimation method, including: Substitute the Gaussian kernel as the kernel function into the kernel density estimation formula to obtain the probability density function for each operating mode; The probability value of each operating mode is obtained by integrating the probability density function of each operating mode. The probability values ​​of each operating mode are used to characterize the probability of each operating mode occurring in the set of operating modes.

5. The method for generating a panoramic power system scene integrating conventional and extreme scenarios according to claim 3, characterized in that, Based on power flow calculations and line power overload, the safety margin index values ​​for each operating mode are calculated using the overload inverse analysis method, including: By performing power flow calculations under various operating modes, the voltage and phase angle of each bus in the power grid are determined, and the active power of each line under each operating mode is calculated. Based on the active power of each line under each operating mode and the maximum active power that it can withstand, the power overload of each line under each operating mode is determined. The power overload of each line is expressed as follows: ； In the formula, For the first Under the first operating mode, the first Line power overload, For the first Under the first operating mode, the first The active power of the line, For the first Under the first operating mode, the first The maximum active power that the line can withstand; Calculate the safety margin index value for each operating mode based on the power overload of each line under each operating mode; The safety margin index values ​​for each operating mode are expressed as follows: ； In the formula, Indicates the first Safety margin index values ​​for each operating mode This represents the set of lines with a power overload greater than zero. Represents the set of all lines. The logical judgment symbol "exists" is represented by... This indicates taking the maximum value. A larger safety margin value indicates a more stable current system operation. The logical judgment symbol is "arbitrary".

6. The method for generating a panoramic power system scene integrating conventional and extreme scenarios according to claim 1, characterized in that, Based on fault data, an expanded set of fault scenarios is generated using a trained generative adversarial network, including: Historical fault data is acquired and preprocessed to obtain historical fault samples. Random sampling is then performed from a standard normal distribution to obtain a random noise vector. The random noise vector is used as the input to the generator in the generative adversarial network. The following steps are repeated until the generator converges, and a trained generator is obtained. Use a generator to generate candidate fault samples; Historical fault samples and candidate fault samples are used together as input to the discriminator, and the discriminator outputs the discrimination result. The parameters of the discriminator are updated based on the discrimination results, which enhances the discriminator's ability to distinguish between candidate fault samples and historical fault samples. The parameters of the generator are updated based on the discrimination results, so that the candidate fault samples generated by the generator can approximate the statistical characteristics of historical fault samples, and reduce the differences between candidate fault samples and historical fault samples in terms of temporal correlation and probability distribution. Based on the fault data, a target number of generated fault samples are generated using a trained generator model. Fault scenarios are extracted using a hybrid fault sample that includes historical fault samples and generated fault samples, forming an expanded fault scenario set.

7. The method for generating a panoramic power system scene integrating conventional and extreme scenarios according to claim 1, characterized in that, Calculate the failure probability value and risk index value for each failure scenario in the failure scenario set, and select a target number of failure scenarios to construct a high-probability failure subset and a high-risk failure subset, including: The probability values ​​of fault occurrence for each fault scenario are calculated based on the line load rate and regression analysis. The severity of each fault scenario is calculated based on the line power overload, and the risk index value of each fault scenario is calculated based on the fault occurrence probability value and the fault severity value. A high-probability fault subset is constructed by selecting a target number of fault scenarios based on their probability of occurrence from high to low. A high-risk fault subset is constructed by selecting a target number of fault scenarios based on risk index values ​​from high to low.

8. The method for generating a panoramic power system scene integrating conventional and extreme scenarios according to claim 7, characterized in that, The probability values ​​of fault occurrence for each fault scenario are calculated based on line load rate and regression analysis, including: The load rate of each line under each fault scenario is calculated based on the actual power flow and the rated allowable power. The load rate of each line under each fault scenario is expressed as follows: ； In the formula, Indicates the first In the fault scenario, the first Load rate of the line Indicates the first In the fault scenario, the first The active power of the line, For the first The maximum active power that each line can withstand; based on the load rate of each line under each fault scenario, the probability of each fault scenario occurring is calculated by regression analysis; the regression analysis method includes: determining adjustment parameters by minimizing a preset objective function, so that the error between the probability of each fault scenario occurring calculated based on the adjustment parameters and the actual probability of each fault scenario occurring is minimized; the fault probability of each line is calculated based on the load rate of each line under each fault scenario and the adjustment parameters. The preset objective function is expressed as follows: ； In the formula, This represents the preset objective function. Indicates the first The actual probability of a fault scenario occurring Indicates based on adjustment parameters The calculated probability of each failure scenario occurring, Indicates the total number of failure scenarios; The failure probability of each line is expressed as follows: ； In the formula, Indicates the first In the fault scenario, the first The probability of failure on each line. Indicates the adjustment parameter. This represents an exponential function; for all fault scenarios, the fault probabilities of each line under each fault scenario are multiplied together to obtain the fault occurrence probability value of each fault scenario. The probability values ​​of each fault scenario are expressed as follows: ； In the formula, Indicates the first The probability value of failure occurrence for each failure scenario. Indicates the first In a fault scenario, the set of lines that are in a fault state at the initial stage of the system fault state. Indicates the first Under each fault scenario The failure probabilities of each line in the line set are multiplied together.

9. The method for generating a panoramic power system scene integrating conventional and extreme scenarios according to claim 7, characterized in that, The severity of each fault scenario is calculated based on the line power overload. Risk index values ​​for each fault scenario are calculated based on the fault occurrence probability and fault severity, including: For each failure scenario, perform the following steps to obtain the risk indicator value for each failure scenario: Calculate the power overload of each line; if the power overload of all lines is not greater than zero, then the maximum value of the power overload of each line is taken as the fault severity of the current fault scenario; if the power overload of any line is greater than zero, then the sum of the power overloads of all lines with power overloads greater than zero is taken as the fault severity of the current fault scenario; multiply the fault occurrence probability value of the current fault scenario by the fault severity value of the current fault scenario to obtain the risk index value of the current fault scenario. The severity of each fault scenario is expressed as follows: ； In the formula, Indicates the first The severity of the fault in each fault scenario; The risk index values ​​for each failure scenario are expressed as follows: ； In the formula, Indicates the first Risk indicator values ​​for each failure scenario.

10. The method for generating a panoramic power system scene integrating conventional and extreme scenarios according to claim 7, characterized in that, Using a power flow matching method, each operating mode in the high-probability operating mode subset is combined with each fault scenario in the high-probability fault subset to construct a set of normal scenarios. Each operating mode in the low-margin operating mode subset is combined with each fault scenario in the high-risk fault subset to construct a set of extreme scenarios. Based on the set of normal scenarios and the set of extreme scenarios, a panoramic power system scenario set is formed, including: Based on the set of normal scenarios and the set of extreme scenarios, the power difference of each line under each operating mode and under each fault scenario is calculated using the power flow matching method. Based on the principle of minimizing power differences, fault scenarios are matched for each operating mode; Based on the matched operating mode and the corresponding fault scenario, the following two types of combinations are formed: The operation mode with a probability value exceeding the first preset threshold is matched with the fault scenario with a fault occurrence probability value exceeding the second preset threshold and defined as a typical scenario. The operation mode with a safety margin index value lower than the third preset threshold and the failure scenario with a risk index value exceeding the fourth preset threshold are matched and defined as extreme scenarios. According to the set operation rules, the typical scenario set and the extreme scenario set are combined to form a panoramic scenario set of the power system. The formula for calculating the power difference is as follows: ； In the formula, Indicates the first The operating mode and the first Power difference between different fault scenarios Indicates the first The set of lines in the power grid that remain operational under a fault scenario.

Images