A method and apparatus for modeling and configuring power system operation scenarios based on the spatiotemporal characteristics of typhoons.

By constructing a power system operation scenario model based on the spatiotemporal characteristics of typhoons, the shortcomings of existing technologies in addressing the long-term impact of typhoons on power systems have been overcome. This has enabled more accurate typhoon path reconfiguration and load forecasting, thereby improving the power system's disaster resistance and operational efficiency under extreme weather conditions.

CN120874568BActive Publication Date: 2026-11-10GUANGDONG POWER GRID CO LTD
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202510990769.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2025-07-18
Publication Date
2026-11-10
Estimated Expiration
2045-07-18

AI Technical Summary

Technical Problem

Existing technologies cannot accurately and comprehensively address the long-term impact of typhoons on power systems, especially in terms of wind farm output characteristics and power system operation planning. They also lack systematic analysis of the uncertainty of typhoon paths and the evolution characteristics of meteorological stages.

Method used

By acquiring historical meteorological data, historical load data, and historical typhoon data of the power system, a meteorological feature matrix and a typhoon path feature matrix are constructed. A neural network is used to fit the mapping relationship between meteorological parameters and load. A random typhoon path is generated by combining a multivariate joint probability distribution model and a Monte Carlo simulation algorithm. The typhoon path information for the planning year is reconstructed. Finally, a power system operation scenario model is constructed by combining a multi-component coupling model of the typhoon wind field.

Benefits of technology

It has improved the disaster resistance and operational efficiency of the power system under extreme weather conditions such as typhoons, ensured the continuity and reliability of power supply, and provided a scientific basis for operation and planning.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120874568B_ABST
    Figure CN120874568B_ABST
Patent Text Reader

Abstract

This invention discloses a method and apparatus for modeling and configuring power system operation scenarios based on the spatiotemporal characteristics of typhoons. This application acquires historical power system data to construct meteorological and typhoon path feature matrices and load datasets. Based on the three stages of typhoon passage (before, during, and after), a pre-set neural network is used to fit the mapping relationship between meteorological parameters and load, generating a fitted curve. A multivariate joint probability distribution model and Monte Carlo simulation algorithm are combined to generate random typhoon paths and reconstruct typhoon path information for the planned year. Based on the fitted curves, typhoon path information, and historical meteorological data, monthly hourly load data and meteorological data for the planned year's typhoon are generated. Finally, combined with a pre-set multi-component coupling model of the typhoon wind field, multiple power system operation scenario models based on the spatiotemporal characteristics of typhoons are constructed, and the power supply of the power system is controlled accordingly. This effectively solves the problem that existing technologies cannot accurately and comprehensively address the long-term impact of typhoons on power systems.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of power systems, and in particular to a method and apparatus for modeling and configuring power system operation scenarios based on the spatiotemporal characteristics of typhoons. Background Technology

[0002] With the intensification of global climate change, the frequency and intensity of extreme weather events have increased significantly. Typhoons, as a high-impact, low-frequency natural disaster, pose a serious threat to infrastructure and energy systems in coastal areas. Especially with the large-scale grid integration of renewable energy sources such as wind power, wind farms are typically located along the coast or offshore, areas with a high probability of being affected by typhoons. Typhoons present severe challenges to the power output characteristics of wind farms and the operation and planning of power systems. Therefore, studying the impact of typhoons on wind farm output and its influence on power system operation and planning has significant practical implications.

[0003] Typhoons exhibit complex spatiotemporal characteristics, with significant uncertainties in parameters such as their path, intensity, and speed. This presents substantial challenges to wind farm output calculations and power system operation and planning. The uncertainty of typhoon paths not only affects short-term wind farm output but can also lead to power and load fluctuations in the power system during typhoon passage. Furthermore, changes in meteorological conditions during a typhoon can significantly impact wind farm output and power system load. Therefore, accurately simulating typhoon paths, analyzing wind farm characteristics, and reconstructing load conditions during typhoon passage are crucial foundations for power system operation and planning.

[0004] In recent years, scholars both domestically and internationally have conducted extensive research on typhoon track modeling, wind field characteristic analysis, and load reconfiguration during typhoon passage, achieving certain results. However, existing research still has the following shortcomings:

[0005] Typhoon track modeling: It is mostly based on statistical analysis of historical data, lacks a systematic reconstruction method for the evolution characteristics of typhoon tracks and related parameters in the planning year, and is difficult to accurately reflect the uncertainty of typhoon tracks.

[0006] Historical load data reconstruction: mostly focused on short-term impact analysis, lacking systematic analysis and fitting models of the evolution characteristics of meteorological stages within a typhoon month.

[0007] The coupling relationship between wind farms and typhoon paths: mainly focuses on radial wind speed, lacking analysis of the combined impact of typhoon radial wind speed and moving wind speed.

[0008] These shortcomings are concentrated in the simulation or short-term impact analysis of a single typhoon event, lacking systematic analysis of the evolution characteristics of the meteorological stages within a typhoon month, and the reconstruction model of typhoon paths and historical load data, which makes it impossible for existing technologies to accurately and comprehensively address the long-term impact of typhoons on the power system. Summary of the Invention

[0009] This invention provides a method and apparatus for modeling and configuring power system operation scenarios based on the spatiotemporal characteristics of typhoons, in order to solve the problem that existing technologies cannot accurately and comprehensively address the long-term impact of typhoons on power systems.

[0010] Firstly, this application provides a method for modeling power system operation scenarios based on the spatiotemporal characteristics of typhoons, including:

[0011] Acquire historical meteorological data, historical load data, and historical typhoon data for the power system;

[0012] Based on the historical meteorological data, historical load data, and historical typhoon data, a meteorological feature matrix, a typhoon path feature matrix, and a load dataset are constructed.

[0013] Based on the meteorological feature matrix and load dataset divided into three stages before, during and after the typhoon, a pre-set neural network is used to fit the mapping relationship between meteorological parameters and load to generate a meteorological-load fitting curve.

[0014] Based on the typhoon path feature matrix, random typhoon paths are generated by combining a multivariate joint probability distribution model and a Monte Carlo simulation algorithm, and the typhoon path information for the planning year is reconstructed.

[0015] Based on the meteorological-load fitting curve, typhoon track information and historical meteorological data, the monthly hourly typhoon load data and monthly hourly typhoon meteorological data for the planning year are generated.

[0016] Based on the typhoon path information, monthly meteorological data and monthly load data, and combined with the preset multi-component coupling model of the typhoon wind field, multiple power system operation scenario models based on the spatiotemporal characteristics of typhoons are constructed.

[0017] This application constructs meteorological feature matrices, typhoon path feature matrices, and load datasets by acquiring historical meteorological, load, and typhoon data from the power system, providing fundamental data support for subsequent modeling. The meteorological feature matrices and load datasets are divided into three stages: before, during, and after a typhoon's passage. A neural network is used to fit the mapping relationship between meteorological parameters and load, generating a meteorological-load fitting curve. This allows the model to more accurately reflect the impact of different typhoon stages on the load. A multivariate joint probability distribution model and Monte Carlo simulation algorithm are combined to generate random typhoon paths, reconstructing typhoon path information for the planning year and effectively capturing the uncertainty of typhoon paths. Based on the reconstructed typhoon path information and the meteorological-load fitting curve, monthly hourly load and meteorological data for the planning year are generated, providing a detailed data foundation for constructing power system operation scenarios. Finally, combined with a pre-defined multi-component coupled typhoon wind field model, multiple power system operation scenario models based on the spatiotemporal characteristics of typhoons are constructed. These models comprehensively reflect the impact of typhoons on the power system, providing a scientific basis for power system operation and planning, thereby significantly improving the power system's disaster resistance and operational efficiency under extreme weather conditions such as typhoons. This application effectively solves the problem that existing technologies cannot accurately and comprehensively address the long-term impact of typhoons on power systems.

[0018] As a preferred embodiment of the first aspect, the step of generating random typhoon paths based on the typhoon path feature matrix, combined with a multivariate joint probability distribution model and a Monte Carlo simulation algorithm, and reconstructing the typhoon path information for the planning year specifically involves:

[0019] The central pressure, moving speed and direction in the typhoon path feature matrix are discretized to divide multiple state intervals, and a multivariate joint state transition probability matrix is ​​constructed by combining the historical typhoon data.

[0020] Based on the multivariate joint state transition probability matrix, a random typhoon path state sequence is generated using the Monte Carlo simulation algorithm. The random typhoon path state sequence contains the hourly state code of the typhoon.

[0021] The random typhoon path state sequence is decoded, and the state code is converted into continuous spatiotemporal parameters to generate typhoon path information for the planned year.

[0022] The typhoon path information for the planned year is specifically as follows:

[0023] The typhoon's path information is as follows:

[0024] X = X n,min +u(X n,max -X n,min )

[0025] In the formula, X represents the simulated typhoon path information at a certain moment. n,max X n,min These are the upper and lower limits of the typhoon path range covered by state n, respectively, and u is a preset value.

[0026] In this preferred embodiment, the central pressure, speed, and direction of the typhoon path feature matrix are discretized and divided into multiple state intervals. A multivariate joint state transition probability matrix is ​​constructed by combining historical typhoon data. Based on this probability matrix, a Monte Carlo simulation algorithm is used to generate a random typhoon path state sequence containing hourly state codes. Subsequently, these state sequences are decoded, converting the discrete state codes into continuous spatiotemporal parameters, thereby generating typhoon path information for the planning year. This process not only effectively captures the uncertainty of typhoon paths but also improves the accuracy and reliability of path reconstruction through discretization and stochastic simulation. Ultimately, the generated typhoon path information provides more accurate data support for the operation and planning of the power system, significantly enhancing the power system's disaster resistance and operational efficiency in the face of extreme weather such as typhoons.

[0027] As a preferred embodiment of the first aspect, the step of generating monthly hourly typhoon load data and monthly hourly typhoon meteorological data for the planning year based on the meteorological-load fitting curve, typhoon track information, and historical meteorological data specifically includes:

[0028] Based on the typhoon path information, the actual number of days the typhoon affected the typhoon was counted, and the actual number of days the typhoon affected the typhoon was divided into three stages: the pre-typhoon, mid-typhoon, and post-typhoon durations.

[0029] Hourly meteorological data for the period before and after the typhoon are extracted from the historical meteorological data, and the average value of the hourly meteorological data is taken to generate monthly hourly meteorological data for the typhoon before and after the typhoon.

[0030] If the duration of the typhoon phase is less than a preset threshold, the monthly hourly meteorological data of the typhoon in the historical meteorological data is recorded sequentially; otherwise, the combination of the meteorological data before and after the typhoon in the historical meteorological data is extracted, and the monthly hourly meteorological data of the typhoon in the typhoon is generated according to the preset mean calculation method.

[0031] Based on the monthly hourly meteorological data of typhoons before and after the typhoon, the monthly hourly meteorological data of typhoons during the typhoon, and the meteorological-load fitting curve, and by regressing and fitting the planned annual peak based on the maximum value of the annual historical load data, the monthly hourly load data of typhoons is generated.

[0032] In this preferred embodiment, this application accurately locates the time range of each stage of the typhoon by statistically analyzing the actual number of days the typhoon affects the area and dividing the duration into three stages: before, during, and after the typhoon. Furthermore, hourly meteorological data before and after the typhoon are extracted from historical meteorological data and averaged to generate hourly meteorological data before and after the typhoon. This process effectively integrates historical data, providing a stable data foundation for subsequent analysis. For the mid-typhoon stage, if the duration is less than a preset threshold, the mid-typhoon meteorological data from historical data is directly recorded; if the duration is greater than or equal to the threshold, the mid-typhoon meteorological data is generated by extracting combined data before and after the typhoon from historical data and calculating the average. This flexible processing method ensures the integrity and accuracy of the data. Finally, by combining the hourly meteorological data before, during, and after the typhoon, as well as the meteorological-load fitting curve, and by regressing the planned annual peak value based on the maximum value of the annual historical load data, monthly hourly load data for the typhoon is generated. This series of steps not only improves the accuracy of load data forecasting, but also provides more accurate and reliable meteorological and load data support for the operation and planning of the power system, thereby enhancing the power system's disaster resistance and operational efficiency under extreme weather conditions such as typhoons.

[0033] As a preferred embodiment of the first aspect, the step of constructing multiple power system operation scenario models based on the typhoon's spatiotemporal characteristics, using the typhoon path information, monthly hourly meteorological data of the typhoon, and monthly hourly load data of the typhoon, combined with a preset multi-component coupling model of the typhoon wind field, specifically involves:

[0034] Based on the center location, speed and direction of movement of the typhoon path information, and combined with the multi-component coupling model of the typhoon wind field, the wind speed in the typhoon-affected area is decomposed into radial and moving components, and the actual hourly wind speed of the wind farm is calculated by vector superposition.

[0035] Using the typhoon center as the center and the preset influence radius as the threshold, the geometric circle determination method is used to determine whether the wind farm is within the typhoon's influence range. If it is within the range, it is marked as a typhoon-affected scenario.

[0036] Based on the actual hourly wind speed, and combined with the preset wind turbine cut-in wind speed, cut-out wind speed and rated power characteristic curve, the hourly output data of the typhoon-affected scenario is calculated in segments.

[0037] By aligning the monthly hourly meteorological data, monthly hourly load data, and hourly power output data of typhoons in time and space, multiple operational scenario models are generated, which include typhoon path intensity, meteorological fluctuations, and load changes.

[0038] The calculation formula for the hourly power output data of the typhoon-affected scenario is as follows:

[0039]

[0040] In the formula, Vci V represents the cut-in wind speed. co V represents the cut-out wind speed. r P represents the rated wind speed. r P represents the active power generated by the wind turbine at rated wind speed. t Let V be the output power of the wind farm at time t. t Let t be the wind speed at time t, and A, B, and C be the parameters of the wind turbine power characteristic curve.

[0041] In this preferred embodiment, this application combines the center location, moving speed, and direction from the typhoon path information, and utilizes a multi-component coupling model of the typhoon wind field to decompose the wind speed within the typhoon's influence area into radial and moving components. The actual hourly wind speed of the wind farm is then calculated through vector superposition. This process accurately simulates the impact of the typhoon on the wind speed of the wind farm, providing accurate wind speed data for subsequent wind farm output calculations. Furthermore, using the typhoon center as the center and a preset influence radius as a threshold, a geometric circle determination method is employed to determine whether the wind farm is within the typhoon's influence range, thereby filtering out scenarios affected by the typhoon. For these scenarios, hourly output data is calculated segmentally based on the actual hourly wind speed and the characteristic curves of the wind turbines. This step accurately assesses the impact of the typhoon on the wind farm's power generation capacity. Finally, the monthly hourly meteorological data, load data, and wind farm output data are aligned spatiotemporally to generate multiple operational scenario models that include typhoon path intensity, meteorological fluctuations, and load changes. These models comprehensively reflect the impact of typhoons on the power system, providing a scientific basis for the operation and planning of the power system. They significantly improve the disaster resistance and operational efficiency of the power system under extreme weather conditions such as typhoons, and have important scientific significance and practical application value.

[0042] As a preferred embodiment of the first aspect, the acquisition of historical meteorological data and historical typhoon data of the power system specifically includes:

[0043] The historical meteorological data includes hourly data on temperature, precipitation, wind speed, and humidity.

[0044] The historical typhoon data includes latitude and longitude, central pressure, speed of movement, and direction of movement.

[0045] In this preferred embodiment, this application acquires historical meteorological and typhoon data of the power system, providing a comprehensive and detailed data foundation for subsequent modeling and analysis. Specifically, the historical meteorological data covers hourly data on temperature, precipitation, wind speed, and humidity. This data accurately reflects the dynamic changes in meteorological conditions, providing the necessary meteorological background for the operation and planning of the power system. Simultaneously, the historical typhoon data includes key parameters such as latitude and longitude, central pressure, movement speed, and direction of movement. These parameters are crucial for understanding and predicting the path and intensity of typhoons. Combining this data, this application can more accurately simulate the impact of typhoons on the power system, thereby providing a scientific basis for the disaster-resistant design and operational optimization of the power system, and enhancing the reliability and stability of the power system under extreme weather conditions.

[0046] Secondly, this application also provides a power supply configuration method based on the spatiotemporal characteristics of typhoons, including:

[0047] Based on the power system operation scenario modeling method based on the spatiotemporal characteristics of typhoons, multiple first power system operation scenario models of the power system to be controlled are constructed.

[0048] Based on the multiple first power system operation scenario models, assess the load reduction risk of the power system under typhoon power supply in each scenario. If the load reduction is non-zero, iteratively correct the first power system operation scenario model by adjusting the generator capacity or optimizing the generator layout.

[0049] Based on the revised first power system operation scenario model, output a power planning scheme that meets the power supply requirements of multiple climates during the typhoon month;

[0050] Based on the power supply planning schemes for the various climate power supply requirements, the power supply of the power system is configured.

[0051] This application employs a power system operation scenario modeling method based on the spatiotemporal characteristics of typhoons. First, it constructs multiple first-stage power system operation scenario models, which comprehensively reflect the potential impact of typhoons on power system operation. Based on these models, the load reduction risk of the power system under each scenario is assessed. If the load reduction is found to be non-zero, indicating a risk to power supply, the model is iteratively corrected by adjusting generator capacity or optimizing generator layout until the load reduction is zero, ensuring the reliability of power supply under typhoon conditions. Subsequently, based on the corrected model, a power supply planning scheme that meets the power supply requirements of multiple climates during the typhoon month is output, providing scientific guidance for power system power configuration. Finally, based on these planning schemes, the power supply of the power system is precisely controlled, thereby effectively improving the disaster resistance and operational stability of the power system under extreme weather conditions such as typhoons, ensuring the continuity and reliability of power supply.

[0052] Thirdly, this application provides a power system operation scenario modeling device based on the spatiotemporal characteristics of typhoons. The power system operation scenario modeling device based on the spatiotemporal characteristics of typhoons includes an acquisition module, a construction module, a fitting module, a data reconstruction module, and a modeling module;

[0053] The acquisition module is used to acquire historical meteorological data, historical load data, and historical typhoon data of the power system;

[0054] The construction module is used to construct a meteorological feature matrix, a typhoon path feature matrix, and a load dataset based on the historical meteorological data, historical load data, and historical typhoon data.

[0055] The fitting module is used to divide the meteorological feature matrix and load dataset into three stages: before, during and after the typhoon, and to use a preset neural network to fit the mapping relationship between meteorological parameters and load to generate a meteorological-load fitting curve.

[0056] The data reconstruction module is used to generate random typhoon paths based on the typhoon path feature matrix, combined with a multivariate joint probability distribution model and a Monte Carlo simulation algorithm, and to reconstruct the typhoon path information for the planning year.

[0057] Based on the meteorological-load fitting curve, typhoon track information and historical meteorological data, the monthly hourly typhoon load data and monthly hourly typhoon meteorological data for the planning year are generated.

[0058] The modeling module is used to construct multiple power system operation scenario models based on the spatiotemporal characteristics of typhoons, based on the typhoon path information, monthly meteorological data of typhoons, and monthly load data of typhoons, combined with a preset multi-component coupling model of typhoon wind fields.

[0059] This device utilizes five modules that work in a coordinated manner to more accurately construct power system operation scenarios. This application acquires historical meteorological data, historical load data, and historical typhoon data from the power system to construct a meteorological feature matrix, a typhoon path feature matrix, and a load dataset, providing fundamental data support for subsequent modeling. The meteorological feature matrix and load dataset are divided into three stages: before, during, and after a typhoon's passage. A neural network is used to fit the mapping relationship between meteorological parameters and load, generating a meteorological-load fitting curve. This allows the model to more accurately reflect the impact of different typhoon stages on the load. A multivariate joint probability distribution model and Monte Carlo simulation algorithm are combined to generate random typhoon paths, reconstructing typhoon path information for the planning year and effectively capturing the uncertainty of typhoon paths. Based on the reconstructed typhoon path information and the meteorological-load fitting curve, monthly hourly load data and meteorological data for the planning year are generated, providing a detailed data foundation for constructing power system operation scenarios. Finally, by combining a pre-defined multi-component coupled model of typhoon wind fields, multiple power system operation scenario models based on the spatiotemporal characteristics of typhoons are constructed. These models can comprehensively reflect the impact of typhoons on the power system, providing a scientific basis for the operation and planning of the power system, thereby significantly improving the disaster resistance and operational efficiency of the power system under extreme weather conditions such as typhoons. This application effectively solves the problem that existing technologies cannot accurately and comprehensively address the long-term impact of typhoons on power systems.

[0060] Fourthly, this application also provides a power supply configuration device based on the spatiotemporal characteristics of typhoons, including a model building module, a model correction module, a model output module, and a model configuration module:

[0061] The model building module is used to construct multiple first power system operation scenario models of the power system to be controlled, based on the aforementioned power system operation scenario modeling method based on the spatiotemporal characteristics of typhoons.

[0062] The model correction module is used to assess the load reduction risk of the power system under typhoon power supply in each scenario based on the multiple first power system operation scenario models. If the load reduction is non-zero, the first power system operation scenario model is iteratively corrected by adjusting the generator capacity or optimizing the generator layout.

[0063] The model output module is used to output a power planning scheme that meets the power supply requirements of multiple climates during the typhoon month, based on the corrected first power system operation scenario model.

[0064] The model configuration module is used to configure the power supply of the power system according to the power planning schemes for the multiple climate power supply requirements.

[0065] This application employs a power system operation scenario modeling method based on the spatiotemporal characteristics of typhoons. First, it constructs multiple first-stage power system operation scenario models, which comprehensively reflect the potential impact of typhoons on power system operation. Based on these models, the load reduction risk of the power system under each scenario is assessed. If the load reduction is found to be non-zero, indicating a risk to power supply, the model is iteratively corrected by adjusting generator capacity or optimizing generator layout until the load reduction is zero, ensuring the reliability of power supply under typhoon conditions. Subsequently, based on the corrected model, a power supply planning scheme that meets the power supply requirements of multiple climates during the typhoon month is output, providing scientific guidance for power system power configuration. Finally, based on these planning schemes, the power supply of the power system is precisely controlled, thereby effectively improving the disaster resistance and operational stability of the power system under extreme weather conditions such as typhoons, ensuring the continuity and reliability of power supply. Attached Figure Description

[0066] Figure 1 A flowchart illustrating an embodiment of the power system operation scenario modeling method based on the spatiotemporal characteristics of typhoons provided in this application;

[0067] Figure 2 A schematic diagram of a structural embodiment of the wind field model that comprehensively considers the radial wind speed and moving wind speed of a typhoon provided in this application;

[0068] Figure 3 A schematic diagram of a structural embodiment of the geometric circle determination method for typhoon impact range provided in this application;

[0069] Figure 4 A schematic diagram of a structure of one embodiment of the typhoon track map for the planning year provided in this application;

[0070] Figure 5 A schematic diagram of the structure of one embodiment of the load simulation curve comparison chart during typhoons provided in this application;

[0071] Figure 6 A schematic diagram of the structure of one embodiment of the error histogram provided in this application;

[0072] Figure 7 A schematic diagram of a structure of one embodiment of the simulated load curve for the year to be planned provided in this application;

[0073] Figure 8 A schematic diagram of the structure of one embodiment of the typhoon-month wind farm output curve provided in this application;

[0074] Figure 9 A flowchart illustrating an embodiment of the power supply configuration method based on the spatiotemporal characteristics of typhoons provided in this application;

[0075] Figure 10 A schematic diagram of the structure of an embodiment of the power system operation scenario modeling device based on the spatiotemporal characteristics of typhoons provided in this application;

[0076] Figure 11 This is a schematic diagram of one embodiment of the power supply configuration device based on the spatiotemporal characteristics of typhoons provided in this application. Detailed Implementation

[0077] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0078] Example 1

[0079] Please refer to Figure 1 In order to address the problem that existing technologies cannot accurately and comprehensively address the long-term impact of typhoons on power systems, this invention provides a power system operation scenario modeling method based on the spatiotemporal characteristics of typhoons.

[0080] In this embodiment, steps S01-S06 detail the process of a power system operation scenario modeling method based on the spatiotemporal characteristics of typhoons in this application.

[0081] Step S01: Obtain historical meteorological data, historical load data, and historical typhoon data of the power system.

[0082] As a preferred embodiment of Example 1, the acquisition of historical meteorological data, historical load data, and historical typhoon data of the power system specifically includes:

[0083] Historical meteorological data for the year is obtained from the meteorological data center, including hourly data on temperature, precipitation, wind speed, and humidity. Historical power grid load data is obtained from the power grid operation center. Historical typhoon path data affecting a specific area is obtained from the typhoon data center, including the typhoon's origin latitude and longitude coordinates, hourly central pressure, east-west movement speed, and north-south movement speed. This data provides the basic input for subsequent modeling. This application requires data cleaning and alignment to ensure the temporal consistency of meteorological, load, and typhoon data. For example, if meteorological data for a certain period is missing, interpolation using the average of adjacent periods is used to complete the data.

[0084] Step S02: Based on the historical meteorological data, historical load data, and historical typhoon data, construct a meteorological feature matrix, a typhoon path feature matrix, and a load dataset.

[0085] As a preferred embodiment of Example 1, the step of constructing a meteorological feature matrix, a typhoon path feature matrix, and a load dataset based on the historical meteorological data, historical load data, and historical typhoon data specifically involves:

[0086] Define a meteorological feature matrix, where each row represents meteorological data at a given time point, including temperature, precipitation, wind speed, and relative humidity. Define a typhoon track feature matrix, where each row represents typhoon track data at a given time point, including the typhoon's central pressure, east-west movement speed, and north-south movement speed. Organize historical load data into a load dataset for subsequent load fitting.

[0087] More specifically, the meteorological feature matrix is ​​defined as follows:

[0088]

[0089] In the formula, n represents the number of hours in a year, generally n = 8760; T, R, Wi, and H represent the hourly data of temperature, precipitation, wind speed, and relative humidity in the i-th year, respectively. Load data L corresponding to the time series of meteorological data is also obtained.

[0090] The typhoon path feature matrix is ​​defined as follows:

[0091]

[0092] In the formula, d represents the number of hours the typhoon lasts. P, U, and V represent the hourly data of the typhoon's central pressure, east-west (radial) movement speed, and north-south (zonal) movement speed, respectively. The formula also obtains the latitude and longitude coordinates of the starting points of historical typhoons.

[0093] Step S03: Based on the three stages of the typhoon's passage (before, during, and after), the meteorological feature matrix and load dataset are divided, and a preset neural network is used to fit the mapping relationship between meteorological parameters and load to generate a meteorological-load fitting curve.

[0094] As a preferred embodiment of Example 1, the step of dividing the meteorological feature matrix and load dataset based on the three stages of the typhoon's passage (before, during, and after the typhoon) and using a preset neural network to fit the mapping relationship between meteorological parameters and load to generate a meteorological-load fitting curve is as follows:

[0095] Based on typhoon track forecast data from the National Meteorological Center, the typhoon month is divided into three stages: before, during, and after typhoon passage, and each data set is labeled. A feedforward neural network is used to fit the meteorological-load data, where the typhoon month stage labels and meteorological data are used as input features, and load data is used as the output target. The Levenberg-Marquardt algorithm is used as the training function to configure the network, and a hyperbolic tangent sigmoid function is used as the activation function. The mean squared error (MSE) and coefficient of determination (R²) are used to measure the results. 2 The performance of neural networks is evaluated.

[0096] More specifically, the data input into the feedforward neural network is:

[0097]

[0098] Where l represents the number of years in the historical data, F represents the label matrix, where "1" in F means the data belongs to the period before the typhoon, "2" means the data belongs to the period during the typhoon, and "3" means the data belongs to the period after the typhoon.

[0099] At this point, the data to be fitted is: load data. The historical data for each year are normalized according to the maximum load of each year.

[0100] Step S04: Based on the typhoon path feature matrix, generate random typhoon paths by combining a multivariate joint probability distribution model and a Monte Carlo simulation algorithm, and reconstruct the typhoon path information for the planning year.

[0101] As a preferred embodiment of Example 1, the step of generating random typhoon paths based on the typhoon path feature matrix, combined with a multivariate joint probability distribution model and a Monte Carlo simulation algorithm, and reconstructing the typhoon path information for the planning year specifically involves:

[0102] This study utilizes Multivariate Markov Chain Monte Carlo (MCMC) technology to simulate the paths of historical typhoons by integrating hourly data. First, core parameter sequences, including central pressure, latitude and longitude coordinates, direction of movement, and velocity, of historical typhoons in the target area are obtained from the China Meteorological Administration's typhoon network. A joint probabilistic model is then constructed using the MCMC method to capture the spatiotemporal correlations and dynamic evolution patterns among the parameters, generating a typhoon path dataset for the planning year that maintains historical statistical characteristics. The reconstructed results include hourly data on typhoon central pressure, latitude and longitude, direction of movement, and velocity.

[0103] The specific steps are as follows:

[0104] Construct the joint state transition probability matrix:

[0105] The multivariable joint state transition matrix P1 is an N Num_index ×NNum_index The matrix, P1, each element p mn The meaning is the probability that the typhoon state will change to n in the next moment, given that the current typhoon state is m.

[0106] p mn =P1(X t =n|X t-1 =m)

[0107] In the formula, X t and X t-1 These represent the state of the typhoon at time t and time t-1, respectively.

[0108] The specific steps for constructing the joint state transition probability matrix are as follows:

[0109] 1) Define the typhoon random variable state. First, convert the individual typhoon path information into discrete state points, and then encode the path information into states. For individual typhoon path information, an appropriate number of states N should be selected based on the distribution range. Then, based on the magnitude of the path information indicators at each time point, the data should be assigned to different states, converting the original data into discrete state points. For a typhoon, assuming the number of path information indicators is Num_index, for a certain moment, the typhoon path state is a Num_index dimensional vector s = [s1, s2, ..., s...]. Num_index State encoding is achieved by mapping the value of each dimension to a unique integer index. The state index, stateindex, can be calculated using the following formula:

[0110]

[0111] In the formula, s i -1 maps the values ​​of each dimension from 1 to N to 0 to N-1 (except for the last dimension), where N refers to the maximum possible value for each dimension; N Num_index-1 Assign a position offset to each dimension, where the first dimension (i=1) has the largest position offset N. Num_index-1 The second dimension has N Num_index-2 This process continues until the last dimension has no additional positional offset, meaning the units digit is retained; the +1 is to ensure that the index starts from 1, not 0. After state encoding, the typhoon path state is transformed from a Num_index dimensional vector s into a vector from 0 to N. Num_index Integers between [a certain range].

[0112] 2) Construct an N array to record the frequency of state transitions. Num_index ×N Num_index The sparse matrix F1 will be used to track the number of transitions between different states. Constructing a sparse matrix can save memory space. Specifically, first construct an N...Num_index ×N Num_index The zero matrix F0 is defined as follows: if the initial state in the historical typhoon path sequence is m, and the subsequent states are n, then the corresponding element f in matrix F0 is... mn The value will increment by 1. By iteratively analyzing all adjacent data pairs in the original sequence, matrix F1 will eventually summarize the transition frequencies between different states.

[0113] 3) The joint state transition probability matrix P1 is also a sparse matrix. It is constructed by dividing each element in the joint state transition frequency matrix F1 by the sum of the elements in its row, i.e.:

[0114]

[0115] In the formula, N refers to the maximum value that each dimension can achieve.

[0116] Reconstructing the typhoon path for the planned year:

[0117] 1) Construct a multivariate joint cumulative probability transition matrix, P. cum The construction method is shown in the following formula:

[0118]

[0119] If the dimension of P is N Num_index ×N Num_index Then P cum The dimension is N Num_index ×N Num_index +1. The first column of the cumulative probability transition matrix is ​​all zeros, and starting from the second column, each element P... cum,mn The value of (m rows and n columns) is the sum of the elements in matrix P whose corresponding row and column number is less than n.

[0120] 2) Generate a random state sequence of typhoon paths. Use the Monte Carlo method to generate a typhoon path sequence that meets the requirements for the planning year. First, set the initial path state as m, then continuously generate uniform random numbers u between 0 and 1, and combine them with the cumulative probability transition matrix P. cum Compare the elements in the m-th row; if p cum,mn <p cum,m(n+1) Then, the state n at the next moment. This process is repeated until a random state sequence of the typhoon path that meets the length requirement is generated.

[0121] 3) Constructing the typhoon path information sequence: The random state sequence obtained in the previous step is the numbering of each state, i.e., the sequence is encoded and discrete. It needs to be decoded using a specific method to convert it into a typhoon path information sequence. First, the random state sequence is decoded, i.e., the state index of each time step is inversely transformed into a Num_index dimensional vector s, which is then converted into a discrete state sequence for each region. Then, the discrete states are converted into the corresponding typhoon path information according to the following formula:

[0122] X = X n,min +u(X n,max -X n,min )

[0123] In the formula, X represents the simulated typhoon path information at that moment. n,max X n,min These are the upper and lower limits of the typhoon path range covered by state n, respectively.

[0124] Based on the reconstructed typhoon track information for the planning year, the hourly typhoon center latitude and longitude, hourly typhoon moving speed, and moving direction are calculated.

[0125] In this preferred embodiment, the central pressure, speed, and direction of the typhoon path feature matrix are discretized and divided into multiple state intervals. A multivariate joint state transition probability matrix is ​​constructed by combining historical typhoon data. Based on this probability matrix, a Monte Carlo simulation algorithm is used to generate a random typhoon path state sequence containing hourly state codes. Subsequently, these state sequences are decoded, converting the discrete state codes into continuous spatiotemporal parameters, thereby generating typhoon path information for the planning year. This process not only effectively captures the uncertainty of typhoon paths but also improves the accuracy and reliability of path reconstruction through discretization and stochastic simulation. Ultimately, the generated typhoon path information provides more accurate data support for the operation and planning of the power system, significantly enhancing the power system's disaster resistance and operational efficiency in the face of extreme weather such as typhoons.

[0126] Step S05: Based on the meteorological-load fitting curve, typhoon path information and historical meteorological data, generate the monthly hourly typhoon load data and monthly hourly typhoon meteorological data for the planning year.

[0127] As a preferred embodiment of Example 1, the step of generating monthly hourly typhoon load data and monthly hourly typhoon meteorological data for the planning year based on the meteorological-load fitting curve, typhoon path information, and historical meteorological data specifically involves:

[0128] By statistically reconstructing the duration of the typhoon, the duration of the pre-typhoon and post-typhoon phases can be determined.

[0129] Hourly meteorological data before and after a typhoon are extracted from historical meteorological data, and the average of these data is taken hourly to generate hourly meteorological data before and after a typhoon.

[0130] The statistically reconstructed typhoon duration (i.e., the number of days within a typhoon) is d. typhoon The duration d before and after the typhoon to be reconstructed is then... before and d after for:

[0131]

[0132] in, This indicates rounding down. Based on historical data, we obtain the value d before the i-th typhoon. before Historical meteorological data for the day:

[0133]

[0134] Extracting d after the i-th typhoon after Historical meteorological data for the day:

[0135]

[0136] By averaging hourly data before and after multiple typhoons, reconstructed data before and after the typhoons are obtained:

[0137]

[0138] In the formula, n before and n after T represents the number of data segments collected before and after the typhoon, respectively. before,i R before,i W before,i and H before,i Let d represent the number of typhoons before the i-th typhoon. before Hourly data on temperature, precipitation, wind speed, and relative humidity. after,i R after,i W after,i and H after,i Let d represent the number of typhoons after the i-th typhoon. after Hourly data on temperature, precipitation, wind speed, and relative humidity.

[0139] If the duration of the typhoon phase is less than a preset threshold, the meteorological data during the typhoon in the historical meteorological data will be recorded sequentially; otherwise, the combined data before and after the typhoon in the historical data will be extracted, and hourly meteorological data during the typhoon will be generated according to the preset mean calculation method.

[0140] Analyze historical meteorological data for typhoons. If the number of days in the historical typhoon meteorological data is less than d...typhoon Record historical typhoon day data in chronological order; if the number of days of meteorological data in historical typhoons is greater than or equal to d... typhoon Before extracting historical data After splicing The data is used as a record of historical typhoon days. The d value of the typhoon is derived from this. typhoon Weather data W for the day typhoon The average value was then taken using the method described above to obtain the reconstructed meteorological data from the typhoon.

[0141] Based on hourly meteorological data and meteorological-load fitting curves before, during, and after the typhoon, combined with the maximum values ​​of historical load data, hourly load data for the planning year are generated using a regression fitting method.

[0142] In this preferred embodiment, this application accurately locates the time range of each stage of the typhoon by statistically analyzing the actual number of days the typhoon affects the area and dividing the duration into three stages: before, during, and after the typhoon. Furthermore, hourly meteorological data before and after the typhoon are extracted from historical meteorological data and averaged to generate hourly meteorological data before and after the typhoon. This process effectively integrates historical data, providing a stable data foundation for subsequent analysis. For the mid-typhoon stage, if the duration is less than a preset threshold, the mid-typhoon meteorological data from historical data is directly recorded; if the duration is greater than or equal to the threshold, the mid-typhoon meteorological data is generated by extracting combined data before and after the typhoon from historical data and calculating the average. This flexible processing method ensures the integrity and accuracy of the data. Finally, by combining the hourly meteorological data before, during, and after the typhoon, as well as the meteorological-load fitting curve, and by regressing the planned annual peak value based on the maximum value of the annual historical load data, monthly hourly load data for the typhoon is generated. This series of steps not only improves the accuracy of load data forecasting, but also provides more accurate and reliable meteorological and load data support for the operation and planning of the power system, thereby enhancing the power system's disaster resistance and operational efficiency under extreme weather conditions such as typhoons.

[0143] Step S06: Based on the typhoon path information, monthly meteorological data of the typhoon, and monthly load data of the typhoon, and combined with the preset multi-component coupling model of the typhoon wind field, multiple power system operation scenario models based on the spatiotemporal characteristics of the typhoon are constructed.

[0144] As a preferred embodiment of Example 1, the step of constructing multiple power system operation scenario models based on the typhoon's spatiotemporal characteristics, using the typhoon path information, monthly hourly meteorological data of the typhoon, and monthly hourly load data of the typhoon, combined with a preset multi-component coupling model of the typhoon wind field, specifically involves:

[0145] The construction of the preset typhoon wind field multi-component coupling model is specifically as follows:

[0146] In the Batts wind field model, the maximum gradient wind speed is:

[0147]

[0148] In the formula, K is an empirical constant, taken as 6.72;

[0149] ΔP—Typhoon center pressure difference (hPa), generally refers to the difference between the central pressure of the typhoon and the outer pressure of the typhoon.

[0150] R max — Maximum wind speed radius (km);

[0151] f—Chriolis coefficient. f = 2Ωsinψ, where Ω is the Earth's rotational angular velocity (rad / s) and ψ is the latitude of the typhoon's center.

[0152] The average maximum wind speed over 10 minutes at a height of 10 meters above sea level is:

[0153]

[0154] In the formula, V T This represents the speed at which the center of the typhoon moves.

[0155] The wind speed at the wind turbine hub height z of the wind farm at a distance r from the typhoon center is:

[0156]

[0157] In the formula, β is an empirical coefficient, taken as 0.7;

[0158] γ – empirical coefficient, taken as 0.09.

[0159] The radius of the typhoon's maximum wind speed is:

[0160]

[0161] Based on the typhoon circulation wind field model, the wind speed and direction of the circulation at a certain location can be calculated. Furthermore, based on the moving wind field, the speed and direction of the typhoon's movement at that location can be obtained. Let the circulation wind speed be a vector. Moving wind speed is a vector Their sizes are respectively and The directions are angles. and (These angles are the angles between the vector and the positive x-axis, with longitude as the x-axis and latitude as the y-axis.) For example... Figure 2 The figure shows a wind field model that takes into account both the radial wind speed and the moving wind speed of the typhoon.

[0162] The geometric circle criterion method is used to determine the typhoon's impact range. A simulated circle is constructed with the typhoon's center as the center and R as the radius. If the wind farm is located inside the circle, it is considered to be affected by the typhoon; if it is located outside the circle, it is considered unaffected. Figure 3 As shown.

[0163] During the typhoon month, the wind speed at the wind farm is represented by a line number equal to the number of hours (d) in the typhoon month. typhoon ×24, where the number of columns is num (the number of wind farms). wind_farm matrix V windspeed The values ​​in the matrix Let be the wind speed at the j-th wind farm at time i in the typhoon month. If the wind farm is not affected by the typhoon, then... To reconstruct the typhoon month wind speed, if the wind farm is not affected by the typhoon, then To calculate the wind speed when a wind farm is affected by a typhoon.

[0164] Based on the actual hourly wind speed, and combined with the preset cut-in wind speed, cut-out wind speed, and rated power characteristic curves of the wind turbine, the hourly output data for the typhoon-affected scenario is calculated in segments, specifically as follows:

[0165] The power output data of Taifengyue Wind Farm was calculated using the wind speed-wind power formula, as shown in the following formula.

[0166]

[0167] In the formula, V ci V represents the cut-in wind speed. co V represents the cut-out wind speed. r P represents the rated wind speed. r P represents the active power generated by the wind turbine at rated wind speed. t Let V be the output power of the wind farm at time t. t Let be the wind speed at time t, and A, B, and C be the parameters of the wind turbine power characteristic curve, as shown in the following expression:

[0168]

[0169] By aligning the monthly hourly meteorological data, monthly hourly load data, and hourly power output data of typhoons in time and space, multiple operational scenario models are generated, which include typhoon path intensity, meteorological fluctuations, and load changes.

[0170] In this preferred embodiment, this application combines the center location, moving speed, and direction from the typhoon path information, and utilizes a multi-component coupling model of the typhoon wind field to decompose the wind speed within the typhoon's influence area into radial and moving components. The actual hourly wind speed of the wind farm is then calculated through vector superposition. This process accurately simulates the impact of the typhoon on the wind speed of the wind farm, providing accurate wind speed data for subsequent wind farm output calculations. Furthermore, using the typhoon center as the center and a preset influence radius as a threshold, a geometric circle determination method is employed to determine whether the wind farm is within the typhoon's influence range, thereby filtering out scenarios affected by the typhoon. For these scenarios, hourly output data is calculated segmentally based on the actual hourly wind speed and the characteristic curves of the wind turbines. This step accurately assesses the impact of the typhoon on the wind farm's power generation capacity. Finally, the monthly hourly meteorological data, load data, and wind farm output data are aligned spatiotemporally to generate multiple operational scenario models that include typhoon path intensity, meteorological fluctuations, and load changes. These models comprehensively reflect the impact of typhoons on the power system, providing a scientific basis for the operation and planning of the power system. They significantly improve the disaster resistance and operational efficiency of the power system under extreme weather conditions such as typhoons, and have important scientific significance and practical application value.

[0171] As a preferred embodiment, this application establishes a typhoon path reconstruction model based on the historical typhoon trajectory database of a coastal province in southern China from 2017 to 2020, using the multivariate MCMC method. The number of discrete states is selected as 10. The historical typhoon paths and the planned typhoon paths are as follows: Figure 4 As shown.

[0172] By comparing and analyzing the planned typhoon paths with historical typhoon paths, it was found that the generated paths showed a high degree of agreement with historical data in terms of starting position and movement trend. This agreement indicates that the path generation method of this invention can effectively capture the basic characteristics of typhoons, and therefore can represent the possible paths of typhoons in the planned year to a certain extent. In addition, the generation of planned typhoon paths takes into account various possible typhoon development trends, thus generating multiple different paths. These paths show certain differences in movement direction and intensity changes, but overall still conform to the statistical characteristics of historical typhoons. This uncertainty provides a more comprehensive reference for planned typhoon paths.

[0173] Based on meteorological monitoring data and historical typhoon trajectory databases from a coastal province in southern China from 2017 to 2020, this study divides the research period into three stages: before, during, and after typhoon passage, integrating the time dimension of typhoon events. A feedforward neural network model is constructed using the province's power grid load time-series dataset, yielding the load simulation curve. (See figure) Figure 5 .

[0174] Figure 5The comparative analysis of the simulated load and the actual load shown in the model indicates that, despite some deviations and errors, the simulated load generally reflects the changing trends and fluctuations of the actual load well. Model coefficient of determination R0 2 =0.70159, which basically meets the load reconfiguration accuracy requirements in extreme weather such as typhoons.

[0175] To quantitatively evaluate the predictive performance of the neural network model, an error distribution histogram was plotted for visual analysis. Experimental results are as follows: Figure 6 The results show that the model prediction error (predicted value - actual value) is mainly distributed in the range of [-0.1058, 0.09679], with its distribution center approaching the zero error axis, exhibiting an approximately symmetrical distribution. This statistical characteristic indicates that the model has a local accuracy concentration characteristic, and the error distribution does not show significant systematic deviation characteristics; the frequency distribution of positive and negative errors is basically balanced. The above error distribution pattern verifies the predictive effectiveness of the established neural network model under the target working condition, meeting the basic requirements for model stability in practical engineering applications.

[0176] Based on the meteorological reconstruction data during the typhoon mentioned above Substituting the data into the feedforward neural network, we obtain the normalized time series of the load for the year to be planned, as follows: Figure 7 As shown;

[0177] In the lead-up to a typhoon, residents and businesses generally take preventative measures, such as charging equipment and stockpiling supplies, leading to an increase in electricity load at certain times. Furthermore, to mitigate potential typhoon damage, some areas implemented preventative power outages, resulting in a decrease in overall electricity load. Simulation results show that the load curve fluctuated slightly before the typhoon's arrival, then gradually decreased, consistent with the actual observed load change trend.

[0178] During the typhoon, strong winds and torrential rains directly damaged power facilities, causing widespread power outages and a sharp drop in power load. However, as critical facilities such as hospitals and emergency centers activated emergency or backup power supplies, power load in some areas recovered. The simulated load curve showed a significant decline during the typhoon's passage, with slight rebounds at times, consistent with the actual load trend.

[0179] After the typhoon passed, the power company quickly launched repair work and gradually restored power supply, with the power load showing a slow recovery trend. The simulated load curve gradually rose after the typhoon, consistent with the actual load change trend. This indicates that the simulation model can capture the power load recovery process well after the typhoon.

[0180] Based on the reconstructed typhoon tracks for the planned year and the reconstructed meteorological data for the typhoon months, and considering the physical relationship between the typhoon tracks and the impact on wind farms, as well as the wind power-wind energy conversion relationship, the wind farm output during the typhoon period was calculated as follows: Figure 8 As shown.

[0181] Detailed analysis of the wind farm output curves during typhoon months reveals distinct phased variations in wind farm output before, during, and after typhoons, consistent with actual conditions. Specifically: Before a typhoon, air pressure is typically high, while wind speeds are weak and unstable, resulting in lower overall wind farm output. When a typhoon enters the wind farm area, the strong winds, often exceeding normal wind speeds, drive wind turbine blades to rotate rapidly, significantly increasing power generation, potentially reaching or even exceeding full capacity. After a typhoon, as the typhoon moves away, wind speeds decrease rapidly. Furthermore, potential precipitation and air pressure changes may further impact wind turbine efficiency, causing wind farm output to decline rapidly, gradually returning to pre-typhoon levels and stabilizing. In conclusion, the method proposed in this application not only reflects the direct impact of typhoons on wind farm operation but also reveals the potential operational patterns of wind farms at different stages.

[0182] This application constructs meteorological feature matrices, typhoon path feature matrices, and load datasets by acquiring historical meteorological, load, and typhoon data from the power system, providing fundamental data support for subsequent modeling. The meteorological feature matrices and load datasets are divided into three stages: before, during, and after a typhoon's passage. A neural network is used to fit the mapping relationship between meteorological parameters and load, generating a meteorological-load fitting curve. This allows the model to more accurately reflect the impact of different typhoon stages on the load. A multivariate joint probability distribution model and Monte Carlo simulation algorithm are combined to generate random typhoon paths, reconstructing typhoon path information for the planning year and effectively capturing the uncertainty of typhoon paths. Based on the reconstructed typhoon path information and the meteorological-load fitting curve, monthly hourly load and meteorological data for the planning year are generated, providing a detailed data foundation for constructing power system operation scenarios. Finally, combined with a pre-defined multi-component coupled typhoon wind field model, multiple power system operation scenario models based on the spatiotemporal characteristics of typhoons are constructed. These models comprehensively reflect the impact of typhoons on the power system, providing a scientific basis for power system operation and planning, thereby significantly improving the power system's disaster resistance and operational efficiency under extreme weather conditions such as typhoons. This application effectively solves the problem that existing technologies cannot accurately and comprehensively address the long-term impact of typhoons on power systems.

[0183] Example 2

[0184] Please refer to Figure 9 , Figure 9This is a flowchart illustrating an embodiment of a power supply configuration method based on the spatiotemporal characteristics of typhoons provided in this application. Steps S101-S103 detail the process of a power supply configuration method based on the spatiotemporal characteristics of typhoons in this application.

[0185] S101: Based on the power system operation scenario modeling method based on the spatiotemporal characteristics of typhoons, construct multiple first power system operation scenario models for the power system to be controlled;

[0186] S102: Based on the multiple first power system operation scenario models, assess the load reduction risk of the power system under typhoon power supply in each scenario. If the load reduction is non-zero, iteratively correct the first power system operation scenario model by adjusting the generator capacity or optimizing the generator layout.

[0187] As a preferred embodiment of Embodiment Two, the load reduction risk of the power system under typhoon power supply in each scenario is assessed based on the multiple first power system operation scenario models. If the load reduction is non-zero, the first power system operation scenario model is iteratively corrected by adjusting the generator capacity or optimizing the generator layout. Specifically:

[0188] In the conventional power planning phase, with economic efficiency and basic reliability as the core objectives, an initial power scheme for the planning year is constructed. First, a basic load demand curve and peak capacity requirements for the planning year are established based on historical load data. Second, the investment and construction costs, operational constraints, and environmental indicators of conventional power sources such as thermal power, hydropower, and nuclear power are integrated, while the output fluctuation characteristics of renewable energy sources such as wind power and photovoltaics are quantified. Subsequently, a multi-objective optimization algorithm is used to find the optimal combination of power source type, installed capacity, and spatial layout under the condition of satisfying power supply reliability constraints, ensuring that the planning scheme has the lowest power supply cost and achieves power balance standards under conventional scenarios (the method can refer to existing power system power planning technologies). The conventional planning scheme for the planning year output in this phase serves as the benchmark input for subsequent typhoon scenario verification.

[0189] To address the impact of typhoon extreme weather on the power system, a verification of the conventional planning scheme for the planning year was conducted based on a typhoon monthly scenario set. The verification process was as follows: 1) Based on the wind farm output data and load data for the typhoon months of the planning year, a verification set containing Nsce typical scenarios was constructed; 2) The load reduction risk under each typhoon scenario was assessed (the method can refer to existing power system time-series operation simulation technology), and the typhoon monthly load reduction was calculated. The verification rule required that the typhoon monthly load reduction be 0; otherwise, the scheme was deemed to have failed the verification.

[0190] If the planned power supply scheme fails the typhoon month scenario verification, a multi-dimensional correction mechanism will be initiated to iteratively optimize the scheme. Correction strategies include increasing generator capacity and optimizing generator locations. The modified power supply plan needs to be re-verified under the typhoon month scenario. This verification-correction process will be repeated until the planned year's power supply plan meets the power supply reliability requirements under the typhoon month scenario.

[0191] S103: Based on the revised first power system operation scenario model, output a power planning scheme that meets the power supply requirements of multiple climates during the typhoon month;

[0192] S104: Configure the power supply of the power system according to the power supply planning scheme of the multiple climate power supply requirements.

[0193] This application employs a power system operation scenario modeling method based on the spatiotemporal characteristics of typhoons. First, it constructs multiple first-stage power system operation scenario models, which comprehensively reflect the potential impact of typhoons on power system operation. Based on these models, the load reduction risk of the power system under each scenario is assessed. If the load reduction is found to be non-zero, indicating a risk to power supply, the model is iteratively corrected by adjusting generator capacity or optimizing generator layout until the load reduction is zero, ensuring the reliability of power supply under typhoon conditions. Subsequently, based on the corrected model, a power supply planning scheme that meets the power supply requirements of multiple climates during the typhoon month is output, providing scientific guidance for power system power configuration. Finally, based on these planning schemes, the power supply of the power system is precisely controlled, thereby effectively improving the disaster resistance and operational stability of the power system under extreme weather conditions such as typhoons, ensuring the continuity and reliability of power supply.

[0194] Example 3

[0195] Please refer to Figure 10 This application provides a power system operation scenario modeling device based on the spatiotemporal characteristics of typhoons.

[0196] In this embodiment, the power system operation scenario modeling device based on the spatiotemporal characteristics of typhoons includes an acquisition module 10, a construction module 20, a fitting module 30, a data reconstruction module 40, and a modeling module 50.

[0197] The acquisition module 10 is used to acquire historical meteorological data, historical load data and historical typhoon data of the power system.

[0198] As a preferred embodiment of Example 3, the acquisition of historical meteorological data, historical load data, and historical typhoon data of the power system specifically includes:

[0199] Historical meteorological data for the year is obtained from the meteorological data center, including hourly data on temperature, precipitation, wind speed, and humidity. Historical power grid load data is obtained from the power grid operation center. Historical typhoon path data affecting a specific area is obtained from the typhoon data center, including the typhoon's origin latitude and longitude coordinates, hourly central pressure, east-west movement speed, and north-south movement speed. This data provides the basic input for subsequent modeling. This application requires data cleaning and alignment to ensure the temporal consistency of meteorological, load, and typhoon data. For example, if meteorological data for a certain period is missing, interpolation using the average of adjacent periods is used to complete the data.

[0200] The construction module 20 is used to construct a meteorological feature matrix, a typhoon path feature matrix, and a load dataset based on the historical meteorological data, historical load data, and historical typhoon data.

[0201] As a preferred embodiment of Example 3, the step of constructing a meteorological feature matrix, a typhoon path feature matrix, and a load dataset based on the historical meteorological data, historical load data, and historical typhoon data specifically involves:

[0202] Define a meteorological feature matrix, where each row represents meteorological data at a given time point, including temperature, precipitation, wind speed, and relative humidity. Define a typhoon track feature matrix, where each row represents typhoon track data at a given time point, including the typhoon's central pressure, east-west movement speed, and north-south movement speed. Organize historical load data into a load dataset for subsequent load fitting.

[0203] More specifically, the meteorological feature matrix is ​​defined as follows:

[0204]

[0205] In the formula, n represents the number of hours in a year, generally n = 8760; T, R, Wi, and H represent the hourly data of temperature, precipitation, wind speed, and relative humidity in the i-th year, respectively. Load data L corresponding to the time series of meteorological data is also obtained.

[0206] The typhoon path feature matrix is ​​defined as follows:

[0207]

[0208] In the formula, d represents the number of hours the typhoon lasts. P, U, and V represent the hourly data of the typhoon's central pressure, east-west (radial) movement speed, and north-south (zonal) movement speed, respectively. The formula also obtains the latitude and longitude coordinates of the starting points of historical typhoons.

[0209] The fitting module 30 is used to divide the meteorological feature matrix and load dataset into three stages: before, during and after the typhoon, and to use a preset neural network to fit the mapping relationship between meteorological parameters and load to generate a meteorological-load fitting curve.

[0210] As a preferred embodiment of Example 3, the step of dividing the meteorological feature matrix and load dataset based on the three stages of typhoon passage (before, during, and after the typhoon) and using a preset neural network to fit the mapping relationship between meteorological parameters and load to generate a meteorological-load fitting curve is as follows:

[0211] Based on typhoon track forecast data from the National Meteorological Center, the typhoon month is divided into three stages: before, during, and after typhoon passage, and each data set is labeled. A feedforward neural network is used to fit the meteorological-load data, where the typhoon month stage labels and meteorological data are used as input features, and load data is used as the output target. The Levenberg-Marquardt algorithm is used as the training function to configure the network, and a hyperbolic tangent sigmoid function is used as the activation function. The mean squared error (MSE) and coefficient of determination (R²) are used to measure the results. 2 The performance of neural networks is evaluated.

[0212] More specifically, the data input into the feedforward neural network is:

[0213]

[0214] Where l represents the number of years in the historical data, F represents the label matrix, where "1" in F means the data belongs to the period before the typhoon, "2" means the data belongs to the period during the typhoon, and "3" means the data belongs to the period after the typhoon.

[0215] At this point, the data to be fitted is: load data The historical data for each year are normalized according to the maximum load of each year.

[0216] The data reconstruction module 40 is used to generate random typhoon paths based on the typhoon path feature matrix, combined with a multivariate joint probability distribution model and a Monte Carlo simulation algorithm, and to reconstruct the typhoon path information for the planning year.

[0217] As a preferred embodiment of Example 3, the step of generating random typhoon paths based on the typhoon path feature matrix, combined with a multivariate joint probability distribution model and a Monte Carlo simulation algorithm, and reconstructing the typhoon path information for the planning year specifically involves:

[0218] This study utilizes Multivariate Markov Chain Monte Carlo (MCMC) technology to simulate the paths of historical typhoons by integrating hourly data. First, core parameter sequences, including central pressure, latitude and longitude coordinates, direction of movement, and velocity, of historical typhoons in the target area are obtained from the China Meteorological Administration's typhoon network. A joint probabilistic model is then constructed using the MCMC method to capture the spatiotemporal correlations and dynamic evolution patterns among the parameters, generating a typhoon path dataset for the planning year that maintains historical statistical characteristics. The reconstructed results include hourly data on typhoon central pressure, latitude and longitude, direction of movement, and velocity.

[0219] The specific steps are as follows:

[0220] Construct the joint state transition probability matrix:

[0221] The multivariable joint state transition matrix P1 is an N Num_index ×N Num_index The matrix, P1, each element p mn The meaning is the probability that the typhoon state will change to n in the next moment, given that the current typhoon state is m.

[0222] p mn =P1(X t =n|X t-1 =m)

[0223] In the formula, X t and X t-1 These represent the state of the typhoon at time t and time t-1, respectively.

[0224] The specific steps for constructing the joint state transition probability matrix are as follows:

[0225] 1) Define the typhoon random variable state. First, convert the individual typhoon path information into discrete state points, and then encode the path information into states. For individual typhoon path information, an appropriate number of states N should be selected based on the distribution range. Then, based on the magnitude of the path information indicators at each time point, the data should be assigned to different states, converting the original data into discrete state points. For a typhoon, assuming the number of path information indicators is Num_index, for a certain moment, the typhoon path state is a Num_index dimensional vector s = [s1, s2, ..., s...]. Num_index State encoding is achieved by mapping the value of each dimension to a unique integer index. The state index, stateindex, can be calculated using the following formula:

[0226]

[0227] In the formula, s i-1 maps the values ​​of each dimension from 1 to N to 0 to N-1 (except for the last dimension), where N refers to the maximum possible value for each dimension; N Num_index-1 Assign a position offset to each dimension, where the first dimension (i=1) has the largest position offset N. Num_index-1 The second dimension has N Num_index-2 This process continues until the last dimension has no additional positional offset, meaning the units digit is retained; the +1 is to ensure that the index starts from 1, not 0. After state encoding, the typhoon path state is transformed from a Num_index dimensional vector s into a vector from 0 to N. Num_index Integers between [a certain range].

[0228] 2) Construct an N array to record the frequency of state transitions. Num_index ×N Num_index The sparse matrix F1 will be used to track the number of transitions between different states. Constructing a sparse matrix can save memory space. Specifically, first construct an N... Num_index ×N Num_index The zero matrix F0 is defined as follows: if the initial state in the historical typhoon path sequence is m, and the subsequent states are n, then the corresponding element f in matrix F0 is... mn The value will increment by 1. By iteratively analyzing all adjacent data pairs in the original sequence, matrix F1 will eventually summarize the transition frequencies between different states.

[0229] 3) The joint state transition probability matrix P1 is also a sparse matrix. It is constructed by dividing each element in the joint state transition frequency matrix F1 by the sum of the elements in its row, i.e.:

[0230]

[0231] In the formula, N refers to the maximum value that each dimension can achieve.

[0232] Reconstructing the typhoon path for the planned year:

[0233] 1) Construct a multivariate joint cumulative probability transition matrix, P. cum The construction method is shown in the following formula:

[0234]

[0235] If the dimension of P is N Num_index ×N Num_index Then P cum The dimension is N Num_index ×N Num_index +1. The first column of the cumulative probability transition matrix is ​​all zeros, and starting from the second column, each element P... cum,mnThe value of (m rows and n columns) is the sum of the elements in matrix P whose corresponding row and column number is less than n.

[0236] 2) Generate a random state sequence of typhoon paths. Use the Monte Carlo method to generate a typhoon path sequence that meets the requirements for the planning year. First, set the initial path state as m, then continuously generate uniform random numbers u between 0 and 1, and combine them with the cumulative probability transition matrix P. cum Compare the elements in the m-th row; if p cum,mn <p cum,m(n+1) Then, the state n at the next moment. This process is repeated until a random state sequence of the typhoon path that meets the length requirement is generated.

[0237] 3) Constructing the typhoon path information sequence: The random state sequence obtained in the previous step is the numbering of each state, i.e., the sequence is encoded and discrete. It needs to be decoded using a specific method to convert it into a typhoon path information sequence. First, the random state sequence is decoded, i.e., the state index of each time step is inversely transformed into a Num_index dimensional vector s, which is then converted into a discrete state sequence for each region. Then, the discrete states are converted into the corresponding typhoon path information according to the following formula:

[0238] X = X n,min +u(X n,max -X n,min )

[0239] In the formula, X represents the simulated typhoon path information at that moment. n,max X n,min These are the upper and lower limits of the typhoon path range covered by state n, respectively.

[0240] Based on the reconstructed typhoon track information for the planning year, the hourly typhoon center latitude and longitude, hourly typhoon moving speed, and moving direction are calculated.

[0241] In this preferred embodiment, the central pressure, speed, and direction of the typhoon path feature matrix are discretized and divided into multiple state intervals. A multivariate joint state transition probability matrix is ​​constructed by combining historical typhoon data. Based on this probability matrix, a Monte Carlo simulation algorithm is used to generate a random typhoon path state sequence containing hourly state codes. Subsequently, these state sequences are decoded, converting the discrete state codes into continuous spatiotemporal parameters, thereby generating typhoon path information for the planning year. This process not only effectively captures the uncertainty of typhoon paths but also improves the accuracy and reliability of path reconstruction through discretization and stochastic simulation. Ultimately, the generated typhoon path information provides more accurate data support for the operation and planning of the power system, significantly enhancing the power system's disaster resistance and operational efficiency in the face of extreme weather such as typhoons.

[0242] The data reconstruction module 40 is also used to generate monthly hourly typhoon load data and monthly hourly typhoon meteorological data for the planning year based on the meteorological-load fitting curve, typhoon path information and historical meteorological data.

[0243] As a preferred embodiment of Example 3, the step of generating monthly hourly typhoon load data and monthly hourly typhoon meteorological data for the planning year based on the meteorological-load fitting curve, typhoon path information, and historical meteorological data specifically involves:

[0244] By statistically reconstructing the duration of the typhoon, the duration of the pre-typhoon and post-typhoon phases can be determined.

[0245] Hourly meteorological data before and after a typhoon are extracted from historical meteorological data, and the average of these data is taken hourly to generate hourly meteorological data before and after a typhoon.

[0246] The statistically reconstructed typhoon duration (i.e., the number of days within a typhoon) is d. typhoon The duration d before and after the typhoon to be reconstructed is then... before and d after for:

[0247]

[0248] in, This indicates rounding down. Based on historical data, we obtain the value d before the i-th typhoon. before Historical meteorological data for the day:

[0249]

[0250] Extracting d after the i-th typhoon after Historical meteorological data for the day:

[0251]

[0252] By averaging hourly data before and after multiple typhoons, reconstructed data before and after the typhoons are obtained:

[0253]

[0254] In the formula, n before and n after T represents the number of data segments collected before and after the typhoon, respectively. before,i R before,i W before,i and H before,i Let d represent the number of typhoons before the i-th typhoon. before Hourly data on temperature, precipitation, wind speed, and relative humidity. after,i R after,i W after,i and H after,i Let d represent the number of typhoons after the i-th typhoon. after Hourly data on temperature, precipitation, wind speed, and relative humidity.

[0255] If the duration of the typhoon phase is less than a preset threshold, the meteorological data during the typhoon in the historical meteorological data will be recorded sequentially; otherwise, the combined data before and after the typhoon in the historical data will be extracted, and hourly meteorological data during the typhoon will be generated according to the preset mean calculation method.

[0256] Analyze historical meteorological data for typhoons. If the number of days in the historical typhoon meteorological data is less than d... typhoon Record historical typhoon day data in chronological order; if the number of days of meteorological data in historical typhoons is greater than or equal to d... typhoon Before extracting historical data After splicing The data is used as a record of historical typhoon days. The d value of the typhoon is derived from this. typhoon Weather data W for the day typhoon The average value was then taken using the method described above to obtain the reconstructed meteorological data from the typhoon.

[0257] Based on hourly meteorological data and meteorological-load fitting curves before, during, and after the typhoon, combined with the maximum values ​​of historical load data, hourly load data for the planning year are generated using a regression fitting method.

[0258] In this preferred embodiment, this application accurately locates the time range of each stage of the typhoon by statistically analyzing the actual number of days the typhoon affects the area and dividing the duration into three stages: before, during, and after the typhoon. Furthermore, hourly meteorological data before and after the typhoon are extracted from historical meteorological data and averaged to generate hourly meteorological data before and after the typhoon. This process effectively integrates historical data, providing a stable data foundation for subsequent analysis. For the mid-typhoon stage, if the duration is less than a preset threshold, the mid-typhoon meteorological data from historical data is directly recorded; if the duration is greater than or equal to the threshold, the mid-typhoon meteorological data is generated by extracting combined data before and after the typhoon from historical data and calculating the average. This flexible processing method ensures the integrity and accuracy of the data. Finally, by combining the hourly meteorological data before, during, and after the typhoon, as well as the meteorological-load fitting curve, and by regressing the planned annual peak value based on the maximum value of the annual historical load data, monthly hourly load data for the typhoon is generated. This series of steps not only improves the accuracy of load data forecasting, but also provides more accurate and reliable meteorological and load data support for the operation and planning of the power system, thereby enhancing the power system's disaster resistance and operational efficiency under extreme weather conditions such as typhoons.

[0259] The modeling module 50 is used to construct multiple power system operation scenario models based on the spatiotemporal characteristics of typhoons, based on the typhoon path information, monthly meteorological data of typhoons, and monthly load data of typhoons, combined with a preset multi-component coupling model of typhoon wind fields.

[0260] As a preferred embodiment of Example 3, the step of constructing multiple power system operation scenario models based on the typhoon's spatiotemporal characteristics, using the typhoon path information, monthly hourly meteorological data, and monthly hourly load data, combined with a preset multi-component coupling model of the typhoon wind field, specifically involves:

[0261] The construction of the preset typhoon wind field multi-component coupling model is specifically as follows:

[0262] In the Batts wind field model, the maximum gradient wind speed is:

[0263]

[0264] In the formula, K is an empirical constant, taken as 6.72;

[0265] ΔP—Typhoon center pressure difference (hPa), generally refers to the difference between the central pressure of the typhoon and the outer pressure of the typhoon.

[0266] R max — Maximum wind speed radius (km);

[0267] f—Chriolis coefficient. f = 2Ωsinψ, where Ω is the Earth's rotational angular velocity (rad / s) and ψ is the latitude of the typhoon's center.

[0268] The average maximum wind speed over 10 minutes at a height of 10 meters above sea level is:

[0269]

[0270] In the formula, V T This represents the speed at which the center of the typhoon moves.

[0271] The wind speed at the wind turbine hub height z of the wind farm at a distance r from the typhoon center is:

[0272]

[0273] In the formula, β is an empirical coefficient, taken as 0.7;

[0274] γ – empirical coefficient, taken as 0.09.

[0275] The radius of the typhoon's maximum wind speed is:

[0276]

[0277] Based on the typhoon circulation wind field model, the wind speed and direction of the circulation at a certain location can be calculated. Furthermore, based on the moving wind field, the speed and direction of the typhoon's movement at that location can be obtained. Let the circulation wind speed be a vector. Moving wind speed is a vector Their sizes are respectively and The directions are angles. and (These angles are the angles between the vector and the positive x-axis, with longitude as the x-axis and latitude as the y-axis.) For example... Figure 2 The figure shows a wind field model that takes into account both the radial wind speed and the moving wind speed of the typhoon.

[0278] The geometric circle criterion method is used to determine the typhoon's impact range. A simulated circle is constructed with the typhoon's center as the center and R as the radius. If the wind farm is located inside the circle, it is considered to be affected by the typhoon; if it is located outside the circle, it is considered unaffected. Figure 3 As shown.

[0279] During the typhoon month, the wind speed at the wind farm is represented by a line number equal to the number of hours (d) in the typhoon month. typhoon ×24, where the number of columns is num (the number of wind farms). wind_farm matrix V windspeed The values ​​in the matrix Let be the wind speed at the j-th wind farm at time i in the typhoon month. If the wind farm is not affected by the typhoon, then... To reconstruct the typhoon month wind speed, if the wind farm is not affected by the typhoon, then To calculate the wind speed when a wind farm is affected by a typhoon.

[0280] Based on the actual hourly wind speed, and combined with the preset cut-in wind speed, cut-out wind speed, and rated power characteristic curves of the wind turbine, the hourly output data for the typhoon-affected scenario is calculated in segments, specifically as follows:

[0281] The power output data of Taifengyue Wind Farm was calculated using the wind speed-wind power formula, as shown in the following formula.

[0282]

[0283] In the formula, V ci V represents the cut-in wind speed. co V represents the cut-out wind speed. r P represents the rated wind speed. r P represents the active power generated by the wind turbine at rated wind speed. t Let V be the output power of the wind farm at time t. t Let be the wind speed at time t, and A, B, and C be the parameters of the wind turbine power characteristic curve, as shown in the following expression:

[0284]

[0285] By aligning the monthly hourly meteorological data, monthly hourly load data, and hourly power output data of typhoons in time and space, multiple operational scenario models are generated, which include typhoon path intensity, meteorological fluctuations, and load changes.

[0286] In this preferred embodiment, this application combines the center location, moving speed, and direction from the typhoon path information, and utilizes a multi-component coupling model of the typhoon wind field to decompose the wind speed within the typhoon's influence area into radial and moving components. The actual hourly wind speed of the wind farm is then calculated through vector superposition. This process accurately simulates the impact of the typhoon on the wind speed of the wind farm, providing accurate wind speed data for subsequent wind farm output calculations. Furthermore, using the typhoon center as the center and a preset influence radius as a threshold, a geometric circle determination method is employed to determine whether the wind farm is within the typhoon's influence range, thereby filtering out scenarios affected by the typhoon. For these scenarios, hourly output data is calculated segmentally based on the actual hourly wind speed and the characteristic curves of the wind turbines. This step accurately assesses the impact of the typhoon on the wind farm's power generation capacity. Finally, the monthly hourly meteorological data, load data, and wind farm output data are aligned spatiotemporally to generate multiple operational scenario models that include typhoon path intensity, meteorological fluctuations, and load changes. These models comprehensively reflect the impact of typhoons on the power system, providing a scientific basis for the operation and planning of the power system. They significantly improve the disaster resistance and operational efficiency of the power system under extreme weather conditions such as typhoons, and have important scientific significance and practical application value.

[0287] As a preferred embodiment of Example 3, this application establishes a typhoon path reconstruction model based on the historical typhoon trajectory database of a coastal province in southern China from 2017 to 2020, using the multivariate MCMC method. The number of discrete states is selected as 10. The historical typhoon paths and the planned typhoon paths are as follows: Figure 4 As shown.

[0288] By comparing and analyzing the planned typhoon paths with historical typhoon paths, it was found that the generated paths showed a high degree of agreement with historical data in terms of starting position and movement trend. This agreement indicates that the path generation method of this invention can effectively capture the basic characteristics of typhoons, and therefore can represent the possible paths of typhoons in the planned year to a certain extent. In addition, the generation of planned typhoon paths takes into account various possible typhoon development trends, thus generating multiple different paths. These paths show certain differences in movement direction and intensity changes, but overall still conform to the statistical characteristics of historical typhoons. This uncertainty provides a more comprehensive reference for planned typhoon paths.

[0289] Based on meteorological monitoring data and historical typhoon trajectory databases from a coastal province in southern China from 2017 to 2020, this study divides the research period into three stages: before, during, and after typhoon passage, integrating the time dimension of typhoon events. A feedforward neural network model is constructed using the province's power grid load time-series dataset, yielding the load simulation curve. (See figure) Figure 5 .

[0290] Figure 5 The comparative analysis of the simulated load and the actual load shown in the model indicates that, despite some deviations and errors, the simulated load generally reflects the changing trends and fluctuations of the actual load well. Model coefficient of determination R0 2 =0.70159, which basically meets the load reconfiguration accuracy requirements in extreme weather such as typhoons.

[0291] To quantitatively evaluate the predictive performance of the neural network model, an error distribution histogram was plotted for visual analysis. Experimental results are as follows: Figure 6 The results show that the model prediction error (predicted value - actual value) is mainly distributed in the range of [-0.1058, 0.09679], with its distribution center approaching the zero error axis, exhibiting an approximately symmetrical distribution. This statistical characteristic indicates that the model has a local accuracy concentration characteristic, and the error distribution does not show significant systematic deviation characteristics; the frequency distribution of positive and negative errors is basically balanced. The above error distribution pattern verifies the predictive effectiveness of the established neural network model under the target working condition, meeting the basic requirements for model stability in practical engineering applications.

[0292] Based on the meteorological reconstruction data during the typhoon mentioned above Substituting the data into the feedforward neural network, we obtain the normalized time series of the load for the year to be planned, as follows: Figure 7 As shown;

[0293] In the lead-up to a typhoon, residents and businesses generally take preventative measures, such as charging equipment and stockpiling supplies, leading to an increase in electricity load at certain times. Furthermore, to mitigate potential typhoon damage, some areas implemented preventative power outages, resulting in a decrease in overall electricity load. Simulation results show that the load curve fluctuated slightly before the typhoon's arrival, then gradually decreased, consistent with the actual observed load change trend.

[0294] During the typhoon, strong winds and torrential rains directly damaged power facilities, causing widespread power outages and a sharp drop in power load. However, as critical facilities such as hospitals and emergency centers activated emergency or backup power supplies, power load in some areas recovered. The simulated load curve showed a significant decline during the typhoon's passage, with slight rebounds at times, consistent with the actual load trend.

[0295] After the typhoon passed, the power company quickly launched repair work and gradually restored power supply, with the power load showing a slow recovery trend. The simulated load curve gradually rose after the typhoon, consistent with the actual load change trend. This indicates that the simulation model can capture the power load recovery process well after the typhoon.

[0296] Based on the reconstructed typhoon tracks for the planned year and the reconstructed meteorological data for the typhoon months, and considering the physical relationship between the typhoon tracks and the impact on wind farms, as well as the wind power-wind energy conversion relationship, the wind farm output during the typhoon period was calculated as follows: Figure 8 As shown.

[0297] Detailed analysis of the wind farm output curves during typhoon months reveals distinct phased variations in wind farm output before, during, and after typhoons, consistent with actual conditions. Specifically: Before a typhoon, air pressure is typically high, while wind speeds are weak and unstable, resulting in lower overall wind farm output. When a typhoon enters the wind farm area, the strong winds, often exceeding normal wind speeds, drive wind turbine blades to rotate rapidly, significantly increasing power generation, potentially reaching or even exceeding full capacity. After a typhoon, as the typhoon moves away, wind speeds decrease rapidly. Furthermore, potential precipitation and air pressure changes may further impact wind turbine efficiency, causing wind farm output to decline rapidly, gradually returning to pre-typhoon levels and stabilizing. In conclusion, the method proposed in this application not only reflects the direct impact of typhoons on wind farm operation but also reveals the potential operational patterns of wind farms at different stages.

[0298] This application constructs meteorological feature matrices, typhoon path feature matrices, and load datasets by acquiring historical meteorological, load, and typhoon data from the power system, providing fundamental data support for subsequent modeling. The meteorological feature matrices and load datasets are divided into three stages: before, during, and after a typhoon's passage. A neural network is used to fit the mapping relationship between meteorological parameters and load, generating a meteorological-load fitting curve. This allows the model to more accurately reflect the impact of different typhoon stages on the load. A multivariate joint probability distribution model and Monte Carlo simulation algorithm are combined to generate random typhoon paths, reconstructing typhoon path information for the planning year and effectively capturing the uncertainty of typhoon paths. Based on the reconstructed typhoon path information and the meteorological-load fitting curve, monthly hourly load and meteorological data for the planning year are generated, providing a detailed data foundation for constructing power system operation scenarios. Finally, combined with a pre-defined multi-component coupled typhoon wind field model, multiple power system operation scenario models based on the spatiotemporal characteristics of typhoons are constructed. These models comprehensively reflect the impact of typhoons on the power system, providing a scientific basis for power system operation and planning, thereby significantly improving the power system's disaster resistance and operational efficiency under extreme weather conditions such as typhoons. This application effectively solves the problem that existing technologies cannot accurately and comprehensively address the long-term impact of typhoons on power systems.

[0299] Example 4

[0300] Please refer to Figure 11 , Figure 11 This is a schematic diagram of an embodiment of a power supply configuration device based on the spatiotemporal characteristics of typhoons provided in this application; the power supply configuration device based on the spatiotemporal characteristics of typhoons includes a model building module 101, a model correction module 102, a model output module 103, and a model configuration module 104;

[0301] The model building module 101 is used to build multiple first power system operation scenario models of the power system to be controlled according to the power system operation scenario modeling method based on the spatiotemporal characteristics of typhoons.

[0302] The model correction module 102 is used to assess the load reduction risk of the power system under typhoon power supply in each scenario based on the multiple first power system operation scenario models. If the load reduction is non-zero, the first power system operation scenario model is iteratively corrected by adjusting the generator set capacity or optimizing the unit layout.

[0303] As a preferred embodiment of Example 4, the load reduction risk of the power system under typhoon power supply in each scenario is assessed based on the plurality of first power system operation scenario models. If the load reduction is non-zero, the first power system operation scenario model is iteratively corrected by adjusting the generator capacity or optimizing the generator layout. Specifically:

[0304] In the conventional power planning phase, with economic efficiency and basic reliability as the core objectives, an initial power scheme for the planning year is constructed. First, a basic load demand curve and peak capacity requirements for the planning year are established based on historical load data. Second, the investment and construction costs, operational constraints, and environmental indicators of conventional power sources such as thermal power, hydropower, and nuclear power are integrated, while the output fluctuation characteristics of renewable energy sources such as wind power and photovoltaics are quantified. Subsequently, a multi-objective optimization algorithm is used to find the optimal combination of power source type, installed capacity, and spatial layout under the condition of satisfying power supply reliability constraints, ensuring that the planning scheme has the lowest power supply cost and achieves power balance standards under conventional scenarios (the method can refer to existing power system power planning technologies). The conventional planning scheme for the planning year output in this phase serves as the benchmark input for subsequent typhoon scenario verification.

[0305] To address the impact of typhoon extreme weather on the power system, a verification of the conventional planning scheme for the planning year was conducted based on a typhoon monthly scenario set. The verification process was as follows: 1) Based on the wind farm output data and load data for the typhoon months of the planning year, a verification set containing Nsce typical scenarios was constructed; 2) The load reduction risk under each typhoon scenario was assessed (the method can refer to existing power system time-series operation simulation technology), and the typhoon monthly load reduction was calculated. The verification rule required that the typhoon monthly load reduction be 0; otherwise, the scheme was deemed to have failed the verification.

[0306] If the planned power supply scheme fails the typhoon month scenario verification, a multi-dimensional correction mechanism will be initiated to iteratively optimize the scheme. Correction strategies include increasing generator capacity and optimizing generator locations. The modified power supply plan needs to be re-verified under the typhoon month scenario. This verification-correction process will be repeated until the planned year's power supply plan meets the power supply reliability requirements under the typhoon month scenario.

[0307] The model output module 103 is used to output a power planning scheme that meets the power supply requirements of multiple climates during the typhoon month, based on the modified first power system operation scenario model.

[0308] The model configuration module 104 is used to configure the power supply of the power system according to the power planning scheme of the multiple climate power supply requirements.

[0309] This application employs a power system operation scenario modeling method based on the spatiotemporal characteristics of typhoons. First, it constructs multiple first-stage power system operation scenario models, which comprehensively reflect the potential impact of typhoons on power system operation. Based on these models, the load reduction risk of the power system under each scenario is assessed. If the load reduction is found to be non-zero, indicating a risk to power supply, the model is iteratively corrected by adjusting generator capacity or optimizing generator layout until the load reduction is zero, ensuring the reliability of power supply under typhoon conditions. Subsequently, based on the corrected model, a power supply planning scheme that meets the power supply requirements of multiple climates during the typhoon month is output, providing scientific guidance for power system power configuration. Finally, based on these planning schemes, the power supply of the power system is precisely controlled, thereby effectively improving the disaster resistance and operational stability of the power system under extreme weather conditions such as typhoons, ensuring the continuity and reliability of power supply.

[0310] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A method for modeling power system operation scenarios based on the spatiotemporal characteristics of typhoons, characterized in that, include: Acquire historical meteorological data, historical load data, and historical typhoon data for the power system; Based on the historical meteorological data, historical load data, and historical typhoon data, a meteorological feature matrix, a typhoon path feature matrix, and a load dataset are constructed. Based on the meteorological feature matrix and load dataset divided into three stages before, during and after the typhoon, a pre-set neural network is used to fit the mapping relationship between meteorological parameters and load to generate a meteorological-load fitting curve. Based on the typhoon path feature matrix, random typhoon paths are generated using a multivariate joint probability distribution model and Monte Carlo simulation algorithm to reconstruct the typhoon path information for the planning year, specifically including: The central pressure, moving speed and direction in the typhoon path feature matrix are discretized to divide multiple state intervals, and a multivariate joint state transition probability matrix is ​​constructed by combining the historical typhoon data. Based on the multivariate joint state transition probability matrix, a random typhoon path state sequence is generated using the Monte Carlo simulation algorithm. The random typhoon path state sequence contains the hourly state code of the typhoon. The random typhoon path state sequence is decoded, and the state code is converted into continuous spatiotemporal parameters to generate typhoon path information for the planning year; specifically, generating typhoon path information for the planning year involves: The typhoon path information is as follows: In the formula, This is the simulated typhoon path information at a certain moment. , They are states n The upper and lower limits of the typhoon's path range covered. u This is a preset value; Based on the aforementioned meteorological-load fitting curve, typhoon track information, and historical meteorological data, the monthly hourly typhoon load data and monthly hourly typhoon meteorological data for the planning year are generated, specifically as follows: Based on the typhoon path information, the actual number of days the typhoon affected the typhoon was calculated, and the actual number of days the typhoon affected the typhoon was divided into three stages: the pre-typhoon, mid-typhoon, and post-typhoon durations. Hourly meteorological data for the period before and after the typhoon are extracted from the historical meteorological data, and the average value of the hourly meteorological data is taken to generate monthly hourly meteorological data for the typhoon before and after the typhoon. If the duration of the typhoon phase is less than a preset threshold, the monthly hourly meteorological data of the typhoon in the historical meteorological data is recorded sequentially; otherwise, the combination of the meteorological data before and after the typhoon in the historical meteorological data is extracted, and the monthly hourly meteorological data of the typhoon in the typhoon is generated according to the preset mean calculation method. Based on the monthly hourly meteorological data of typhoons before and after the typhoon, the monthly hourly meteorological data of typhoons during the typhoon, and the meteorological-load fitting curve, and by regressing and fitting the planned annual peak based on the maximum value of the annual historical load data, the monthly hourly load data of typhoons is generated. Based on the typhoon path information, monthly hourly meteorological data of the typhoon, and monthly hourly load data of the typhoon, and combined with a pre-set multi-component coupling model of the typhoon wind field, several power system operation scenario models based on the spatiotemporal characteristics of typhoons are constructed, specifically: Based on the center location, speed and direction of movement of the typhoon path information, and combined with the multi-component coupling model of the typhoon wind field, the wind speed in the typhoon-affected area is decomposed into radial and moving components, and the actual hourly wind speed of the wind farm is calculated by vector superposition. Using the typhoon center as the center and the preset influence radius as the threshold, the geometric circle determination method is used to determine whether the wind farm is within the typhoon's influence range. If it is within the range, it is marked as a typhoon-affected scenario. Based on the actual hourly wind speed, and combined with the preset wind turbine cut-in wind speed, cut-out wind speed and rated power characteristic curve, the hourly output data of the typhoon-affected scenario is calculated in segments. By aligning the monthly hourly meteorological data, monthly hourly load data, and hourly power output data of typhoons in time and space, multiple operational scenario models are generated, which include typhoon path intensity, meteorological fluctuations, and load changes.

2. The power system operation scenario modeling method based on the spatiotemporal characteristics of typhoons according to claim 1, characterized in that, The calculation formula for the hourly power output data of the typhoon-affected scenario is as follows: In the formula, Represents the cut-in wind speed. This represents the cut-out wind speed. Represents the rated wind speed. This represents the active power generated by the wind turbine at its rated wind speed. for Wind farm output power at all times for Wind speed at all times , and These are the parameters for the power characteristic curve of the wind turbine.

3. The power system operation scenario modeling method based on the spatiotemporal characteristics of typhoons according to claim 1, characterized in that, The acquisition of historical meteorological data and historical typhoon data of the power system specifically includes: The historical meteorological data includes hourly data on temperature, precipitation, wind speed, and humidity. The historical typhoon data includes latitude and longitude, central pressure, speed of movement, and direction of movement.

4. A power supply configuration method based on the spatiotemporal characteristics of typhoons, characterized in that, include: According to any one of claims 1 to 3, the power system operation scenario modeling method based on the spatiotemporal characteristics of typhoons is used to construct multiple first power system operation scenario models for the power system to be controlled. Based on the multiple first power system operation scenario models, assess the load reduction risk of the power system under typhoon power supply in each scenario. If the load reduction is non-zero, iteratively correct the first power system operation scenario model by adjusting the generator capacity or optimizing the generator layout. Based on the revised first power system operation scenario model, output a power planning scheme that meets the power supply requirements of multiple climates during the typhoon month; Based on the power supply planning schemes for the various climate power supply requirements, the power supply of the power system is configured.

5. A power system operation scenario modeling device based on the spatiotemporal characteristics of typhoons, characterized in that, It includes modules for acquisition, construction, fitting, data reconstruction, and modeling. The acquisition module is used to acquire historical meteorological data, historical load data, and historical typhoon data of the power system; The construction module is used to construct a meteorological feature matrix, a typhoon path feature matrix, and a load dataset based on the historical meteorological data, historical load data, and historical typhoon data. The fitting module is used to divide the meteorological feature matrix and load dataset into three stages: before, during and after the typhoon, and to use a preset neural network to fit the mapping relationship between meteorological parameters and load to generate a meteorological-load fitting curve. The data reconstruction module is used to generate random typhoon paths based on the typhoon path feature matrix, combined with a multivariate joint probability distribution model and a Monte Carlo simulation algorithm, and to reconstruct the typhoon path information for the planning year, specifically including: The central pressure, moving speed and direction in the typhoon path feature matrix are discretized to divide multiple state intervals, and a multivariate joint state transition probability matrix is ​​constructed by combining the historical typhoon data. Based on the multivariate joint state transition probability matrix, a random typhoon path state sequence is generated using the Monte Carlo simulation algorithm. The random typhoon path state sequence contains the hourly state code of the typhoon. The random typhoon path state sequence is decoded, and the state code is converted into continuous spatiotemporal parameters to generate typhoon path information for the planning year; specifically, generating typhoon path information for the planning year involves: The typhoon path information is as follows: In the formula, This is the simulated typhoon path information at a certain moment. , They are states n The upper and lower limits of the typhoon's path range covered. u This is a preset value; Based on the aforementioned meteorological-load fitting curve, typhoon track information, and historical meteorological data, the monthly hourly typhoon load data and monthly hourly typhoon meteorological data for the planning year are generated, specifically as follows: Based on the typhoon path information, the actual number of days the typhoon affected the typhoon was calculated, and the actual number of days the typhoon affected the typhoon was divided into three stages: the pre-typhoon, mid-typhoon, and post-typhoon durations. Hourly meteorological data for the period before and after the typhoon are extracted from the historical meteorological data, and the average value of the hourly meteorological data is taken to generate monthly hourly meteorological data for the typhoon before and after the typhoon. If the duration of the typhoon phase is less than a preset threshold, the monthly hourly meteorological data of the typhoon in the historical meteorological data is recorded sequentially; otherwise, the combination of the meteorological data before and after the typhoon in the historical meteorological data is extracted, and the monthly hourly meteorological data of the typhoon in the typhoon is generated according to the preset mean calculation method. Based on the monthly hourly meteorological data of typhoons before and after the typhoon, the monthly hourly meteorological data of typhoons during the typhoon, and the meteorological-load fitting curve, and by regressing and fitting the planned annual peak based on the maximum value of the annual historical load data, the monthly hourly load data of typhoons is generated. The modeling module is used to construct multiple power system operation scenario models based on the spatiotemporal characteristics of typhoons, using the typhoon path information, monthly hourly meteorological data of typhoons, and monthly hourly load data of typhoons, combined with a preset multi-component coupling model of typhoon wind fields. Specifically: Based on the center location, speed and direction of movement of the typhoon path information, and combined with the multi-component coupling model of the typhoon wind field, the wind speed in the typhoon-affected area is decomposed into radial and moving components, and the actual hourly wind speed of the wind farm is calculated by vector superposition. Using the typhoon center as the center and the preset influence radius as the threshold, the geometric circle determination method is used to determine whether the wind farm is within the typhoon's influence range. If it is within the range, it is marked as a typhoon-affected scenario. Based on the actual hourly wind speed, and combined with the preset wind turbine cut-in wind speed, cut-out wind speed and rated power characteristic curve, the hourly output data of the typhoon-affected scenario is calculated in segments. By aligning the monthly hourly meteorological data, monthly hourly load data, and hourly power output data of typhoons in time and space, multiple operational scenario models are generated, which include typhoon path intensity, meteorological fluctuations, and load changes.

6. A power supply configuration device based on the spatiotemporal characteristics of typhoons, characterized in that, It includes a model building module, a model correction module, a model output module, and a model configuration module: The model building module is used to construct multiple first power system operation scenario models of the power system to be controlled by the power system operation scenario modeling method based on the spatiotemporal characteristics of typhoons according to any one of claims 1 to 3. The model correction module is used to assess the load reduction risk of the power system under typhoon power supply in each scenario based on the multiple first power system operation scenario models. If the load reduction is non-zero, the first power system operation scenario model is iteratively corrected by adjusting the generator capacity or optimizing the generator layout. The model output module is used to output a power planning scheme that meets the power supply requirements of multiple climates during the typhoon month, based on the corrected first power system operation scenario model. The model configuration module is used to configure the power supply of the power system according to the power planning schemes for the multiple climate power supply requirements.

Citation Information

Patent Citations

  • Double-layer evaluation method and system for evaluating resilience of elastic power distribution network

    CN114358619A

  • Source network collaborative planning method and system considering typhoon and wind power capacity credibility

    CN119627894A

  • Monte Carlo Markov chain-based strong typhoon wind speed and wind direction joint probability distribution generation method

    CN119761199A

  • Integrated energy system distributed scheduling method and system considering typhoon uncertainty

    CN120013208A