Power system dispatching methods, devices, equipment, and media based on line transmission quota-based tiered switching.

Abstract

This application discloses a power system dispatching method, apparatus, equipment, and medium based on line transmission quota tiered switching, relating to the field of power system operation technology. The method includes: constructing a continuous-time economic dispatching model and converting it into a finite-dimensional linear programming model through Bernstein polynomial interpolation; performing power flow calculation and verification on the sample set; combining historical data of renewable energy and loads; determining feature centers by clustering operation mode types; using a support vector machine classifier to identify the stability-instability boundary; eliminating unstable samples; dividing the power region of the remaining samples into grids; setting the maximum power value of each grid as the line transmission quota; calculating the distance between the renewable energy and load curves and each feature center over time; determining the category switching time point based on the time-varying distance; dividing the area into sub-intervals; selecting the minimum quota of the nearest sub-type as the time period quota; and generating a dispatching scheme to improve the efficiency and reliability of power system dispatching.

Description

Technical Field

[0001] This invention relates to the field of power system operation technology, and in particular to a power system dispatching method, apparatus, equipment and medium based on line transmission quota tiered switching. Background Technology

[0002] Currently, with the continuous construction of new power systems, the penetration rate of new energy installed capacity is constantly increasing. Due to the randomness and volatility of wind and solar power output, the system power distribution may change significantly within minute-level or even shorter timescales, resulting in stronger time-varying characteristics in the power grid's security and stability boundaries. Against this backdrop, security and stability constraints are gradually becoming one of the key bottlenecks restricting the optimization of power grid operation.

[0003] However, safety and stability constraints are often characterized by high dimensionality, non-convexity, and nonlinearity. In engineering practice, line transmission limits are often used as a low-complexity approximation of the stability boundary and applied to scheduling operations.

[0004] Existing research on the acquisition and evaluation of transmission limits mainly focuses on two approaches: one is mechanistic model-driven limit calculation methods, such as sensitivity analysis, continuous power flow, repetitive power flow, and optimal power flow, which can maintain accuracy to a certain extent, but usually require a large number of power flow / transient simulation iterations, limiting computational efficiency; the other is data-driven rapid evaluation methods, which use machine learning, deep learning, etc., to establish a nonlinear mapping from operating states to limits, achieving online prediction capabilities. Furthermore, to address the diversity and uncertainty of system operating modes, some studies have also introduced clustering and other methods to classify operating states, thereby improving the generalization and applicability of evaluation models.

[0005] On the other hand, continuous-time economic dispatch (CTED), as an important tool for characterizing continuous power changes over short timescales, has attracted attention in recent years.

[0006] As can be seen from the above, how to improve the efficiency and reliability of power system dispatching in the process of power system dispatching based on line transmission quota tiered switching is an urgent problem to be solved. Summary of the Invention

[0007] In view of this, the purpose of this invention is to provide a power system dispatching method, apparatus, equipment, and medium based on line transmission quota-based tiered switching, which can improve the efficiency and reliability of power system dispatching during the process of power system dispatching based on line transmission quota-based tiered switching. The specific solution is as follows: In a first aspect, this application provides a power system dispatching method based on line transmission quota-based tiered switching, including: Based on the Bernstein interpolation coefficients, the continuous-time variables in the continuous-time economic dispatch model of the power system are transformed to obtain a finite-dimensional linear programming model. Based on historical power system data, operation mode samples are generated, and the linear programming model is used to perform power flow calculation and safety and stability verification on each operation mode sample to obtain an operation mode sample dataset including line power and stability labels. The K-means++ clustering algorithm is used to classify the operation mode sample dataset based on historical new energy output data and historical load data to obtain sample classification results. Then, the feature center corresponding to each sample classification result is determined. Based on the sample partitioning results and the stability-instability boundary determined by the support vector machine classifier, the sample partitioning results are judged and eliminated by using the preset safe and stable point neighborhood search method and based on the stability-instability boundary. The power region corresponding to the eliminated result is divided into grids, and then the maximum power value in each grid is set as the corresponding line transmission limit. The distance between the continuously changing new energy curve and load curve in each time period and each of the aforementioned feature centers is calculated to obtain the time-varying distance, and the category switching time point corresponding to each time period is determined based on the time-varying distance. Based on the switching time points of each category, the continuous time period is divided into several sub-intervals, and the operating mode sub-type closest to each sub-interval is determined. Then, the minimum value of the line transmission limit corresponding to each operating mode sub-type is set as the line transmission limit corresponding to the continuous time period, so as to generate a target economic scheduling scheme using the linear programming model and based on the line transmission limit.

[0008] Optionally, the transformation of continuous-time variables in the continuous-time economic dispatch model of the power system based on Bernstein interpolation coefficients to obtain a finite-dimensional linear programming model includes: The total generation cost of the power system is determined by the integral between the active power of the unit and the cost function of the unit. Then, an objective function is constructed with the goal of minimizing the total generation cost. The unit operation constraints are determined based on the upper limit of active power, the lower limit of active power, the upper limit of reactive power, the lower limit of reactive power, and the ramp rate of the unit. Linearized AC power flow constraints are constructed based on line power equations, node power balance equations, line conductance, line susceptance, and load sets. Node voltage and phase angle constraints are determined based on the square of voltage amplitude, upper limit of phase angle difference, and lower limit of phase angle difference. The line transmission limit constraint is determined based on the current line power flow and the preset line power threshold, and the target constraint is constructed based on the unit operation constraint, the linearized AC power flow constraint, the node voltage and phase angle constraint and the line transmission limit constraint. By using Bernstein interpolation coefficients and based on the objective constraints, the continuous-time variables in the infinite-dimensional continuous-time economic scheduling model are transformed to obtain a finite-dimensional linear programming model.

[0009] Optionally, the step of generating operation mode samples based on historical power system data, and using the linear programming model to perform power flow calculations and safety and stability checks on each operation mode sample, yields an operation mode sample dataset including line power and stability labels, comprising: The system acquires historical renewable energy output data and historical load data corresponding to the power system, and randomly perturbs the historical renewable energy output data within a first preset amplitude range to obtain the renewable energy output curve after perturbation. Then, it randomly perturbs the historical load data within a second preset amplitude range to obtain the load curve after perturbation. Based on the Monte Carlo method and the post-disturbance renewable energy output curve and the post-disturbance load curve, operation mode samples are generated, and the linear programming model is used to process each operation mode sample to obtain the corresponding continuous-time economic dispatch solution; the continuous-time economic dispatch solution includes the active power output and reactive power output of the unit. Based on the continuous-time economic dispatch solution, the net injected active power and net injected reactive power of each node in the power system are determined, and the continuous-time economic dispatch solution is sampled according to a preset time resolution to obtain several dispatch time sections. Power flow calculations are performed on the net injected active power and net injected reactive power corresponding to each scheduling time segment to obtain the active power and reactive power of each line at the corresponding scheduling time segment. Stability checks are performed on each of the aforementioned scheduling time segments to obtain a safety and stability label indicating whether the circuit system corresponding to the scheduling time segment is safe and stable. Then, based on the disturbance-induced renewable energy output curve, the disturbance-induced load curve, the net injected active power, the net injected reactive power, the continuous-time economic scheduling solution, and the safety and stability label corresponding to each of the aforementioned scheduling time segments, an operation mode sample dataset is constructed.

[0010] Optionally, the step of using the K-means++ clustering algorithm and classifying the operation mode sample dataset based on historical new energy output data and historical load data to obtain sample classification results, and then determining the feature centers corresponding to each sample classification result, includes: Feature vectors characterizing the operating mode state are extracted from the operating mode sample dataset; the feature vectors include active power output from new energy sources, reactive power output from new energy sources, active power output from load, and reactive power output from load. Minimizing the sum of squared intra-cluster errors of each cluster is set as the clustering objective. Then, the clustering objective is used, and the K-means++ clustering algorithm is used to perform cluster analysis on the feature vectors to obtain the target number of clusters. The operating mode sample dataset is then divided into operating mode types based on the target number of clusters to obtain the sample division results. Then, the feature centers corresponding to each sample division result are determined. The sum of squared intra-cluster errors is the sum of the squared Euclidean distances from each feature vector in each cluster to the feature center. The feature centers are used to characterize the common features of the operating mode types.

[0011] Optionally, the step of determining the stability-instability boundary based on the sample partitioning results and a support vector machine classifier, using a preset safe and stable point neighborhood search method and based on the stability-instability boundary to judge and remove the sample partitioning results, and dividing the power region corresponding to the removed results into grids, and then setting the maximum power value in each grid as the corresponding line transmission limit, includes: Extract the line power and label corresponding to each transmission line in the sample segmentation results, and construct a two-dimensional feature vector and classification label based on the line power and the label to characterize the mapping relationship between the line operating point and stability; The initial support vector machine classifier is trained using the sample partitioning results to obtain the target support vector machine classifier. The target support vector machine classifier and the radial basis function are then used to map the two-dimensional feature vector to a high-dimensional feature space that satisfies the preset high-dimensional conditions to obtain the mapping result. Based on the mapping results, a stability-instability boundary is determined to distinguish between the stable and unstable regions. A preset safe and stable point neighborhood search method is used to perform a safe and stable point neighborhood search on the line operation points corresponding to each of the operation modes in the sample division results based on the stability-instability boundary. This results in a power region to be eliminated, including the first stable sample point and the unstable sample point. The power region to be eliminated is then eliminated from the power region. Finally, the remaining power region, including each second stable sample point, is divided into grids to obtain each grid. In each of the grids, a stable sample point with the maximum active power amplitude and the maximum reactive power amplitude is determined, and the maximum power amplitude among the stable sample points corresponding to each of the grids is determined. Then, the maximum power amplitude is set as the line transmission limit of the transmission line in the sample division result.

[0012] Optionally, the step of calculating the distance between the continuously changing new energy curves and load curves within each time period and each of the characteristic centers to obtain time-varying distances, and determining the category switching time points corresponding to each time period based on the time-varying distances, includes: Based on the active power output curve of new energy sources, reactive power output curve of new energy sources, active power output curve of load, and reactive power output curve of load corresponding to each scheduling period, a time-varying feature vector is determined, and the Euclidean distance between the time-varying feature vector and each feature center is determined, so as to obtain a time-varying distance that changes continuously with time and is used to characterize the difference between the current operating point and each operating mode. Determine the time switching equation corresponding to each operating mode category, and use the time switching equation and the time-varying distance between the time-varying feature vector and each feature center to determine the category switching time point within the scheduling period.

[0013] Optionally, the step of dividing the continuous time period into several sub-intervals based on the switching time points of each category, determining the operating mode sub-type closest to each sub-interval, and then setting the minimum value among the line transmission limits corresponding to each operating mode sub-type as the line transmission limit corresponding to the continuous time period, to generate a target economic scheduling scheme using the linear programming model and based on the line transmission limit, includes: Based on the switching time points of each category, the continuous scheduling period is divided into several continuous sub-intervals, and the Euclidean distance between the running feature vector of each sub-interval and the corresponding feature center is determined. The running mode category corresponding to the feature center with the smallest distance is set as the running mode sub-type corresponding to the sub-interval. Obtain the line transmission limit corresponding to the sub-type of the operation mode, and set the minimum value among the line transmission limits corresponding to each sub-interval within the same continuous scheduling period as the line transmission limit for the continuous scheduling period. Based on the line transmission limit and the linear programming model, a unit output plan that meets the preset stability and economic conditions is determined, and a target economic dispatch scheme is generated based on the unit output plan.

[0014] Secondly, this application provides a power system dispatching device based on line transmission quota tiered switching, comprising: The time variable transformation module is used to transform the continuous time variables in the continuous time economic dispatch model of the power system based on the Bernstein interpolation coefficients to obtain a finite-dimensional linear programming model. The sample dataset generation module is used to generate operation mode samples based on historical power system data, and to use the linear programming model to perform power flow calculation and safety and stability verification on each operation mode sample to obtain an operation mode sample dataset including line power and stability labels. The feature center determination module is used to classify the operation mode sample dataset by operation mode type using the K-means++ clustering algorithm and based on historical new energy output data and historical load data, to obtain sample classification results, and then determine the feature center corresponding to each sample classification result. The stability-instability boundary generation module is used to determine the stability-instability boundary based on the sample partitioning results and the support vector machine classifier. It uses a preset safe and stable point neighborhood search method and the stability-instability boundary to judge and remove the sample partitioning results. It then divides the power region corresponding to the removed results into grids and sets the maximum power value in each grid as the corresponding line transmission limit. The time-varying distance determination module is used to calculate the distance between the continuously changing new energy curve and load curve in each time period and each of the feature centers to obtain the time-varying distance, so as to determine the category switching time point corresponding to each time period based on the time-varying distance; The economic scheduling scheme generation module is used to divide a continuous time period into several sub-intervals based on the switching time points of each category, determine the operating mode sub-type that is closest to each sub-interval, and then set the minimum value of the line transmission limit corresponding to each operating mode sub-type as the line transmission limit corresponding to the continuous time period, so as to generate a target economic scheduling scheme using the linear programming model and based on the line transmission limit.

[0015] Thirdly, this application provides an electronic device, comprising: Memory, used to store computer programs; A processor is used to execute the computer program to implement the aforementioned power system dispatching method based on line transmission quota tiered switching.

[0016] Fourthly, this application provides a computer-readable medium for storing a computer program, wherein the computer program, when executed by a processor, implements the aforementioned power system dispatching method based on line transmission quota tiered switching.

[0017] As can be seen from the above, before performing power system dispatching based on line transmission quota tiering, this application needs to transform the continuous-time variables in the continuous-time economic dispatch model of the power system using Bernstein interpolation coefficients to obtain a finite-dimensional linear programming model; generate operation mode samples based on historical power system data, and use the linear programming model to perform power flow calculation and safety and stability verification on each operation mode sample to obtain an operation mode sample dataset including line power and stability labels; use the K-means++ clustering algorithm and historical renewable energy output data and historical load data to classify the operation mode sample dataset into operation mode types to obtain sample classification results, and then determine the feature centers corresponding to each sample classification result; based on each sample classification result and a support vector machine classifier, determine the stability-disruption... The system employs a stable boundary approach, utilizing a pre-defined neighborhood search method based on the stable-instability boundary to determine and eliminate samples from each partition. The power regions corresponding to the eliminated samples are then divided into grids, with the maximum power value in each grid set as the corresponding line transmission limit. The system calculates the time-varying distances between the continuously changing renewable energy curves and load curves within each time period and each feature center, using these time-varying distances to determine the category switching time points for each time period. Based on these category switching time points, the continuous time period is divided into several sub-intervals, and the operating mode sub-type closest to each sub-interval is determined. The minimum value among the line transmission limits corresponding to each operating mode sub-type is then set as the line transmission limit for the continuous time period. A target economic dispatch scheme is then generated using a linear programming model based on the line transmission limit.

[0018] Therefore, this application first needs to transform the continuous-time variables in the continuous-time economic dispatch model of the power system based on Bernstein interpolation coefficients to obtain a finite-dimensional linear programming model; generate operation mode samples based on historical power system data, and use the linear programming model to perform power flow calculation and safety and stability verification on each operation mode sample to obtain an operation mode sample dataset including line power and stability labels; secondly, use the K-means++ clustering algorithm and historical renewable energy output data and historical load data to classify the operation mode sample dataset into operation mode types to obtain sample classification results, and then determine the feature centers corresponding to each sample classification result; then, based on each sample classification result and a support vector machine classifier, determine the stability-instability boundary to utilize the preset safety and stability... A fixed-point neighborhood search method is used to determine and eliminate samples based on the stability-instability boundary. The power regions corresponding to the eliminated samples are then divided into grids, and the maximum power value in each grid is set as the corresponding line transmission limit. Furthermore, the distances between the continuously changing renewable energy curves and load curves within each time period and each feature center are calculated to obtain time-varying distances, which are used to determine the category switching time points for each time period. Finally, based on the category switching time points, the continuous time period is divided into several sub-intervals, and the operating mode sub-type closest to each sub-interval is determined. The minimum value of the line transmission limit corresponding to each operating mode sub-type is then set as the line transmission limit for the continuous time period. A linear programming model is then used to generate a target economic dispatch scheme based on the line transmission limit. This improves the efficiency of power system dispatching in the process of power system dispatching based on line transmission limit-based tiered switching, thereby enhancing the safety and reliability of the production process. Attached Figure Description

[0019] 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 embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0020] Figure 1 This is a flowchart of a power system dispatching method based on line transmission quota tiered switching disclosed in this application; Figure 2 This is a schematic diagram of a specific power system dispatching process based on line transmission quota tiered switching disclosed in this application; Figure 3 This is a schematic diagram illustrating a specific method for obtaining the stability-instability boundary of a transmission line as disclosed in this application; Figure 4This is a schematic diagram of a specific stable-instability confusion region disclosed in this application; Figure 5 This is a schematic diagram illustrating the overall process of performing a detailed calculation of line transmission limits as disclosed in this application; Figure 6 This is a schematic diagram of a specific set of continuous time sub-interval partitioning-category mapping rules disclosed in this application; Figure 7 This is a schematic diagram illustrating the calculation process of a specific continuous-time sub-interval partitioning-category mapping rule set disclosed in this application; Figure 8 This is a schematic diagram of a power system dispatching device based on line transmission quota tiered switching disclosed in this application; Figure 9 This is a structural diagram of an electronic device disclosed in this application. Detailed Implementation

[0021] 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.

[0022] Currently, with the continuous construction of new power systems, the penetration rate of new energy installed capacity is constantly increasing. Affected by the randomness and volatility of wind and solar power output, the system power distribution may change significantly within minute-level or even shorter time scales, resulting in stronger time-varying characteristics of the power grid's security and stability boundary. Against this backdrop, security and stability constraints have gradually become one of the key bottlenecks restricting power grid operation optimization. However, security and stability constraints often have high-dimensional, non-convex, and nonlinear characteristics. In engineering practice, line transmission limits are often used as a low-complexity approximation of the stability boundary and applied to dispatching operations. Therefore, this application provides a power system dispatching method based on line transmission limit-based tiered switching, which can improve the efficiency and reliability of power system dispatching in the process of power system dispatching based on line transmission limit-based tiered switching.

[0023] See Figure 1 As shown, this embodiment of the invention discloses a power system dispatching method based on line transmission quota tiered switching, including: Step S11: Based on the Bernstein interpolation coefficients, transform the continuous-time variables in the continuous-time economic dispatch model of the power system to obtain a finite-dimensional linear programming model.

[0024] In this embodiment, this application proposes a continuous-time economic scheduling model that considers tiered switching of line transmission quotas. First, this application constructs a continuous-time economic scheduling model and transforms the infinite-dimensional problem into a finite-dimensional linear programming problem using Bernstein polynomials. Second, it generates a massive number of operating modes and uses K-means++ clustering algorithm to classify several system operating mode categories. Further, for the line transmission power of each category, this application pre-divides the stability-instability boundary using Support Vector Machine (SVM) and uses a safe and stable point neighborhood search method to eliminate the stability-instability confusion region, obtaining a refined line transmission quota for each operating mode category. Based on this, it further derives the relationship between continuous-time load and the system operating mode category classification, dividing the continuous-time load into several sub-intervals belonging to different operating mode categories, obtaining the line transmission quota for each sub-interval, and substituting it into the continuous-time model for solution.

[0025] In this embodiment, the flowchart of power system dispatching based on line transmission quota tiered switching is as follows: Figure 2 As shown: First, a continuous-time economic scheduling model is constructed, and the infinite-dimensional problem is transformed into a finite-dimensional linear programming problem using Bernstein polynomials. This is specifically divided into the following two sub-steps: Continuous-time economic scheduling modeling: The objective function of the model corresponds to the goal of minimizing the total cost. ; in, For scheduling periods; For the set of scheduling periods; For the assembly of generator units; For the unit At any moment Those who have made meritorious contributions; , , The units The coefficients of the quadratic, linear, and constant terms of the active cost function.

[0026] The model constraints include unit constraints, linearized AC power flow constraints, load shedding constraints, and line transmission limit constraints. (1) Unit constraints: The unit must meet the upper and lower limits of output constraints: ; ; in, For the unit At any moment Unproductive efforts; , The units The upper and lower limits of effective contribution; , The units The upper and lower limits of reactive power output.

[0027] In addition, the unit must meet the ramping constraints: ; in, For the unit At any moment The rate of ascent; , The units The maximum rate of climbing uphill and downhill.

[0028] (2) Linearized AC power flow constraints: ; ; ; ; in, For the line At any moment The active power; For the line At any moment reactive power; For nodes At any moment The voltage; For the node To the node The phase angle difference; For the line The electrical conductivity; For the line susceptance; Belonging to a node A collection of generators; Belonging to a node The load set; Belonging to a node A collection of new energy generating units; For nodes There are pairs of nodes connected by a line. gather; , For each new energy unit at time The effort exerted, whether meritorious or not; , They are time points Active and reactive loads; This represents the total number of nodes.

[0029] (3) Voltage amplitude and phase angle constraints: ; ; in, , They are nodes The lower and upper limits of the square of the voltage; , They are slave nodes To the node The lower and upper limits of the phase angle difference.

[0030] (4) Power flow constraints of the line: ; in, For at any time node , The power flow passing through the line at any given moment; For at any time node , The maximum power allowed to flow through the line.

[0031] Furthermore, the embodiments of this application require model transformation based on Bernstein polynomials. That is, the embodiments of this application use Bernstein polynomials to transform the optimization problem in algebraic space into an easily solvable optimization problem in function space. The variables mentioned above are simply modified by adding superscripts. Bk To characterize the first continuous variable corresponding to k The Bernstein interpolation coefficients have the same physical meaning.

[0032] Furthermore, the expression corresponding to the transformed objective function is shown below: ; The expressions corresponding to the transformed constraints are shown below: (1) The expressions corresponding to the converted upper and lower limits of unit output constraints in the unit constraints are as follows: ; ; The expression corresponding to the converted unit ramp-up constraint is shown below: ; (2) The expression for the transformed line power flow constraint in the linearized AC power flow constraint is shown below: ; ; The expression corresponding to the transformed node power balance constraint is shown below: ; ; (3) The expressions corresponding to the voltage amplitude and phase angle constraints are as follows: ; ; (4) The expression corresponding to the line power flow constraint is as follows: ; (5) The expression corresponding to the variable continuity constraint is shown below: ; in, These are the Bernstein interpolation coefficient variables; This represents the set of Bernstein interpolation coefficient variables. The above constraints are first-order continuity constraints, ensuring that the magnitude and slope of the power curves at the end of the previous time period and the beginning of the next time period are equal. In summary, the embodiments of this application use Bernstein polynomials to transform continuous-time variables, thereby transforming an infinite-dimensional problem into a finite-dimensional linear programming problem.

[0033] Specifically, based on Bernstein interpolation coefficients, the continuous-time variables in the continuous-time economic dispatch model of the power system are transformed to obtain a finite-dimensional linear programming model. This model can include: determining the total generation cost of the power system based on the integral between the unit active power and the unit cost function; then constructing an objective function with the goal of minimizing the total generation cost; and determining the unit operating constraints based on the unit active power upper limit, unit active power lower limit, unit reactive power upper limit, unit reactive power lower limit, and unit ramp rate; and determining the line power equation, nodal power balance equation, line conductance, line susceptance, and load settling. Linearized AC power flow constraints are constructed, and node voltage and phase angle constraints are determined based on the square of voltage amplitude, upper and lower limits of phase angle difference. Line transmission limit constraints are determined based on the current line power flow and preset line power threshold. Target constraints are constructed based on unit operation constraints, linearized AC power flow constraints, node voltage and phase angle constraints, and line transmission limit constraints. Using Bernstein interpolation coefficients and based on the target constraints, the continuous-time variables in the infinite-dimensional continuous-time economic scheduling model are transformed to obtain a finite-dimensional linear programming model.

[0034] Step S12: Generate operating mode samples based on historical power system data, and use the linear programming model to perform power flow calculation and safety and stability verification on each operating mode sample to obtain an operating mode sample dataset including line power and stability labels.

[0035] In this embodiment, the present application requires generating a massive number of operating modes and using K-means++ clustering to classify several system operating mode categories, specifically divided into the following two sub-steps: First, the embodiments of this application require the generation of operating modes: In order to alleviate the shortcomings of traditional cross-sectional quota assessment, which is conservative and difficult to characterize the time-varying nature of operating boundaries, the embodiments of this application construct a cross-sectional quota method. This method automatically extracts the similarity and difference between different modes from a large-scale sample of operating modes, and then identifies typical operating modes dominated by load level and new energy output factors; then, using cluster centers, it realizes rapid mode discrimination and attribution mapping of actual operating modes, providing a basis for the dynamic selection of cross-sectional quotas.

[0036] Furthermore, to support the clustering model partitioning, this application embodiment generates a massive dataset of operating modes, the specific process of which is as follows: (1) Random disturbances within ±30% range are made based on historical new energy output data, and random disturbances within ±20% range are made based on historical load data. Then, the Monte Carlo method is used to generate diverse new energy and load curve samples. The total number of samples generated is denoted as N.

[0037] (2) For the generated N sets of new energy and load sample data, the above formula is used to solve the problem and obtain the active power output and reactive power output of the unit in the N sets of continuous time economic dispatch solutions, thereby calculating the net active power and reactive power injected into each node of the system.

[0038] (3) For each continuous-time scheduling solution, the scheduling time segments are sampled with a resolution of 5 minutes. Therefore, for a 24-hour continuous-time scheduling solution, a total of 288 scheduling time segments are sampled. For N sets of continuous-time economic scheduling solutions, a total of 288 scheduling time segments are sampled. There are N economic dispatching sections. For each dispatching time section, power flow calculation is performed to obtain the active power and reactive power of each line.

[0039] (4) For each continuous time scheduling section, a stability check is performed to determine whether the scheduling section meets the system's safety and stability requirements.

[0040] In summary, during the operation mode generation phase, this application's embodiments yielded a total of 288... There are N operating modes, each of which includes: active power output of new energy sources, reactive power output of new energy sources, active power output of load, reactive power output of load, active power output of generator units, reactive power output of generator units, active power of each line, reactive power of each line, and whether the system is operating safely and stably.

[0041] Specifically, operating mode samples are generated based on historical power system data, and power flow calculations and safety and stability checks are performed on each operating mode sample using a linear programming model. This yields an operating mode sample dataset including line power and stability labels. This dataset may include: acquiring historical renewable energy output data and historical load data corresponding to the power system; randomly perturbing the historical renewable energy output data within a first preset amplitude range to obtain the perturbed renewable energy output curve; and then randomly perturbing the historical load data within a second preset amplitude range to obtain the perturbed load curve. Operating mode samples are generated based on the Monte Carlo method and the perturbed renewable energy output curve and load curve, and each operating mode sample is processed using a linear programming model to obtain the corresponding continuous-time economic dispatch solution. The continuous-time economic dispatch solution includes... The system calculates the active power output and reactive power output of generating units; it determines the net injected active power and net injected reactive power of each node in the power system based on the continuous-time economic dispatch solution, and samples the continuous-time economic dispatch solution according to a preset time resolution to obtain several dispatch time sections; it performs power flow calculation on the net injected active power and net injected reactive power corresponding to each dispatch time section to obtain the active power and reactive power of each line at the corresponding dispatch time section; it performs stability verification on each dispatch time section to obtain a safety and stability label to indicate whether the circuit system corresponding to the dispatch time section is safe and stable; and then constructs an operation mode sample dataset based on the post-disturbance renewable energy output curve, post-disturbance load curve, net injected active power, net injected reactive power, continuous-time economic dispatch solution, and safety and stability label corresponding to each dispatch time section.

[0042] Step S13: Use the K-means++ clustering algorithm and historical new energy output data and historical load data to classify the operation mode sample dataset into operation mode types, obtain the sample classification results, and then determine the feature center corresponding to each sample classification result.

[0043] In this embodiment, based on the generated dataset, the present application embodiment needs to further employ the K-means++ algorithm to cluster and divide the operating modes. The input features of the K-means++ algorithm include: active power output from new energy sources. Reactive power output of new energy sources Load active power output Reactive power output of load ,Right now: ; in, These are the input feature vectors for the K-means++ algorithm.

[0044] It is worth mentioning that the goal of K-means+ is to solve the cluster partitioning problem. With cluster center To minimize the sum of squared errors within the cluster: ; in, This represents the sum of squared errors within the cluster. The number of clusters is determined using the silhouette coefficient in this embodiment of the application; This indicates taking the 2-norm.

[0045] Furthermore, after obtaining the cluster partitioning With cluster center Subsequently, in this embodiment of the application, the shortest Euclidean distance between each sample and the selected cluster center is calculated so that all running mode samples are assigned to the class corresponding to the cluster center with the closest Euclidean distance, thereby achieving running mode reduction.

[0046] Specifically, the K-means++ clustering algorithm is used to classify the operation mode sample dataset based on historical renewable energy output data and historical load data, resulting in sample classification results. Then, the feature centers corresponding to each sample classification result are determined. This can include: extracting feature vectors from the operation mode sample dataset to characterize the operation mode status; the feature vectors include renewable energy active power output, renewable energy reactive power output, load active power output, and load reactive power output; setting the minimization of the sum of squared intra-cluster errors of each cluster as the clustering objective; then using the clustering objective and the K-means++ clustering algorithm to perform cluster analysis on the feature vectors to obtain the target number of clusters; classifying the operation mode sample dataset based on the target number of clusters to obtain sample classification results; and then determining the feature centers corresponding to each sample classification result; the sum of squared intra-cluster errors is the sum of the squared Euclidean distances from each feature vector in each cluster to the feature center; the feature centers are used to characterize the common features of the operation mode types.

[0047] Step S14: Based on the sample partitioning results and the support vector machine classifier, determine the stable-instability boundary, use the preset safe and stable point neighborhood search method and the stable-instability boundary to judge and remove the sample partitioning results, divide the power region corresponding to the removed results into grids, and then set the maximum power value in each grid as the corresponding line transmission limit.

[0048] In this embodiment, for each type of line transmission power, the present application embodiment pre-divides the stable-instability boundary using support vector machine (SVM) and uses the safe stable point neighborhood search method to eliminate the stable-instability confusion region, thereby obtaining the refined line transmission limit under each operating mode category.

[0049] Furthermore, to achieve refined calculation of line transmission limits, this application embodiment needs to establish a rapid mapping between line power flow level and system stability: given the active and reactive power flow of a line under a certain operating mode, it can quickly determine whether the point is in the stable or unstable region, and further obtain the stability-instability boundary of the transmission line, such as... Figure 3 As shown.

[0050] To address this, this application proposes a method for pre-dividing the stable-instability boundary based on support vector machines and for eliminating the stable-instability confusion region based on the neighborhood search method of safe and stable points. This method aims to achieve refined calculation of line transmission limits and specifically consists of the following two sub-steps: First, pre-partitioning of the stability-instability boundary based on support vector machine.

[0051] In this embodiment, considering that the stability boundary shows significant differences under different operating modes, after dividing the samples into clusters of typical operating modes, this embodiment trains a support vector machine (SVM) classifier on each cluster and each line, so as to use the trained SVM classifier to pre-divide the stable region and the unstable region, and provide candidate boundaries for subsequent confusion region removal.

[0052] In one specific implementation, let there be a total One line. For each sample (Corresponding to a single scheduling solution and its stability check result), its cluster category is known to be The power flow of each line is calculated, and the stability labels of the samples are obtained through transient simulation or stability criteria. : ; For clusters Take its sample index set For clusters Inner Lines, construct two-dimensional feature vectors : ; in, Indicates the first The circuit of each sample Active power; Indicates the first The circuit of each sample Reactive power.

[0053] Using the two-dimensional feature vector of the line as input and the sample stability 0-1 label as output, the cluster... ,line The problem of partitioning the stability region is transformed into a binary classification problem on a two-dimensional plane.

[0054] Furthermore, for the aforementioned binary classification problem, considering that the stability boundary is typically nonlinear, this application embodiment uses a soft-spaced SVM with an RBF kernel for classification: First, to eliminate the influence of dimensional differences on kernel distance, the input two-dimensional features are standardized: ; in, The input features are standardized. The input features before standardization; The mean of the input features before standardization; The variance of the input features before standardization.

[0055] In the feature space, SVM achieves robust classification by maximizing the margin and allowing a small number of misclassifications. Its optimization form can be written as: ; ; in, The weight vector for the separating hyperplane; For bias terms; For the first A slack variable for each sample, used to allow a small number of samples to violate the interval constraint; This is the penalty coefficient; The first prediction for SVM The category labels of each sample; For nonlinear mapping of the RBF kernel function; For the first The input feature vector after normalization of each sample.

[0056] After training, cluster ,line Discriminant function for: ; in, SV It is a set of support vectors; For the first Lagrange multipliers corresponding to each support vector; For the first The class labels of the support vector samples; For the first Features of each support vector; is the feature vector of the sample to be judged.

[0057] After training, the stability of the samples can be determined based on the discriminant function: if If it is safe and stable; if This indicates instability. Therefore, the stability-instability pre-boundary can be determined by the zero-level set. Given: ; In summary, under a given cluster of operating modes Below, SVM is used for each line. It provides a fast pre-delineation of the stable-instability boundary on the PQ plane, laying the foundation for further identification and elimination of stable-instability confusion regions and extraction of reliable transmission limits.

[0058] Secondly, the stable-instability confusion region is eliminated based on the neighborhood search method of safe and stable points.

[0059] like Figure 4 As shown, due to the limited training samples, non-convex stability boundaries, and complex local morphology, the boundary obtained by SVM often exhibits a certain "classification uncertainty zone," meaning that stable and unstable points are intertwined in local areas. Directly calculating line limits based on pre-defined boundaries may introduce overly optimistic or overly conservative biases. Therefore, this application proposes a safe and stable point neighborhood search method. This method performs gridded neighborhood filtering on the PQ plane of each cluster category's lines, thereby eliminating stable-instability confusion regions and obtaining more reliable limit boundaries.

[0060] Clustering categories ,line Select the SVM discriminant function All sample points, the set of sample points is denoted as : ; Based on the above sample point set, the active and reactive power ranges of the line can be calculated in this embodiment of the application: ; ; in, , These are the maximum and minimum active power of the line, respectively. , These represent the maximum and minimum reactive power of the line, respectively.

[0061] Furthermore, in this embodiment, the aforementioned region is divided equally in both the active and reactive power directions of the line. , There are 1 grid, with a step size of: ; ; in, This represents the step size in the direction of active power of the line; This represents the step size in the direction of reactive power of the line.

[0062] No. grid cells Defined as: ; in, , They are grid cells The lower and upper bounds of the x-coordinate; , They are grid cells The lower and upper bounds of the ordinate.

[0063] For each grid cell This application requires determining whether stable and unstable points exist, and classifies them into the following four categories: (1) All grids that contain both stable and unstable points are denoted as stable-unstable mixed regions; (2) A grid that contains at least one stable sample and no unstable samples is called a stable region; (3) A grid containing at least one unstable sample and no stable samples is denoted as an unstable region: (4) All grids that do not contain stable points and unstable points are denoted as blank areas.

[0064] In stable regions, this embodiment selects the point with the highest power as a safe and stable sample point and uses it as the clustering category. ,line Power transmission limits.

[0065] In summary, the embodiments of this application, based on the safe and stable point neighborhood search method, further eliminate stable-instability confusion regions, and the overall process of fine-tuning the line transmission limit calculation is shown in the following diagram. Figure 5 As shown.

[0066] Specifically, based on the sample partitioning results and the stability-instability boundary determined by the support vector machine classifier, a preset safe and stable point neighborhood search method is used to judge and eliminate samples based on the stability-instability boundary. The power regions corresponding to the eliminated results are then divided into grids, and the maximum power value in each grid is set as the corresponding line transmission limit. This can include: extracting the line power and labels corresponding to each transmission line in the sample partitioning results; constructing a two-dimensional feature vector and classification label based on the line power and labels to characterize the mapping relationship between the line operating point and stability; training an initial support vector machine classifier using the sample partitioning results to obtain a target support vector machine classifier; and using the target support vector machine classifier and the radial basis function kernel function to map the two-dimensional feature vector to a value that satisfies the preset high-dimensional conditions. A high-dimensional feature space is used to obtain mapping results. Based on the mapping results, a stability-instability boundary is determined to distinguish between the stable and unstable regions. A safe and stable point neighborhood search method is used to perform a safe and stable point neighborhood search on the line operating points corresponding to each operating mode sample in the sample partitioning results based on the stability-instability boundary. This results in a power region to be eliminated, including the first stable sample point and the unstable sample point. The power region to be eliminated is then eliminated from the power region. The remaining power region, including each second stable sample point, is then divided into grids to obtain each grid. In each grid, a stable sample point with the maximum active power amplitude and the maximum reactive power amplitude is determined. The maximum power amplitude among the stable sample points corresponding to each grid is then determined. Finally, the maximum power amplitude is set as the line transmission limit of the transmission line in the sample partitioning results.

[0067] Step S15: Calculate the distance between the continuously changing new energy curve and load curve in each time period and each of the aforementioned feature centers to obtain the time-varying distance, and determine the category switching time point corresponding to each time period based on the time-varying distance.

[0068] In this embodiment, the relationship between continuous-time load and system operation mode classification needs to be derived. The continuous-time load is divided into several sub-intervals belonging to different operation mode categories. The line transmission limit for each sub-interval is obtained and substituted into the continuous-time model for solution. In the continuous-time scheduling problem, a fault may occur at any time, causing transient stability issues. Therefore, this embodiment divides the continuous time into several sub-intervals, and determines the category of each sub-interval, thereby constructing a "sub-interval division-category" mapping rule set. A schematic diagram of the continuous-time "sub-interval division-category" mapping rule set is shown below. Figure 6 As shown: For time period According to the Bernstein modeling method, the output and load of new energy sources can be expressed as... cubic polynomial ( ): ; ; in, , They are time points The active and reactive power output of new energy sources; , They are time points The active and reactive power output of the load; , , , They are respectively in the time period The Active power output coefficient of new energy sources, reactive power output coefficient of new energy sources, active power output coefficient of load, reactive power output coefficient of load; For the third Bernstein polynomial The basis functions at time t The value of .

[0069] According to the k-means++ algorithm, for any The calculation of the corresponding shortest distance class is essentially an optimization problem: ; in, , They represent the first The squared Euclidean distance between each new energy unit, its load operating mode, and the cluster center. , They represent the first Active power output and active power load characteristics of new energy generating units operating in cluster-center mode , They represent the first The reactive power output and reactive load characteristics of new energy generating units operating under cluster center mode; the meaning of the above formula is: find the class center closest to the current operating mode, and assign the index of this class center to the limit condition mapping rule set to activate the corresponding limit rule. , All are integer decision variables.

[0070] For simplicity, the characteristics of new energy sources and loads are concatenated into a vector form: ; in, Representing the characteristics of new energy sources and loads in vector form; The first vector form represents the vector form. Cluster centers of each category.

[0071] Defined at time To the Distance between cluster categories : ; For any two cluster categories ,kind With class There is a category switching curve, and the time on the curve Points compared to categories The distances are equal: ; Then proceed to the next step: ; Further analysis reveals: ; Furthermore, for each pair of clustering categories... Based on the above formula, the embodiments of this application construct regarding... cubic equation : ; Order regarding cubic equation Solve the cubic equation in the interval The solutions on [0,1] are each cluster category pair. The switching point.

[0072] Iterate through all cluster pairs to find the interval for the above formula. The solutions on [0, 1] are obtained by sorting all solutions in ascending order. ; in, For all cluster category pairs in the interval The solution is sorted in ascending order [0,1] so that, based on the obtained solution... s Each solution will be in the interval [0,1] is divided into s +1 sub-interval.

[0073] Specifically, the distances between the continuously changing renewable energy curves and load curves within each time period and each feature center are calculated to obtain time-varying distances. Based on these time-varying distances, the category switching time points corresponding to each time period are determined. This can include: determining time-varying feature vectors based on the renewable energy active power output curves, renewable energy reactive power output curves, load active power output curves, and load reactive power output curves corresponding to each scheduling time period, and determining the Euclidean distance between the time-varying feature vectors and each feature center to obtain time-varying distances that change continuously over time and are used to characterize the differences between the current operating point and each operating mode; determining the time switching equations corresponding to each operating mode category, and using the time switching equations and the time-varying distances between the time-varying feature vectors and each feature center to determine the category switching time points within the scheduling time period.

[0074] Step S16: Divide the continuous time period into several sub-intervals based on the switching time points of each category, determine the operating mode sub-type that is closest to each sub-interval, and then set the minimum value of the line transmission limit corresponding to each operating mode sub-type as the line transmission limit corresponding to the continuous time period, so as to generate a target economic scheduling scheme using the linear programming model and based on the line transmission limit.

[0075] In this embodiment, for sub-intervals In this application embodiment, any point is selected, such as the midpoint. Calculate the distance from the point to all classes, and select the class with the shortest distance, denoted as . : ; Understandably, sub-intervals The optimal class is The distance of the shortest class is calculated for each sub-interval. This embodiment of the application can determine the interval. The category to which each sub-interval in [0, 1] belongs is determined, thus obtaining the line transmission limit for each sub-interval. The "interval division-category" mapping rule set can be represented as: ; in, For category The route at that time Transmission limits.

[0076] Therefore, for time periods The line transmission limit is set to the minimum value of the line transmission limit for each sub-interval: ; Then, the obtained By incorporating line power flow constraints, the line tiering and limit embedding in the continuous-time economic scheduling model is achieved, ensuring that the system meets safety and stability constraints even when a fault occurs at any time. In summary, the schematic diagram of the calculation process for the mapping rule set of continuous-time sub-interval partitioning and category is shown below. Figure 7 As shown.

[0077] Specifically, based on the switching time points of each category, the continuous time period is divided into several sub-intervals, and the operating mode sub-type closest to each sub-interval is determined. Then, the minimum value of the line transmission limit corresponding to each operating mode sub-type is set as the line transmission limit corresponding to the continuous time period. A target economic dispatch scheme is generated using a linear programming model and based on the line transmission limit. This can include: dividing the continuous dispatch period into several continuous sub-intervals based on the switching time points of each category, determining the Euclidean distance between the operating feature vector of each sub-interval and the corresponding feature center, and setting the operating mode category corresponding to the feature center with the smallest distance as the operating mode sub-type corresponding to the sub-interval; obtaining the line transmission limit corresponding to the operating mode sub-type, and setting the minimum value of the line transmission limit corresponding to each sub-interval within the same continuous dispatch period as the line transmission limit of the continuous dispatch period; determining the unit output plan that meets the preset stability and economic conditions based on the line transmission limit and the linear programming model, and generating a target economic dispatch scheme based on the unit output plan.

[0078] As can be seen from the above, the embodiments of this application first need to transform the continuous-time variables in the continuous-time economic dispatch model of the power system based on the Bernstein interpolation coefficients to obtain a finite-dimensional linear programming model; generate operation mode samples based on historical power system data, and use the linear programming model to perform power flow calculation and safety and stability verification on each operation mode sample to obtain an operation mode sample dataset including line power and stability labels; secondly, use the K-means++ clustering algorithm and historical new energy output data and historical load data to classify the operation mode sample dataset into operation mode types to obtain sample classification results, and then determine the feature centers corresponding to each sample classification result; then, based on each sample classification result and a support vector machine classifier, determine the stability-instability boundary to utilize the preset safety... A neighborhood search method for fully stable points is employed, and the results of sample partitioning are judged and eliminated based on the stability-instability boundary. The power regions corresponding to the eliminated results are then divided into grids, and the maximum power value in each grid is set as the corresponding line transmission limit. Furthermore, the distances between the continuously changing renewable energy curves and load curves within each time period and each feature center are calculated to obtain time-varying distances, which are used to determine the category switching time points for each time period. Finally, based on the category switching time points, the continuous time period is divided into several sub-intervals, and the operating mode sub-type closest to each sub-interval is determined. The minimum value of the line transmission limit corresponding to each operating mode sub-type is then set as the line transmission limit corresponding to the continuous time period. A linear programming model is then used to generate a target economic dispatch scheme based on the line transmission limit. This improves the efficiency of power system dispatching in the process of power system dispatching based on line transmission limit-based tiered switching, thereby enhancing the safety and reliability of the production process.

[0079] Accordingly, see Figure 8 As shown, this application also provides a power system dispatching device based on line transmission quota tiered switching, comprising: The time variable transformation module 11 is used to transform the continuous time variables in the continuous time economic dispatch model of the power system based on the Bernstein interpolation coefficients to obtain a finite-dimensional linear programming model. The sample dataset generation module 12 is used to generate operation mode samples based on historical power system data, and to use the linear programming model to perform power flow calculation and safety and stability verification on each operation mode sample to obtain an operation mode sample dataset including line power and stability labels. The feature center determination module 13 is used to use the K-means++ clustering algorithm and based on historical new energy output data and historical load data to divide the operation mode sample dataset into operation mode types, obtain sample division results, and then determine the feature center corresponding to each sample division result. The stable-instability boundary generation module 14 is used to determine the stable-instability boundary based on the sample division results and the support vector machine classifier, and to use a preset safe and stable point neighborhood search method to judge and remove the sample division results based on the stable-instability boundary, and to divide the power region corresponding to the removed results into grids, and then set the maximum power value in each grid as the corresponding line transmission limit. The time-varying distance determination module 15 is used to calculate the distance between the continuously changing new energy curve and load curve in each time period and each of the feature centers to obtain the time-varying distance, so as to determine the category switching time point corresponding to each time period based on the time-varying distance; The economic scheduling scheme generation module 16 is used to divide the continuous time period into several sub-intervals based on the switching time points of each category, determine the operating mode sub-type that is closest to each sub-interval, and then set the minimum value of the line transmission limit corresponding to each operating mode sub-type as the line transmission limit corresponding to the continuous time period, so as to generate a target economic scheduling scheme using the linear programming model and based on the line transmission limit.

[0080] In some specific embodiments, the time variable conversion module 11 may specifically include: The total generation cost determination unit is used to determine the total generation cost of the power system based on the integral between the active power of the unit and the cost function of the unit. Then, it constructs an objective function with the goal of minimizing the total generation cost, and determines the unit operation constraints based on the upper limit of active power, the lower limit of active power, the upper limit of reactive power output, the lower limit of reactive power output, and the ramp rate of the unit. The AC power flow constraint construction unit is used to construct linearized AC power flow constraints based on line power equations, node power balance equations, line conductance, line susceptance and load set, and to determine node voltage and phase angle constraints based on the square of voltage amplitude, upper limit of phase angle difference and lower limit of phase angle difference. The condition construction unit is used to determine the line transmission limit constraint condition based on the current line power flow and the preset line power threshold, and to construct the target constraint condition based on the unit operation constraint condition, the linearized AC power flow constraint condition, the node voltage and phase angle constraint condition and the line transmission limit constraint condition; The continuous-time economic scheduling model conversion unit is used to convert the continuous-time variables in the infinite-dimensional continuous-time economic scheduling model using Bernstein interpolation coefficients and based on the target constraints, to obtain a finite-dimensional linear programming model.

[0081] In some specific embodiments, the sample dataset generation module 12 may specifically include: The historical data acquisition unit is used to acquire historical renewable energy output data and historical load data corresponding to the power system, and to randomly perturb the historical renewable energy output data within a first preset amplitude range to obtain the perturbed renewable energy output curve, and then to randomly perturb the historical load data within a second preset amplitude range to obtain the perturbed load curve. The operation mode sample generation unit is used to generate operation mode samples based on the Monte Carlo method and the post-disturbance renewable energy output curve and the post-disturbance load curve, and to process each operation mode sample using the linear programming model to obtain the corresponding continuous-time economic scheduling solution; the continuous-time economic scheduling solution includes the active power output and reactive power output of the unit. The scheduling time section generation unit is used to determine the net injected active power and net injected reactive power of each node in the power system based on the continuous time economic scheduling solution, and to sample the continuous time economic scheduling solution according to a preset time resolution to obtain several scheduling time sections. The power determination unit is used to perform power flow calculations on the net injected active power and net injected reactive power corresponding to each of the scheduling time sections, so as to obtain the active power and reactive power of each line at the corresponding scheduling time section. The safety and stability label determination unit is used to perform stability verification on each of the scheduling time segments to obtain a safety and stability label indicating whether the circuit system corresponding to the scheduling time segment is safe and stable. Then, based on the disturbance-after renewable energy output curve, the disturbance-after load curve, the net injected active power, the net injected reactive power, the continuous time economic scheduling solution, and the safety and stability label corresponding to each of the scheduling time segments, an operation mode sample dataset is constructed.

[0082] In some specific embodiments, the feature center determination module 13 may specifically include: The feature vector extraction unit is used to extract feature vectors representing the operating mode state from the operating mode sample dataset; the feature vectors include active power output of new energy sources, reactive power output of new energy sources, active power output of load, and reactive power output of load. The clustering target determination unit is used to set minimizing the sum of squared intra-cluster errors of each cluster as the clustering target, and then use the clustering target and the K-means++ clustering algorithm to perform cluster analysis on the feature vectors to obtain the target number of clusters. Based on the target number of clusters, the running mode sample dataset is divided into running mode types to obtain sample division results. Then, the feature centers corresponding to each sample division result are determined. The sum of squared intra-cluster errors is the sum of the squared Euclidean distances from each feature vector in each cluster to the feature center. The feature centers are used to characterize the common features of the running mode types.

[0083] In some specific embodiments, the stability-instability boundary generation module 14 may specifically include: The line power determination unit is used to extract the line power and label corresponding to each transmission line in the sample division result, so as to construct a two-dimensional feature vector and classification label based on the line power and the label to characterize the mapping relationship between the line operating point and stability. The mapping result generation unit is used to train an initial support vector machine classifier using the sample partitioning result, obtain a target support vector machine classifier, and use the target support vector machine classifier and radial basis kernel function to map the two-dimensional feature vector to a high-dimensional feature space that satisfies a preset high-dimensional condition to obtain the mapping result; A stable-instability boundary generation sub-unit is used to determine the stable-instability boundary for distinguishing between the stable and unstable regions based on the mapping result. It also uses a preset safe and stable point neighborhood search method and performs a safe and stable point neighborhood search on the line operation points corresponding to each of the operation modes in the sample division result based on the stable-instability boundary to obtain a power region to be eliminated, including the first stable sample point and the unstable sample point. The power region to be eliminated is then eliminated from the power region. Finally, the remaining power region including each second stable sample point is divided into grids to obtain each grid. A stable sample point generation unit is used to determine stable sample points with the maximum active power amplitude and the maximum reactive power amplitude in each of the grids, and to determine the maximum power amplitude among the stable sample points corresponding to each of the grids, and then set the maximum power amplitude as the line transmission limit of the transmission line in the sample division result.

[0084] In some specific embodiments, the time-varying distance determination module 15 may specifically include: The time-varying feature vector determination unit is used to determine the time-varying feature vector based on the active power output curve of new energy, the reactive power output curve of new energy, the active power output curve of load, and the reactive power output curve of load corresponding to each scheduling period, and to determine the Euclidean distance between the time-varying feature vector and each feature center, so as to obtain the time-varying distance that changes continuously with time and is used to characterize the difference between the current operating point and each operating mode. The category switching time point determination unit is used to determine the time switching equation corresponding to each operating mode category, so as to use the time switching equation and the time-varying distance between the time-varying feature vector and each feature center to determine the category switching time point in the scheduling period.

[0085] In some specific embodiments, the economic scheduling scheme generation module 16 may specifically include: The continuous scheduling period division unit is used to divide the continuous scheduling period into several continuous sub-intervals based on the switching time points of each category, and to determine the Euclidean distance between the running feature vector of each sub-interval and the corresponding feature center, and to set the running mode category corresponding to the feature center with the smallest distance as the running mode sub-type corresponding to the sub-interval. The line transmission limit determination unit is used to obtain the line transmission limit corresponding to the sub-type of the operation mode, and set the minimum value among the line transmission limits corresponding to each sub-interval within the same continuous scheduling period as the line transmission limit of the continuous scheduling period. The unit output plan determination unit is used to determine the unit output plan that meets the preset stability and economic conditions based on the line transmission limit and the linear programming model, so as to generate a target economic dispatch scheme based on the unit output plan.

[0086] Furthermore, embodiments of this application also disclose an electronic device, Figure 9 This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content of the diagram should not be construed as limiting the scope of this application. Specifically, the electronic device 20 may include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the power system dispatching method based on line transmission quota tiered switching disclosed in any of the foregoing embodiments. Furthermore, the electronic device 20 in this embodiment may specifically be a computer.

[0087] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0088] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored thereon can include operating system 221, computer program 222, etc., and the storage method can be temporary storage or permanent storage.

[0089] The operating system 221 is used to manage and control the various hardware devices on the electronic device 20 and the computer program 222, which may be Windows Server, Netware, Unix, Linux, etc. In addition to including a computer program capable of performing the power system dispatching method based on line transmission quota tiered switching as disclosed in any of the foregoing embodiments, the computer program 222 may further include computer programs capable of performing other specific tasks.

[0090] Furthermore, this application also discloses a computer-readable medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned power system dispatching method based on line transmission quota tiered switching. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.

[0091] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0092] Those skilled in the art will further 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, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. 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.

[0093] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of media known in the art.

[0094] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0095] The technical solutions provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A power system dispatching method based on line transmission quota tiered switching, characterized in that, include: Based on the Bernstein interpolation coefficients, the continuous-time variables in the continuous-time economic dispatch model of the power system are transformed to obtain a finite-dimensional linear programming model. Based on historical power system data, operation mode samples are generated, and the linear programming model is used to perform power flow calculation and safety and stability verification on each operation mode sample to obtain an operation mode sample dataset including line power and stability labels. The K-means++ clustering algorithm is used to classify the operation mode sample dataset based on historical new energy output data and historical load data to obtain sample classification results. Then, the feature center corresponding to each sample classification result is determined. Based on the sample partitioning results and the stability-instability boundary determined by the support vector machine classifier, the sample partitioning results are judged and eliminated by using the preset safe and stable point neighborhood search method and based on the stability-instability boundary. The power region corresponding to the eliminated result is divided into grids, and then the maximum power value in each grid is set as the corresponding line transmission limit. The distance between the continuously changing new energy curve and load curve in each time period and each of the aforementioned feature centers is calculated to obtain the time-varying distance, and the category switching time point corresponding to each time period is determined based on the time-varying distance. Based on the switching time points of each category, the continuous time period is divided into several sub-intervals, and the operating mode sub-type closest to each sub-interval is determined. Then, the minimum value of the line transmission limit corresponding to each operating mode sub-type is set as the line transmission limit corresponding to the continuous time period, so as to generate a target economic scheduling scheme using the linear programming model and based on the line transmission limit.

2. The power system dispatching method based on line transmission quota tiered switching according to claim 1, characterized in that, The transformation of continuous-time variables in the continuous-time economic dispatch model of the power system based on Bernstein interpolation coefficients to obtain a finite-dimensional linear programming model includes: The total generation cost of the power system is determined by the integral between the active power of the unit and the cost function of the unit. Then, an objective function is constructed with the goal of minimizing the total generation cost. The unit operation constraints are determined based on the upper limit of active power, the lower limit of active power, the upper limit of reactive power, the lower limit of reactive power, and the ramp rate of the unit. Linearized AC power flow constraints are constructed based on line power equations, node power balance equations, line conductance, line susceptance, and load sets. Node voltage and phase angle constraints are determined based on the square of voltage amplitude, upper limit of phase angle difference, and lower limit of phase angle difference. The line transmission limit constraint is determined based on the current line power flow and the preset line power threshold, and the target constraint is constructed based on the unit operation constraint, the linearized AC power flow constraint, the node voltage and phase angle constraint and the line transmission limit constraint. By using Bernstein interpolation coefficients and based on the objective constraints, the continuous-time variables in the infinite-dimensional continuous-time economic scheduling model are transformed to obtain a finite-dimensional linear programming model.

3. The power system dispatching method based on line transmission quota tiered switching according to claim 1, characterized in that, The process involves generating operation mode samples based on historical power system data, and then using the linear programming model to perform power flow calculations and safety and stability checks on each operation mode sample to obtain an operation mode sample dataset including line power and stability labels, comprising: The system acquires historical renewable energy output data and historical load data corresponding to the power system, and randomly perturbs the historical renewable energy output data within a first preset amplitude range to obtain the renewable energy output curve after perturbation. Then, it randomly perturbs the historical load data within a second preset amplitude range to obtain the load curve after perturbation. Based on the Monte Carlo method and the post-disturbance renewable energy output curve and the post-disturbance load curve, operation mode samples are generated, and the linear programming model is used to process each operation mode sample to obtain the corresponding continuous-time economic dispatch solution; the continuous-time economic dispatch solution includes the active power output and reactive power output of the unit. Based on the continuous-time economic dispatch solution, the net injected active power and net injected reactive power of each node in the power system are determined, and the continuous-time economic dispatch solution is sampled according to a preset time resolution to obtain several dispatch time sections. Power flow calculations are performed on the net injected active power and net injected reactive power corresponding to each scheduling time segment to obtain the active power and reactive power of each line at the corresponding scheduling time segment. Stability checks are performed on each of the aforementioned scheduling time segments to obtain a safety and stability label indicating whether the circuit system corresponding to the scheduling time segment is safe and stable. Then, based on the disturbance-induced renewable energy output curve, the disturbance-induced load curve, the net injected active power, the net injected reactive power, the continuous-time economic scheduling solution, and the safety and stability label corresponding to each of the aforementioned scheduling time segments, an operation mode sample dataset is constructed.

4. The power system dispatching method based on line transmission quota tiered switching according to claim 1, characterized in that, The K-means++ clustering algorithm is used to classify the operation mode sample dataset based on historical new energy output data and historical load data to obtain sample classification results. Then, the feature centers corresponding to each sample classification result are determined, including: Feature vectors characterizing the operating mode state are extracted from the operating mode sample dataset; the feature vectors include active power output from new energy sources, reactive power output from new energy sources, active power output from load, and reactive power output from load. Minimizing the sum of squared intra-cluster errors of each cluster is set as the clustering objective. Then, the clustering objective is used, and the K-means++ clustering algorithm is used to perform cluster analysis on the feature vectors to obtain the target number of clusters. The operating mode sample dataset is then divided into operating mode types based on the target number of clusters to obtain the sample division results. Then, the feature centers corresponding to each sample division result are determined. The sum of squared intra-cluster errors is the sum of the squared Euclidean distances from each feature vector in each cluster to the feature center. The feature centers are used to characterize the common features of the operating mode types.

5. The power system dispatching method based on line transmission quota tiered switching according to claim 1, characterized in that, The process involves determining the stability-instability boundary based on the sample partitioning results and a support vector machine classifier, then using a preset safe and stable point neighborhood search method and the stability-instability boundary to judge and eliminate the sample partitioning results. The power regions corresponding to the eliminated results are then divided into grids, and the maximum power value in each grid is set as the corresponding line transmission limit. This includes: Extract the line power and label corresponding to each transmission line in the sample segmentation results, and construct a two-dimensional feature vector and classification label based on the line power and the label to characterize the mapping relationship between the line operating point and stability; The initial support vector machine classifier is trained using the sample partitioning results to obtain the target support vector machine classifier. The target support vector machine classifier and the radial basis function are then used to map the two-dimensional feature vector to a high-dimensional feature space that satisfies the preset high-dimensional conditions to obtain the mapping result. Based on the mapping results, a stability-instability boundary is determined to distinguish between the stable and unstable regions. A preset safe and stable point neighborhood search method is used to perform a safe and stable point neighborhood search on the line operation points corresponding to each of the operation modes in the sample division results based on the stability-instability boundary. This results in a power region to be eliminated, including the first stable sample point and the unstable sample point. The power region to be eliminated is then eliminated from the power region. Finally, the remaining power region, including each second stable sample point, is divided into grids to obtain each grid. In each of the grids, a stable sample point with the maximum active power amplitude and the maximum reactive power amplitude is determined, and the maximum power amplitude among the stable sample points corresponding to each of the grids is determined. Then, the maximum power amplitude is set as the line transmission limit of the transmission line in the sample division result.

6. The power system dispatching method based on line transmission quota tiered switching according to claim 1, characterized in that, The step of calculating the distance between the continuously changing new energy curves and load curves within each time period and each of the aforementioned feature centers to obtain time-varying distances, and determining the category switching time points corresponding to each time period based on the time-varying distances, includes: Based on the active power output curve of new energy sources, reactive power output curve of new energy sources, active power output curve of load, and reactive power output curve of load corresponding to each scheduling period, a time-varying feature vector is determined, and the Euclidean distance between the time-varying feature vector and each feature center is determined, so as to obtain a time-varying distance that changes continuously with time and is used to characterize the difference between the current operating point and each operating mode. Determine the time switching equation corresponding to each operating mode category, and use the time switching equation and the time-varying distance between the time-varying feature vector and each feature center to determine the category switching time point within the scheduling period.

7. The power system dispatching method based on line transmission quota tiered switching according to any one of claims 1 to 6, characterized in that, The process of dividing a continuous time period into several sub-intervals based on the switching time points of each category, determining the operating mode sub-type closest to each sub-interval, and then setting the minimum value among the line transmission limits corresponding to each operating mode sub-type as the line transmission limit corresponding to the continuous time period, thereby generating a target economic scheduling scheme using the linear programming model and based on the line transmission limit, includes: Based on the switching time points of each category, the continuous scheduling period is divided into several continuous sub-intervals, and the Euclidean distance between the running feature vector of each sub-interval and the corresponding feature center is determined. The running mode category corresponding to the feature center with the smallest distance is set as the running mode sub-type corresponding to the sub-interval. Obtain the line transmission limit corresponding to the sub-type of the operation mode, and set the minimum value among the line transmission limits corresponding to each sub-interval within the same continuous scheduling period as the line transmission limit for the continuous scheduling period. Based on the line transmission limit and the linear programming model, a unit output plan that meets the preset stability and economic conditions is determined, and a target economic dispatch scheme is generated based on the unit output plan.

8. A power system dispatching device based on line transmission quota tiered switching, characterized in that, include: The time variable transformation module is used to transform the continuous time variables in the continuous time economic dispatch model of the power system based on the Bernstein interpolation coefficients to obtain a finite-dimensional linear programming model. The sample dataset generation module is used to generate operation mode samples based on historical power system data, and to use the linear programming model to perform power flow calculation and safety and stability verification on each operation mode sample to obtain an operation mode sample dataset including line power and stability labels. The feature center determination module is used to classify the operation mode sample dataset by operation mode type using the K-means++ clustering algorithm and based on historical new energy output data and historical load data, to obtain sample classification results, and then determine the feature center corresponding to each sample classification result. The stability-instability boundary generation module is used to determine the stability-instability boundary based on the sample partitioning results and the support vector machine classifier. It uses a preset safe and stable point neighborhood search method and the stability-instability boundary to judge and remove the sample partitioning results. It then divides the power region corresponding to the removed results into grids and sets the maximum power value in each grid as the corresponding line transmission limit. The time-varying distance determination module is used to calculate the distance between the continuously changing new energy curve and load curve in each time period and each of the feature centers to obtain the time-varying distance, so as to determine the category switching time point corresponding to each time period based on the time-varying distance; The economic scheduling scheme generation module is used to divide a continuous time period into several sub-intervals based on the switching time points of each category, determine the operating mode sub-type that is closest to each sub-interval, and then set the minimum value of the line transmission limit corresponding to each operating mode sub-type as the line transmission limit corresponding to the continuous time period, so as to generate a target economic scheduling scheme using the linear programming model and based on the line transmission limit.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor is configured to execute the computer program to implement the power system dispatching method based on line transmission quota tiered switching as described in any one of claims 1 to 7.

10. A computer-readable medium, characterized in that, Used to store a computer program, wherein the computer program, when executed by a processor, implements the power system dispatching method based on line transmission quota tiered switching as described in any one of claims 1 to 7.

Images