Multi-dimensional typical operation mode library generation method for high proportion of new energy and related device

Abstract

The application provides a multi-dimensional typical operation mode library generation method for high-proportion new energy and related devices, including obtaining original data for constructing a typical operation mode library, preprocessing and constructing features to obtain preprocessed data containing key features of power grid operation; constructing a multi-dimensional power grid operation state space containing new energy output dimensions based on the preprocessed data; generating candidate operation modes from the space using a multi-dimensional hierarchical sampling method; performing direct current power flow determination and constraint feasibility correction on the candidate operation modes to obtain feasible operation modes that meet power grid operation constraints; selecting a target number of typical operation modes from the feasible operation modes based on representativeness and diversity indexes to form a multi-dimensional typical operation mode library. The application can better cope with the challenges of new energy grid connection and ensure the safe, stable and economic operation of high-proportion new energy power systems by constructing a mode library through multi-dimensional hierarchical sampling, direct current power flow feasibility and representativeness optimization.

Description

Technical Field

[0001] This invention belongs to the field of power system technology, specifically relating to a method and related apparatus for generating a multi-dimensional typical operation mode library for high proportion of new energy sources. Background Technology

[0002] With increasing global emphasis on sustainable development and its goals, the penetration rate of new energy power generation technologies such as wind and solar power in the global power system continues to grow rapidly, and the role of new energy power generation in the power system is becoming increasingly important.

[0003] The integration of a high proportion of renewable energy sources has brought unprecedented challenges to the planning, operation, and control of the power system. Traditional power systems mainly consist of controllable conventional thermal power units with relatively stable operation. However, renewable energy generation is intermittent, volatile, and highly uncertain, and its output is significantly affected by weather conditions, leading to a significant increase in the complexity and risk of grid operation.

[0004] Current methods for generating typical operating modes generally have limitations. Traditional experience-based methods, based on a small number of typical scenarios, rely heavily on expert experience, are highly subjective, have limited coverage of operating scenarios, and cannot reflect the complex operating modes caused by fluctuations in renewable energy output. Methods based on statistical clustering of historical operating data heavily depend on historical data and are insufficiently representative of future new operating modes. Random scenario generation methods based on Monte Carlo sampling are computationally intensive and inefficient, and the selected scenarios cannot guarantee representativeness of the probability space. Furthermore, generating a large number of infeasible operating modes requires significant resources for correction. These limitations make it difficult to adapt to the complexity of high-proportion renewable energy power systems. Summary of the Invention

[0005] In view of this, the present invention aims to overcome the limitations of existing technologies in power systems with a high proportion of renewable energy, and to provide a method and related apparatus for generating a multi-dimensional typical operation mode library for power systems with a high proportion of renewable energy, thereby improving the intelligence level of power grid planning and operation decision-making, effectively addressing the challenges brought by renewable energy grid connection, and ensuring the safe, stable and economical operation of the power system.

[0006] To achieve the above objectives, the technical solution provided by the present invention is as follows:

[0007] In a first aspect, the present invention provides a method for generating a multi-dimensional typical operation mode library for high-proportion new energy sources, comprising the following steps:

[0008] Obtain the raw data required to build a typical operation mode library, preprocess the raw data, and construct features from the preprocessed raw data to obtain preprocessed data containing key features of power grid operation.

[0009] Based on the preprocessed data, the value range and discretization strategy of each dimension of the multidimensional power grid operation state space are determined, and a multidimensional power grid operation state space including the new energy output dimension is obtained.

[0010] A multidimensional hierarchical sampling method is used to generate candidate operating modes from the multidimensional power grid operating state space;

[0011] The candidate operating modes are modified to meet the constraints of grid operation.

[0012] Based on representativeness and diversity indicators, a target number of typical operating methods are selected from feasible operating methods to form a multi-dimensional typical operating method library.

[0013] Furthermore, the dimensions of the multidimensional power grid operating state space include at least:

[0014] From the source perspective, this includes the output range of conventional power sources and the output range of new energy sources such as wind power and photovoltaics;

[0015] The load-end dimension includes the peak load, mid-load, and valley load intervals of the total system load;

[0016] The water-related dimension includes the power output range of hydropower units during the high-water and low-water seasons based on reservoir water level and inflow.

[0017] The topology dimension includes the switching status of transmission lines and the maintenance status of generator sets.

[0018] Furthermore, a multidimensional hierarchical sampling method is used to generate candidate operating modes from the multidimensional power grid operating state space, including:

[0019] Each dimension of the multidimensional power grid operation state space is divided into several sub-intervals, and the sampling weight of each sub-interval is determined.

[0020] Sample points are drawn according to sampling weights within each sub-interval of each dimension;

[0021] The sample points extracted from each dimension are combined to form candidate operating modes that cover both normal and extreme operating scenarios.

[0022] Furthermore, the candidate operating modes are subject to constraint feasibility modifications to obtain feasible operating modes that satisfy the power grid operating constraints, including:

[0023] Perform DC power flow calculations on candidate operating modes to determine whether they meet preset constraints; preset constraints include system active power balance and line transmission power not exceeding line capacity.

[0024] If the preset constraints are not met, the line transfer distribution factor and the sensitivity of generator output to line power are calculated based on the sensitivity information of DC power flow.

[0025] Based on the sensitivity of the calculation, the output of the peak-shaving unit or the new energy unit is adjusted, and the DC power flow calculation and output adjustment are performed iteratively until the candidate operating mode meets the preset constraints and a feasible operating mode is obtained.

[0026] Furthermore, the indicators of representativeness and diversity should include at least:

[0027] Mutual information index is used to evaluate the information coverage capability of the selected typical operating modes across various dimensions of the multidimensional power grid operating state space.

[0028] The maximum and minimum distance metrics are used to ensure that the selected typical operating modes are widely distributed in multidimensional space;

[0029] The coverage metric is used to assess the extent to which selected typical operating modes cover historical operating scenarios and future predicted scenarios.

[0030] Furthermore, obtain the raw data required to build a typical runtime library, including:

[0031] Historical power grid operation data and actual power output data of new energy sources are obtained from historical data acquisition and monitoring systems and new energy power station monitoring systems.

[0032] Obtain new energy output forecast data and total system load forecast data from meteorological forecasting systems and load forecasting systems;

[0033] Obtain network topology, line parameters, transformer parameters, generator parameters, and load parameters from the power grid basic database.

[0034] Furthermore, it also includes outputting the multi-dimensional typical operation mode library in a preset standardized format, specifically including:

[0035] The default standardized format is a .dat file format readable by power grid analysis software. The .dat file must contain at least the following information: basic system information, node information, generator information, load information, line information, transformer information, and reactive power compensation equipment information.

[0036] Secondly, the present invention provides a multi-dimensional typical operation mode library generation device for high-proportion new energy sources, comprising:

[0037] The data acquisition and preprocessing module is used to acquire the raw data required to build a typical operation mode library, preprocess the raw data, and construct features from the preprocessed raw data to obtain preprocessed data containing key features of power grid operation.

[0038] The multidimensional state space construction module is used to determine the value range and discretization strategy of each dimension of the multidimensional power grid operation state space based on preprocessed data, so as to obtain a multidimensional power grid operation state space that includes the dimension of new energy output.

[0039] The multidimensional hierarchical sampling module is used to generate candidate operating modes from the multidimensional power grid operating state space using a multidimensional hierarchical sampling method.

[0040] The feasibility correction module is used to perform constraint feasibility correction on candidate operating modes to obtain feasible operating modes that meet the power grid operation constraints;

[0041] The typical operation mode selection module is used to select a target number of typical operation modes from feasible operation modes based on representativeness and diversity indicators, forming a multi-dimensional typical operation mode library.

[0042] Thirdly, the present invention provides a computer device, the device including a processor and a memory:

[0043] The memory is used to store computer programs and send the instructions of the computer programs to the processor;

[0044] The processor executes instructions from the computer program, such as the method for generating a multi-dimensional typical operating mode library for high-proportion new energy sources, as described in the first aspect.

[0045] Fourthly, the present invention provides a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, it implements a method for generating a multi-dimensional typical operating mode library for high-proportion new energy sources as described in the first aspect.

[0046] In summary, this invention provides a method for generating a multi-dimensional typical operation mode library for high-proportion renewable energy sources. The method includes acquiring the raw data required to construct the library, preprocessing and feature constructing the raw data to obtain preprocessed data containing key features of power grid operation; determining the value range and discretization strategy of each dimension of the multi-dimensional power grid operation state space based on the preprocessed data, thereby constructing a multi-dimensional power grid operation state space containing renewable energy output; generating candidate operation modes from the multi-dimensional power grid operation state space using a multi-dimensional hierarchical sampling method; performing DC power flow determination and constraint feasibility correction on the candidate operation modes to obtain feasible operation modes that meet power grid operation constraints; and selecting a target number of typical operation modes from the feasible operation modes based on representativeness and diversity indicators to form a multi-dimensional typical operation mode library. This invention, by constructing the mode library through multi-dimensional hierarchical sampling, DC power flow feasibility, and representativeness optimization, effectively overcomes the limitations of existing technologies, thereby improving the intelligence level of power grid planning and operation decision-making, better addressing the challenges of renewable energy grid integration, and ensuring the safe, stable, and economical operation of high-proportion renewable energy power systems.

[0047] The present invention also provides a multi-dimensional typical operation mode library generation device, computer equipment and computer-readable storage medium for high proportion of new energy sources, which have similar technical effects to the above methods when implemented, and will not be described in detail here. Attached Figure Description

[0048] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0049] Figure 1 A flowchart illustrating a method for generating a multi-dimensional typical operation mode library for high-proportion new energy sources, provided in an embodiment of the present invention;

[0050] Figure 2 An overall architecture diagram provided for embodiments of the present invention;

[0051] Figure 3 A flowchart of intelligent processing provided for embodiments of the present invention;

[0052] Figure 4 A data acquisition flowchart provided for embodiments of the present invention;

[0053] Figure 5 A flowchart for generating the multi-dimensional typical operation mode library provided in the embodiments of the present invention;

[0054] Figure 6 This is a block diagram of a multi-dimensional typical operation mode library generation device for high-proportion new energy sources, provided in an embodiment of the present invention.

[0055] Figure 7 This is a block diagram of a computer device provided in an embodiment of the present invention. Detailed Implementation

[0056] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0057] The following section will first introduce some of the technical terms involved in this invention.

[0058] (1) New energy output: refers to the real-time power generation of non-fossil fuel renewable energy sources such as wind power and solar photovoltaic power. Its characteristics are intermittency, volatility and uncertainty.

[0059] (2) Typical operating mode: refers to the representative set of the power grid operating status within a specific period (such as a certain day or a certain season), including generator output, load, network topology, power flow distribution, etc., which are used for power grid planning, operation analysis and safety verification.

[0060] (3) Mode library: a set of typical operating modes that have been screened and optimized to cover various operating scenarios that may occur in the future power grid, and to serve as input for simulation analysis.

[0061] (4) Multidimensional stratified sampling: a statistical sampling method that divides a complex multidimensional state space into multiple independent or partially overlapping subspaces (layers) and performs sampling within each layer to ensure the representativeness and coverage of the sample to the overall space.

[0062] (5) Constraint Feasibility: This refers to adjusting and correcting the initially generated operating mode that may not meet the power grid operation constraints (such as power balance, line capacity, and upper and lower limits of unit output) to make it meet the safety and economic requirements of actual operation.

[0063] (6) DC power flow: A simplified power flow calculation method that ignores the reactive power and voltage amplitude changes in the network and focuses only on the transmission and distribution of active power. It is often used to quickly assess the active power balance of the network and the overload of the lines.

[0064] This invention is primarily applied to power grid planning and operation auxiliary decision-making systems, power market simulation platforms, power system simulation analysis software, and power grid safety verification tools. This technology provides power grid planners and operators with an efficient, accurate, and comprehensive library of typical methods covering future operating scenarios, thereby improving the reliability of power grid planning, operational economy, and security in responding to new energy fluctuations. By integrating this method into existing power grid analysis software, its analytical capabilities for scenarios with a high proportion of new energy sources can be significantly improved, supporting the optimized operation and development of smart grids.

[0065] The following section further introduces the existing technical solutions.

[0066] 1. Traditional empirical methods based on a small number of typical scenarios (such as the traditional load-hydropower typical mode generation method):

[0067] Technical principle: This method is mainly based on historical load data and hydropower output. Through expert experience or simple statistical rules, a few representative operating scenarios (e.g., high load, low load, wet season, dry season, etc.) are selected as typical modes. Usually, only about four typical modes are generated.

[0068] Implementation method: First, cluster or segment historical load data to identify peak and off-peak load patterns; second, combine hydrological conditions to determine hydropower output patterns during the wet and dry seasons; finally, combine these patterns through manual experience to form a few operating modes.

[0069] Analysis of characteristics: This method is simple to implement, requires little computation, and is widely used in traditional power systems. However, its main drawback is its reliance on expert experience, strong subjectivity, and limited coverage of operational scenarios. In the context of high renewable energy penetration, it cannot reflect the complex operational modes caused by fluctuations in renewable energy output, and its coverage of future grid capacity availability, reactive power demand, and over-limit risks is severely inadequate.

[0070] 2. Methods based on statistical clustering of historical operating data (such as the generation method of typical patterns of historical data clustering analysis): Technical principle: This method uses a large amount of historical data collection and monitoring system data to extract key operating parameters such as generator output, load, and power flow as features. It then uses clustering algorithms (such as K-means and hierarchical clustering) to perform pattern recognition on historical operating data, thereby finding several statistically significant typical operating patterns.

[0071] Implementation method: First, acquire data from the power grid historical operation database for several months or even several years from the data acquisition and monitoring system; second, clean, normalize, and perform feature engineering on the data to transform the operating parameters into multi-dimensional feature vectors; then, select an appropriate clustering algorithm and number of clusters to cluster the feature vectors; finally, use the samples of each cluster center as representatives of typical operating modes.

[0072] Feature Analysis: This method objectively reflects the actual historical operating conditions of the power grid, is data-driven, and reduces subjectivity. However, its limitation lies in its heavy reliance on historical data. Historical data may not effectively capture new operating modes that may emerge in the future, such as the characteristics of new energy output (e.g., extreme output, steep slope changes) and changes in grid topology, resulting in insufficient representativeness of the generated typical mode library for future situations.

[0073] 3. Random scene generation methods based on Monte Carlo sampling (such as the Monte Carlo random scene generation method):

[0074] Technical principle: This method models uncertainties in power grid operation (such as new energy output, load fluctuations, unit failures, etc.) as random variables, and uses Monte Carlo sampling technology to randomly generate a large number of operating scenarios that satisfy the probability distribution of each random variable.

[0075] Implementation method: First, establish a probability distribution model for uncertainties such as wind power, photovoltaic output and load; second, use a random number generator to perform a large number of independent samplings in these distributions, and generate a complete power grid operation scenario for each sampling; finally, filter or aggregate the tens of thousands or even hundreds of thousands of scenarios generated.

[0076] Characteristics Analysis: The Monte Carlo method can theoretically cover all possible operating scenarios, and it is particularly adept at handling complex combinations of multiple random variables. However, its main drawbacks are its enormous computational cost and low efficiency. The large number of generated scenarios makes subsequent screening and analysis extremely costly, and it is difficult to guarantee that the selected few scenarios represent the entire probability space. Furthermore, simple random sampling often generates a large number of infeasible operating methods, requiring significant computational resources for feasibility correction.

[0077] However, the above-mentioned existing technical solutions have the following drawbacks:

[0078] 1. Disadvantages of the traditional load-hydropower generation method:

[0079] This method focuses only on a few traditional dimensions such as load and hydropower, completely ignoring the significant impact of fluctuations in renewable energy output. Renewable energy output (such as wind and solar power) exhibits significant intermittency, volatility, and regionality. Its high penetration rate not only alters the power system's power supply structure but also profoundly affects network power flow distribution, available transmission capacity, reactive power demand, and voltage stability. Traditional, typical methods cannot effectively capture these complex changes, potentially leading to the underreporting of risks such as network overreach and voltage instability caused by renewable energy during planning and operational analysis, severely reducing the accuracy and reliability of grid safety verification. This experience-based enumeration method has extremely poor representativeness and adaptability in high-uncertainty environments.

[0080] 2. Disadvantages of the typical generation method for historical data cluster analysis:

[0081] While data-driven, it is essentially based on inductive reasoning from past experience. Faced with the rapidly changing landscape of new energy development, historical operational data may not fully reflect potential extreme new energy output scenarios (e.g., extreme high output combined with low load, or extreme low output combined with high load). Furthermore, continuous optimization of the power grid topology and the introduction of energy storage technologies may lead to significant differences between future operating patterns and historical patterns. Therefore, methodologies generated based on historical data clustering have limitations in predicting and covering future operating trends, making it difficult to effectively address new risks and challenges.

[0082] 3. Disadvantages of the Monte Carlo random scene generation method:

[0083] While this method can theoretically cover all possible scenarios, it suffers from serious efficiency problems in practical applications. Random sampling generates a large number of invalid or infeasible operating modes, requiring time-consuming power flow calculations and feasibility corrections. When dealing with large-scale power systems, this "brute-force" sampling method has excessively high computational complexity, making it difficult to generate a moderately large and representative library of typical modes within a limited time. Furthermore, efficiently selecting a few of the most representative typical modes from massive random scenarios remains an open and challenging problem; existing selection methods often fail to guarantee the globally optimal representativeness of the selection results.

[0084] The present invention aims to overcome the limitations of existing technologies in the face of high-proportion renewable energy power systems, and provides a method and related apparatus for generating a typical operation mode library that is efficient, accurate, highly representative and with high coverage. The present invention will be described below with reference to some embodiments.

[0085] Please see Figure 1 This embodiment provides a method for generating a multi-dimensional typical operation mode library for high-proportion new energy sources, including the following steps:

[0086] S11: Obtain the raw data required to build a typical operation mode library, preprocess the raw data, and construct features from the preprocessed raw data to obtain preprocessed data containing key features of power grid operation.

[0087] It should be noted that the raw data refers to the basic data for constructing the typical operation mode library, including power output data of new energy sources (wind power, photovoltaic, etc.), load data (active / reactive load), power grid topology parameters (line impedance, node connection relationship), power output data of conventional power sources, etc.

[0088] Preprocessing refers to the process of cleaning the raw data (handling missing values ​​and outliers, such as unreasonable data such as sudden increases or decreases in wind power output), standardizing (converting data with different dimensions to a unified scale, such as normalizing new energy output to the 0-1 range), and time-series alignment (ensuring consistency in the time dimension of each data point).

[0089] Feature construction involves extracting key indicators that reflect the characteristics of power grid operation from the raw data, such as the fluctuation rate of new energy output (describing the randomness of new energy), the peak-valley difference of load (reflecting load fluctuation), the voltage deviation rate of key nodes, and the power flow margin of lines.

[0090] S12: Based on the preprocessed data, determine the value range and discretization strategy of each dimension of the multidimensional power grid operation state space to obtain a multidimensional power grid operation state space that includes the dimension of new energy output.

[0091] It should be noted that the multidimensional power grid operation state space is an abstract space constructed using key factors affecting power grid operation as dimensions (such as renewable energy output, load level, conventional power output, network topology mode, etc.). Each combination of dimensions corresponds to a specific power grid operation state.

[0092] The range of values ​​for a dimension is the physical or engineering allowable range for each dimension. For example, the range of values ​​for the wind power output dimension is 0 to the rated capacity, and the range of values ​​for the load dimension is from the minimum load to the maximum load.

[0093] Discretization strategies are methods that divide the continuous range of dimensional values ​​into finite discrete intervals (such as equidistant discretization and cluster discretization based on probability distribution). For example, wind power output can be divided into 10 intervals based on 10% of the rated capacity.

[0094] With a high proportion of renewable energy integrated into the grid, the grid's operating status is influenced by multiple factors, including renewable energy output and load, exhibiting high-dimensional characteristics. By clarifying the value range and discretization strategy for each dimension, the continuous and complex operating status can be transformed into a quantifiable and operable discrete space. This ensures that subsequent sampling systems can cover all possible operating scenarios, especially highlighting the renewable energy output dimension, to adapt to the characteristics of a grid with a high proportion of renewable energy.

[0095] S13: A multidimensional hierarchical sampling method is used to generate candidate operating modes from the multidimensional power grid operating state space.

[0096] It should be noted that multidimensional stratified sampling is a method in which each dimension is divided into "layers" according to discretized intervals in a multidimensional state space, and samples are randomly drawn from the cross-combination of each layer (for example, sampling the cross region between the third layer of wind power output and the second layer of load).

[0097] Candidate operating modes are an initial sample set obtained through sampling that may reflect the operating status of the power grid, including specific values ​​of parameters in each dimension (such as 20% wind power output, 60% load, etc.).

[0098] Traditional random sampling may miss some critical states (such as extreme scenarios of high output and high load in new energy sources). Multidimensional stratified sampling ensures that each subspace (the intersection of layers) is covered by samples by uniformly sampling in discrete layers of each dimension. It can capture both common operating states and extreme scenarios, improving the representativeness of candidate samples to the overall state space. It is especially suitable for the characteristics of large fluctuations in new energy output and complex scenarios.

[0099] S14: Perform constraint feasibility modification on the candidate operating modes to obtain feasible operating modes that meet the power grid operation constraints.

[0100] It should be noted that the constraint feasibility correction is based on DC power flow determination. DC power flow determination is a method that quickly determines whether the active power flow of lines and the power balance of nodes in candidate operating modes meet the constraints based on the DC power flow model (ignoring reactive power and voltage amplitude changes to simplify power flow calculation).

[0101] Constraint feasibility modification is the process of adjusting controllable variables (such as conventional power output and renewable energy output limits) to make a candidate mode meet grid constraints when the candidate mode does not meet grid constraints (such as line power flow exceeding rated capacity or node power imbalance).

[0102] Power grid operation constraints include physical or safety limitations such as line transmission capacity constraints, upper and lower limits of power output constraints, and node power balance constraints.

[0103] Candidate operating modes may have physical infeasibility (such as line overload) and need to be verified through power flow calculations. DC power flow calculations are faster than AC power flow calculations, making them suitable for rapid screening of large-scale candidate samples. The correction of infeasible samples is achieved by fine-tuning controllable parameters to ensure that safety constraints are met while retaining the original scenario characteristics, thus ensuring the engineering practicality of the final samples.

[0104] S15: Based on representativeness and diversity indicators, select the target number of typical operating modes from the feasible operating modes to form a multi-dimensional typical operating mode library.

[0105] It should be noted that representativeness indicators measure the degree of similarity between the sample and the statistical characteristics (such as mean, variance, probability distribution) of the overall operating status, such as the deviation rate between the sample's new energy output distribution and the original data distribution.

[0106] Diversity metrics measure the degree of difference between samples and avoid redundancy, such as the mean Euclidean distance between different samples in a multidimensional space.

[0107] The typical operating mode library is a collection of operating modes that has been finally selected, is concise in number, and can cover the main operating scenarios.

[0108] This embodiment provides a method for generating a multi-dimensional typical operation mode library for high-proportion renewable energy grids. This method constructs a multi-dimensional operating state space with renewable energy output as the core dimension, overcoming the limitations of traditional methods that fail to adequately consider the renewable energy dimension. Through multi-dimensional hierarchical sampling, it achieves uniform coverage of the high-dimensional, highly volatile state space, solving the problem of random sampling easily missing extreme scenarios. Combined with rapid DC power flow determination and constraint feasibility correction, it ensures sample feasibility while maintaining computational efficiency, avoiding redundancy of invalid samples. Finally, a selection strategy based on representativeness and diversity indicators balances the conciseness of the typical library with scenario coverage. Overall, it forms a closed-loop method from data processing to space construction, sampling verification, and library generation, specifically adapted to the operating characteristics of high-proportion renewable energy grids, improving the typical operation mode library's ability to characterize complex scenarios and its engineering practicality.

[0109] Figure 2 The overall architecture for implementing the method based on the above embodiments includes data acquisition, preprocessing, state space, intelligent processing, result output to the analysis system, and storage. The following describes this in conjunction with some embodiments of the present invention.

[0110] In one embodiment of the present invention, the dimensions of the multidimensional power grid operating state space include at least:

[0111] From the source perspective, this includes the output range of conventional power sources and the output range of new energy sources such as wind power and photovoltaics;

[0112] The load-end dimension includes the peak load, mid-load, and valley load intervals of the total system load;

[0113] The water-related dimension includes the power output range of hydropower units during the high-water and low-water seasons based on reservoir water level and inflow.

[0114] The topology dimension includes the switching status of transmission lines and the maintenance status of generator sets.

[0115] In this embodiment, a comprehensive description of the operational characteristics of a high-proportion renewable energy power system is achieved by constructing a four-dimensional collaborative power grid operation state space of "source-load-water-topology". Specifically, the source dimension covers both the output range of conventional power sources (including base load and peak-shaving unit output ranges) and the output ranges of renewable energy sources such as wind and solar power (divided into high, medium, and low levels based on actual output), accurately capturing the impact of renewable energy output fluctuations on power grid operation. The load dimension, based on the system load timing characteristics, subdivides the total load into three intervals: peak load, mid-load, and valley load, matching the load demand characteristics of different time periods. The water dimension, combined with natural conditions such as reservoir water level and inflow, divides the output of hydropower units into two core intervals: high-water season and low-water season, taking into account the regulating function of hydropower in the system. The topology dimension focuses on the switching status of transmission lines (such as the operation / maintenance mode of key inter-regional lines) and the maintenance status of generator units (such as planned maintenance / emergency shutdown of main units), reflecting the constraints of dynamic changes in the power grid's physical structure on the operating mode. This dimensional system can fully cover conventional operating scenarios (such as medium output of new energy + mid-load + normal topology), extreme operating scenarios (such as full / zero power generation of new energy + peak and valley loads + critical line maintenance), and potential risk scenarios (such as sudden drop in new energy output + sudden increase in load), ensuring that the final typical mode library can reflect the complex operating modes of a high proportion of new energy power grids.

[0116] In one embodiment of the present invention, a multidimensional hierarchical sampling method is used to generate candidate operating modes from a multidimensional power grid operating state space, including:

[0117] S21: Divide each dimension of the multidimensional power grid operation state space into several sub-intervals and determine the sampling weight of each sub-interval.

[0118] Each dimension of the multidimensional power grid operation state space (source end, load end, water end, topology) is further divided into sub-intervals. For example, the "high" interval of renewable energy output is subdivided into two sub-intervals: "90%-100% rated output" and "70%-90% rated output". Sampling weights are determined based on the distribution of historical operation data. For example, higher weights are assigned to frequently occurring regular sub-intervals (such as renewable energy 50%-70% rated output, mid-load load) to ensure basic coverage, while the weights are appropriately increased for low-probability extreme sub-intervals (such as renewable energy 0%-10% rated output, extreme load peak) to avoid omissions.

[0119] S22: Sample points are drawn according to sampling weights within each sub-interval of each dimension.

[0120] Sample points are randomly selected within each dimensional sub-interval according to a set weight. Multi-dimensional sampling tasks are processed synchronously through parallel computing technology, which greatly shortens the sampling time and ensures that each sub-interval is covered by samples, thus avoiding sampling bias.

[0121] S23: Combine the sample points extracted from each dimension to form candidate operating modes that cover both normal and extreme operating scenarios.

[0122] The sample points extracted from each dimension are combined, such as "high output of new energy + peak load + hydropower high water season + normal topology" and "low output of new energy + valley load + hydropower low water season + critical line maintenance", to finally form a candidate operation mode that covers both normal and extreme scenarios and has a moderate number (such as 1,000-2,000 sets).

[0123] Compared with traditional random sampling, the multidimensional stratified sampling method for generating candidate operation modes provided in this embodiment can significantly reduce invalid samples (such as samples with high repetition or that do not conform to the actual operation rules of the power grid), reduce the resource consumption of subsequent feasibility correction, and at the same time ensure the uniform distribution of samples in multidimensional space through stratified design.

[0124] In one embodiment of the present invention, the candidate operating modes are subjected to constraint feasibility modification to obtain feasible operating modes that satisfy the power grid operating constraints, including:

[0125] S31: Perform DC power flow calculation on the candidate operating modes to determine whether they meet the preset constraints; the preset constraints include system active power balance and line transmission power not exceeding line capacity.

[0126] DC power flow calculations are performed on candidate operating modes, focusing on two key constraints: first, system active power balance, meaning the total generating power must equal the sum of the total load power and network losses; and second, line transmission power constraint, meaning the calculated line power must not exceed the line's rated transmission capacity. The DC power flow calculation employs sparse matrix operation optimization, enabling rapid solution for power flow in large-scale power grids (e.g., those with thousands of nodes), with single calculation time controlled within milliseconds, achieving efficient preliminary screening of batch of candidate modes.

[0127] S32: If the preset constraints are not met, the line transfer distribution factor and the sensitivity of generator output to line power are calculated based on the sensitivity information of DC power flow.

[0128] If the candidate method does not meet the above constraints, sensitivity information is calculated based on the DC power flow results, including the line transfer distribution factor (PTDF, which reflects the degree of influence of generator output change on line power) and the sensitivity matrix of generator output on line power. The key unit that causes the constraint to exceed the limit is accurately located by sensitivity ranking (such as a peak-shaving unit output too high, causing line overload).

[0129] S33: Based on the calculated sensitivity, adjust the output of peak-shaving units or new energy units, iteratively execute DC power flow calculation and output adjustment until the candidate operating mode meets the preset constraints and obtains a feasible operating mode.

[0130] Prioritize adjusting the output of peak-shaving units or new energy units (new energy units should be selected from stations with regulation capabilities). Each adjustment should be made in small steps (e.g., 5% of the rated output of the unit). After adjustment, re-execute the DC power flow calculation to determine whether the constraints are met. If they are still not met, repeat the sensitivity calculation and output adjustment steps until the candidate mode meets the constraints or reaches the preset maximum number of iterations (e.g., 10 times) to finally obtain a feasible operating mode.

[0131] In this embodiment, the feasibility efficiency is greatly improved by using the "determine first, then correct" logic. Compared with the traditional AC power flow correction method, the processing efficiency is improved. At the same time, the sensitivity-driven precise adjustment reduces the number of invalid adjustments and ensures that the corrected feasible method conforms to the actual operation law of the power grid.

[0132] like Figure 3 As shown, Figure 3 The intelligent processing flow designed according to the aforementioned embodiments includes: first, reading the basic data of the power grid and the forecast data of new energy sources, constructing a space (defining the value range and discretization strategy of each dimension (new energy output, load, hydropower unit output, topology status, etc.)) to generate a candidate set M; then, processing the candidates in a loop, if there are unprocessed candidates, performing power flow determination and constraint checks, if the constraints are not met, performing feasibility correction and determining whether convergence is achieved, if converged, adding it to the feasible mode set, otherwise marking it as infeasible; after there are no unprocessed candidates, calculating the indicators of the feasible mode set, filtering N sets, and finally outputting the mode library; at the end of the process, feedback is also given to the construction space or candidate set generation stage to form iterative optimization logic.

[0133] Furthermore, non-uniform stratified sampling based on historical data distribution can be introduced to focus sampling on high-probability areas while retaining coverage of low-probability extreme areas. Sparse matrix techniques are employed to optimize DC power flow calculations; commonly used sensitivity matrices are pre-calculated; and efficient iterative convergence strategies are designed to avoid unnecessary iterations. The feasibility refinement process for candidate methods is naturally suited for parallel computing and can be accelerated using multi-core CPUs or distributed systems. Heuristic algorithms or machine learning-based clustering / dimensionality reduction techniques are used to improve screening efficiency and result quality.

[0134] In one embodiment of the present invention, the representativeness and diversity indicators include at least:

[0135] Mutual information index is used to evaluate the information coverage capability of the selected typical operating modes across various dimensions of the multidimensional power grid operating state space.

[0136] The maximum and minimum distance metrics are used to ensure that the selected typical operating modes are widely distributed in multidimensional space;

[0137] The coverage metric is used to assess the extent to which selected typical operating modes cover historical operating scenarios and future predicted scenarios.

[0138] In this embodiment, the mutual information index assesses the coverage of key states in each dimension by calculating the information correlation between the selected typical methods and the characteristics of each dimension (source end, load end, water end, and topology) in the multidimensional state space. A higher mutual information value indicates that the typical method reflects the information of that dimension more fully, avoiding the omission of key states in a certain dimension (such as renewable energy output). The maximum and minimum distance index calculates the Euclidean distance between candidate methods in the multidimensional space, selecting the method with the largest distance from the set of selected methods each time to add to the typical method library. This ensures that the selected typical methods are widely distributed in the multidimensional space, avoiding excessive concentration of methods in a certain local scenario (such as a large amount of conventional renewable energy output + load side load), while also ensuring coverage of extreme scenarios. The coverage rate index assesses the coverage ratio of typical methods to actual operating scenarios by comparing the overlap between the selected typical methods and historical operating scenarios (such as typical source-load matching patterns in the past 3 years) and future predicted scenarios (such as extreme renewable energy output scenarios predicted by weather). The coverage rate needs to reach above 85% to ensure that the typical method library can support the summarization of historical patterns and the prediction of future trends. When these three types of indicators are applied in synergy, a subset of methods with complete information coverage is first selected using the mutual information indicator. Then, the spatial distribution of this subset is optimized using the maximum and minimum distance indicators. Finally, the coverage rate indicator is used to verify the coverage effect on actual scenarios, forming a closed loop of "screening-optimization-verification". This indicator system replaces the subjectivity of traditional expert experience screening, ensuring that the final selected target number of typical methods (e.g., 50-100 sets) is both concise and can reflect the operating characteristics of high-proportion renewable energy power grids to the greatest extent, providing a reliable scenario foundation for downstream power flow calculation, stability analysis, risk assessment, and other work.

[0139] In one embodiment of the present invention, obtaining the raw data required to construct a typical runtime library includes:

[0140] S41: Obtain historical power grid operation data and actual power output data of new energy from historical data acquisition and monitoring systems and new energy power station monitoring systems.

[0141] Historical and real-time operational data are used to analyze the historical distribution patterns of various dimensions (such as new energy output and load), providing a basis for state space dimension division and sampling weight setting.

[0142] S42: Obtain new energy output forecast data and total system load forecast data from meteorological forecasting system and load forecasting system.

[0143] Predictive data is used to cover potential future operating scenarios, avoiding the typical approach library being limited to historical patterns.

[0144] S43: Obtain network topology, line parameters, transformer parameters, generator parameters, and load parameters from the power grid basic database.

[0145] The basic power grid parameter data is then used for subsequent power flow calculations and constraint determination.

[0146] After data collection in this embodiment, preliminary processing is required: First, format conversion, which converts non-standardized data from different data sources (such as binary data from SCADA systems and CSV data from new energy power plants) into JSON format for easier processing by subsequent modules; Second, data verification, which ensures data integrity and consistency by marking missing values ​​(marking missing timestamp data as "missing") and filtering outliers (removing data that exceeds a reasonable range, such as data where new energy output exceeds 110% of the rated capacity).

[0147] Please see Figure 4 , Figure 4 A data acquisition process is shown, which includes: after starting the acquisition, selecting the data source; if it is a data acquisition and monitoring system, accessing the database; if it is a prediction system, calling the interface; if it is a topology system, reading the configuration; then uniformly cleaning and formatting, and then verifying and deduplicating, and finally completing the data storage.

[0148] In one embodiment of the present invention, the method further includes outputting the multidimensional typical operation mode library according to a preset standardized format, specifically including:

[0149] The default standardized format is a .dat file format readable by power grid analysis software. The .dat file must contain at least the following information: basic system information, node information, generator information, load information, line information, transformer information, and reactive power compensation equipment information.

[0150] In this embodiment, through standardized file generation and optimized output, the typical method library is made "plug and play," meeting the direct usage requirements of downstream power grid analysis systems. The output format adopts the .dat file format commonly used in power grid analysis software. The file content includes seven types of core information: basic system information (system frequency 50Hz, reference voltage 220kV / 500kV), ensuring a consistent calculation benchmark across different analysis scenarios; node information (node ​​number, node name, node type such as PQ node / PV node, reference voltage), clarifying the basic attributes of power grid nodes; generator information (generator number, node location, active power output, reactive power output, active / reactive power output upper and lower limits, ramp rate), supporting the analysis of unit output regulation capabilities; and negative... Load information (load node number, active load, reactive load) reflects the distribution characteristics of the system load; line information (line start node, end node, resistance, reactance, susceptance, rated transmission capacity, switching status) ensures the accuracy of power flow calculation; transformer information (transformer number, start node, end node, rated transformation ratio, actual transformation ratio, switching status) covers the dynamic adjustment of the power grid topology; reactive power compensation equipment information (equipment number, node, equipment type such as capacitor / reactor, compensation capacity, switching status) supplements the basic data for reactive power balance analysis of the system. To improve output efficiency and practicality, three optimization strategies are adopted: first, batch output, generating .dat files for all typical methods at once (e.g., 50 methods corresponding to 50 .dat files), reducing the number of file operations; second, data compression, using GZIP compression for method libraries with large file sizes (e.g., containing more than 100 methods), reducing disk storage usage and data transmission time; and third, asynchronous processing, with the output process executed in parallel with other system tasks (e.g., subsequent data backup), avoiding blocking the main process. Before output, data consistency verification must be performed to check whether the generator output is within the upper and lower limits, whether the line transmission capacity matches the parameters, etc., to ensure that the file is error-free. In addition, a standardized API interface is provided, allowing downstream power grid analysis systems to directly obtain .dat files through the interface without manual download and import, further improving integration efficiency.

[0151] Please see Figure 5 , Figure 5The process for generating a multi-dimensional typical operation mode library for high-proportion new energy sources is as follows: First, a startup process is initiated, and a candidate set M is generated by constructing a spatial framework. Then, candidates are processed iteratively. If any unprocessed candidates exist, power flow determination and constraint checks are performed. If constraints are not met, feasibility modifications are made, and convergence is assessed. If convergence occurs, the mode is added to the feasible mode set; otherwise, it is marked as infeasible. Once no unprocessed candidates remain, the feasible mode set is evaluated for indicators, and N sets are selected, ultimately outputting the mode library. The process also feeds back to the spatial framework construction or candidate set generation stage, forming an iterative optimization logic. Sampling can utilize sequential sampling or low-difference sequences; feasibility analysis can use AC power flow or (mixed) integer linear programming; selection can use clustering or entropy-based indicators; and edge deployment and market data access are also supported.

[0152] Based on the same inventive concept, this application also provides an apparatus for generating a multi-dimensional typical operation mode library for high-proportion new energy sources, which is used to implement the above-mentioned method for generating such a library. The solution provided by this apparatus is similar to the solution described in the above-described method. Therefore, the specific limitations in the embodiments of the apparatus for generating a multi-dimensional typical operation mode library for high-proportion new energy sources provided below can be found in the limitations of the method for generating such a library above, and will not be repeated here.

[0153] Please see Figure 6 This invention also provides a device for generating a multi-dimensional typical operation mode library for high-proportion new energy sources, comprising:

[0154] The data acquisition and preprocessing module is used to acquire the raw data required to build a typical operation mode library, preprocess the raw data, and construct features from the preprocessed raw data to obtain preprocessed data containing key features of power grid operation.

[0155] This module establishes connections with different data sources and retrieves data using standardized data interfaces (such as industrial control interfaces, fieldbus interfaces, and network service interfaces). For historical data, it synchronizes data from databases or file systems periodically via batch processing tasks. For forecast data, it connects to corresponding forecasting services. A unified data interface protocol is designed, such as a data exchange format based on a common data format, to adapt to the differences in various data sources. Data fields are clearly defined, including metadata such as timestamps, device IDs, measured values, and units.

[0156] The multidimensional state space construction module is used to determine the value range and discretization strategy of each dimension of the multidimensional power grid operation state space based on preprocessed data, so as to obtain a multidimensional power grid operation state space that includes the dimension of new energy output.

[0157] The multidimensional hierarchical sampling module is used to generate candidate operating modes from the multidimensional power grid operating state space using a multidimensional hierarchical sampling method.

[0158] The feasibility correction module is used to perform constraint feasibility correction on candidate operating modes to obtain feasible operating modes that meet the power grid operation constraints;

[0159] The typical operation mode selection module is used to select a target number of typical operation modes from feasible operation modes based on representativeness and diversity indicators, forming a multi-dimensional typical operation mode library.

[0160] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0161] Reference Figure 7 The present invention also provides a computer device, including: a memory and a processor, and a computer program stored in the memory. When the computer program is executed on the processor, it implements the method for generating a multi-dimensional typical operating mode library for high-proportion new energy sources as described in any of the above methods.

[0162] The computer device may be a desktop computer, laptop, handheld computer, or cloud server, etc. This computer device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that... Figure 7 The examples of computer devices are merely examples and do not constitute a limitation on computer devices. They may include more or fewer components than shown in the illustration, or combinations of certain components, or different components. For example, they may also include input / output devices, network access devices, etc.

[0163] The processor referred to can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0164] In some embodiments, the memory may be an internal storage unit of the computer device, such as a hard drive or RAM. In other embodiments, the memory may be an external storage device of the computer device, such as a plug-in hard drive, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card. Furthermore, the memory may include both internal and external storage units of the computer device. The memory is used to store the operating system, applications, boot loader, data, and other programs, such as the program code of the computer program. The memory can also be used to temporarily store data that has been output or will be output.

[0165] This invention also provides a computer-readable storage medium storing a computer program thereon. When the computer program is run by a processor, it implements the method for generating a multi-dimensional typical operating mode library for high-proportion new energy sources as described in any of the above methods.

[0166] In this embodiment, if the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying computer program code to a photographing device / terminal device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electrical carrier signals or telecommunication signals.

[0167] This invention provides a computer program product, including a computer program that, when executed by a processor, implements the method for generating a multi-dimensional typical operating mode library for high-proportion new energy sources as described in any of the above methods.

[0168] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0169] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0170] In the embodiments disclosed in this application, it should be understood that the disclosed devices / terminal equipment and methods can be implemented in other ways. For example, the device / terminal equipment embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling or direct coupling or communication connection may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0171] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for generating a multi-dimensional typical operation mode library for high-proportion new energy sources, characterized in that, Includes the following steps: Obtain the raw data required to construct a typical operation mode library, preprocess the raw data, and construct features from the preprocessed raw data to obtain preprocessed data containing key features of power grid operation. Based on the preprocessed data, the value range and discretization strategy of each dimension of the multidimensional power grid operation state space are determined, and the multidimensional power grid operation state space including the new energy output dimension is obtained. A multidimensional hierarchical sampling method is used to generate candidate operating modes from the multidimensional power grid operating state space; The candidate operating modes are modified to meet the constraints of grid operation to obtain feasible operating modes. Based on representativeness and diversity indicators, a target number of typical operating modes are selected from the feasible operating modes to form a multi-dimensional typical operating mode library; A multidimensional hierarchical sampling method is used to generate candidate operating modes from the multidimensional power grid operating state space, including: Each dimension of the multidimensional power grid operation state space is divided into several sub-intervals, and the sampling weight of each sub-interval is determined. Sample points are drawn within each sub-interval of each dimension according to the sampling weights. The sample points extracted from each dimension are combined to form candidate operating modes that cover both normal and extreme operating scenarios; The candidate operating modes are subjected to constraint feasibility modification to obtain feasible operating modes that satisfy the power grid operating constraints, including: DC power flow calculations are performed on the candidate operating modes to determine whether they meet preset constraints; the preset constraints include system active power balance and line transmission power not exceeding line capacity; If the preset constraints are not met, the line transfer distribution factor and the sensitivity of generator output to line power are calculated based on the sensitivity information of DC power flow. Based on the sensitivity of the calculation, the output of the peak-shaving unit or the new energy unit is adjusted, and the DC power flow calculation and output adjustment are iteratively executed until the candidate operating mode meets the preset constraints, and a feasible operating mode is obtained. The representativeness and diversity indicators include at least the following: Mutual information index is used to evaluate the information coverage capability of selected typical operating modes across various dimensions of the multidimensional power grid operating state space. The maximum and minimum distance metrics are used to ensure that the selected typical operating modes are widely distributed in multidimensional space; The coverage metric is used to assess the extent to which the selected typical operating modes cover historical operating scenarios and future predicted scenarios.

2. The method for generating a multi-dimensional typical operation mode library for high-proportion new energy sources according to claim 1, characterized in that, The dimensions of the multidimensional power grid operating state space include at least: From the source perspective, this includes the output range of conventional power sources and the output range of new energy sources such as wind power and photovoltaics; The load-end dimension includes the peak load, mid-load, and valley load intervals of the total system load; The water-related dimension includes the power output range of hydropower units during the high-water and low-water seasons based on reservoir water level and inflow. The topology dimension includes the switching status of transmission lines and the maintenance status of generator sets.

3. The method for generating a multi-dimensional typical operation mode library for high-proportion new energy sources according to claim 1, characterized in that, Obtain the raw data required to build a typical runtime library, including: Historical power grid operation data and actual power output data of new energy sources are obtained from historical data acquisition and monitoring systems and new energy power station monitoring systems. Obtain new energy output forecast data and total system load forecast data from meteorological forecasting systems and load forecasting systems; Obtain network topology, line parameters, transformer parameters, generator parameters, and load parameters from the power grid basic database.

4. The method for generating a multi-dimensional typical operation mode library for high-proportion new energy sources according to claim 1, characterized in that, This also includes outputting the multidimensional typical operation mode library in a preset standardized format, specifically including: The preset standardized format is a .dat file format readable by power grid analysis software. The .dat file contains at least basic system information, node information, generator information, load information, line information, transformer information, and reactive power compensation equipment information.

5. A device for generating a multi-dimensional typical operation mode library for high-proportion new energy sources, characterized in that: include: The data acquisition and preprocessing module is used to acquire the raw data required to build a typical operation mode library, preprocess the raw data, and construct features on the preprocessed raw data to obtain preprocessed data containing key features of power grid operation. A multidimensional state space construction module is used to determine the value range and discretization strategy of each dimension of the multidimensional power grid operation state space based on the preprocessed data, so as to obtain the multidimensional power grid operation state space including the new energy output dimension. A multidimensional hierarchical sampling module is used to generate candidate operating modes from the multidimensional power grid operating state space using a multidimensional hierarchical sampling method; The feasibility correction module is used to perform constraint feasibility correction on the candidate operating modes to obtain feasible operating modes that meet the power grid operation constraints; The typical operation mode screening module is used to screen out a target number of typical operation modes from the feasible operation modes based on representativeness and diversity indicators, forming a multi-dimensional typical operation mode library; A multidimensional hierarchical sampling method is used to generate candidate operating modes from the multidimensional power grid operating state space, including: Each dimension of the multidimensional power grid operation state space is divided into several sub-intervals, and the sampling weight of each sub-interval is determined. Sample points are drawn within each sub-interval of each dimension according to the sampling weights. The sample points extracted from each dimension are combined to form candidate operating modes that cover both normal and extreme operating scenarios; The candidate operating modes are subjected to constraint feasibility modification to obtain feasible operating modes that satisfy the power grid operating constraints, including: DC power flow calculations are performed on the candidate operating modes to determine whether they meet preset constraints; the preset constraints include system active power balance and line transmission power not exceeding line capacity; If the preset constraints are not met, the line transfer distribution factor and the sensitivity of generator output to line power are calculated based on the sensitivity information of DC power flow. Based on the sensitivity of the calculation, the output of the peak-shaving unit or the new energy unit is adjusted, and the DC power flow calculation and output adjustment are iteratively executed until the candidate operating mode meets the preset constraints, and a feasible operating mode is obtained. The representativeness and diversity indicators include at least the following: Mutual information index is used to evaluate the information coverage capability of selected typical operating modes across various dimensions of the multidimensional power grid operating state space. The maximum and minimum distance metrics are used to ensure that the selected typical operating modes are widely distributed in multidimensional space; The coverage metric is used to assess the extent to which the selected typical operating modes cover historical operating scenarios and future predicted scenarios.

6. A computer device, characterized in that, The device includes a processor and a memory: The memory is used to store computer programs and send the instructions of the computer programs to the processor; The processor executes, according to the instructions of the computer program, a method for generating a multi-dimensional typical operating mode library for high-proportion new energy sources as described in any one of claims 1-4.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a processor, implements a method for generating a multi-dimensional typical operating mode library for high-proportion new energy sources as described in any one of claims 1-4.

Images