A data-driven power distribution network edge side cluster controllable capacity quantitative evaluation method

Abstract

A kind of data-driven power distribution network edge side cluster controllable capacity quantitative evaluation method: according to selected power distribution system, input power distribution system historical operation data, and construct power distribution network edge side cluster controllable capacity evaluation model training set;Based on Koopman operator, construct power distribution network edge side cluster controllable capacity evaluation model, and train power distribution network edge side cluster controllable capacity evaluation model using training set, that is, least square method is used to obtain evaluation matrix, to obtain the power distribution network edge side cluster controllable capacity evaluation model of completion training;According to the trained power distribution network edge side cluster controllable capacity evaluation model obtained in step, the boundary interaction power and voltage of power distribution network cluster are quantitatively analyzed, and the analysis result is output.The present application fully considers the variable access state and operation mode of distributed energy, accurately fits the complex mapping relationship of node injection power and cluster intertie power, and breaks away from the dependence of traditional mechanism model driven method on detailed model parameters.

Description

Technical Field

[0001] This invention relates to a method for quantitatively assessing the controllability of a distribution network. In particular, it relates to a data-driven method for quantitatively assessing the controllability of distribution network edge-side clusters. Background Technology

[0002] With the large-scale integration of diverse and complex distributed resources into the distribution network, the pressure on grid operation continues to increase. The continuous increase in installed capacity and power generation of renewable energy, coupled with limited development of flexible resources on the power generation side, exacerbates the contradiction between the large-scale development of distributed power sources and the insufficient carrying capacity of the distribution network. New types of loads, such as electric vehicles and data centers, are characterized by small individual capacity but large numbers, and the power system lacks efficient interaction mechanisms and regulation methods with users, resulting in a large scale of flexible resources on the user side with low utilization rates. The widespread application of flexible distribution equipment such as intelligent soft switches increases the complexity of the distribution system's operating environment and dispatch control. Changes in the characteristics of the power generation, grid, and load systems place higher demands on the organizational structure of the distribution network and local operation control technologies.

[0003] As a new control architecture and organizational model for power distribution systems, edge-side cluster control of power distribution networks aligns closely with the concept of virtual power plants. Edge computing devices are a new type of integrated power distribution terminal equipment that can measure and aggregate power distribution network operation data such as voltage, current, and power. They also possess certain computing capabilities, enabling local data aggregation, fusion utilization, and control functions. Edge-side cluster control of power distribution networks continues the hierarchical and zonal operation control approach, comprehensively utilizing differentiated resources such as power sources, loads, and storage. By quantitatively evaluating the cluster's external control capabilities through equivalent models, it uses external cluster characteristics to replace individual resource features in grid-level optimization and control, promoting local consumption of distributed energy and improving the operational flexibility of the power system through bidirectional interaction with the grid.

[0004] Traditional power distribution system operation and control methods, based on precise physical models of the power distribution system, face challenges such as difficulty in obtaining or inaccurate acquisition of distribution line parameters. With the rapid development of advanced information technology and the digital transformation of power distribution networks, systems can acquire massive amounts of heterogeneous power distribution data that describe the fine-grained state of the power grid. Data-driven methods can fully leverage the potential value of measurement data, dynamically characterizing the complex nonlinear relationships between input and output variables using historical operating data and real-time measurement information. This effectively avoids the limitations of traditional physical models, providing strong support for efficient operation and control of distribution network edge clusters. Summary of the Invention

[0005] The technical problem to be solved by this invention is to provide a data-driven method for quantitatively evaluating the controllability of distribution network edge-side clusters that can fully consider the variable access status and operation mode of distributed energy and get rid of the dependence on detailed model parameters of traditional mechanism model-driven methods.

[0006] The technical solution adopted in this invention is: a data-driven quantitative evaluation method for the controllability of distribution network edge-side clusters, comprising the following steps:

[0007] 1) Based on the selected power distribution system, input the historical operating data of the power distribution system, including: node injected active power, node injected reactive power, cluster boundary node voltage, and inter-cluster tie line power;

[0008] 2) Based on the historical operation data of the distribution system in step 1), construct a training set for the controllability assessment model of the distribution network edge side cluster, including: setting the node injected active power and node injected reactive power as input variables, and setting the cluster boundary node voltage and inter-cluster tie line power as output variables to form a training set;

[0009] 3) Construct a controllability assessment model for the distribution network edge-side cluster based on the Koopman operator, including: randomly initializing the basis vector matrix, performing dimensionality increase processing on the input variables, and training the controllability assessment model for the distribution network edge-side cluster using the training set, that is, using the least squares method to obtain the assessment matrix, and obtaining the completed training model for the controllability assessment model for the distribution network edge-side cluster.

[0010] 4) Based on the completed training of the distribution network edge-side cluster controllability assessment model obtained in step 3), perform quantitative analysis on the boundary interaction power and voltage of the distribution network cluster and output the analysis results.

[0011] This invention presents a data-driven quantitative assessment method for the controllability of distribution network edge-side clusters. Addressing the challenge of defining cluster boundaries due to increasingly complex coupling relationships among multiple factors, this method establishes a quantitative assessment model for the controllability of distribution network edge-side clusters. A training sample set is constructed using node-injected power as input and cluster operational characteristics as output. This data-driven approach accurately quantifies cluster control capabilities, characterizes the operational boundaries of edge-side clusters, and supports more refined global system-wide optimization and control requirements. This invention addresses the frequent changes in the composition and behavioral characteristics of distribution network edge-side clusters, fully considering the variable access states and operating modes of distributed energy resources. It constructs a training sample set using historical operating data and linearizes nonlinear equations through dimensionality-enhancing processing of input variables. This accurately fits the complex mapping relationship between node-injected power and inter-cluster tie-line power, eliminating the dependence on detailed model parameters in traditional mechanistic model-driven methods. Using this method, the voltage and interactive power of cluster boundary nodes can be accurately quantified and assessed even when distribution network parameters are missing, ensuring that the relative estimation error is less than 1‰. Meanwhile, the method of this invention has a faster training performance for the analysis and computation of massive data, and can characterize the operational characteristics of each distribution network edge cluster in a year in just seconds, making it more suitable for situations where edge computing power is limited. Attached Figure Description

[0012] Figure 1 This is a flowchart of a data-driven quantitative evaluation method for the controllability of distribution network edge-side clusters according to the present invention.

[0013] Figure 2 It is an improved IEEE 33-node distribution network structure diagram;

[0014] Figure 3 It is the local load power curve;

[0015] Figure 4 It is the output curve of the local distributed power source;

[0016] Figure 5 It is the active power distribution of the link between cluster 1 and cluster 2 on the edge side;

[0017] Figure 6 It is the active power distribution of the link between cluster 2 and cluster 3 on the edge side. Detailed Implementation

[0018] The following describes in detail, with reference to embodiments and accompanying drawings, a data-driven quantitative evaluation method for the controllability of distribution network edge-side clusters according to the present invention.

[0019] like Figure 1 As shown, the present invention provides a data-driven quantitative evaluation method for the controllability of distribution network edge-side clusters, comprising the following steps:

[0020] 1) Based on the selected power distribution system, input the historical operating data of the power distribution system, including: node injected active power, node injected reactive power, cluster boundary node voltage, and inter-cluster tie line power;

[0021] 2) Based on the historical operation data of the distribution system in step 1), construct a training set for the controllability assessment model of the distribution network edge-side cluster, including: setting node injected active power and node injected reactive power as input variables, and setting cluster boundary node voltage and inter-cluster tie line power as output variables to form a training set; wherein,

[0022] The input variables for setting the node injected active power and node injected reactive power are:

[0023]

[0024] In the formula, X is the input variable of the controllability assessment model of the distribution network edge-side cluster, x h Here, G represents the number of nodes in the power distribution system, and H represents the number of time segments, which are the input variables corresponding to the h-th time segment. These represent the active power and reactive power injected into the i-th node at the h-th time segment, respectively.

[0025] The settings for cluster boundary node voltage and inter-cluster tie-line power as output variables are as follows:

[0026]

[0027] In the formula, Y is the output variable of the controllability assessment model of the distribution network edge-side cluster, y h Here, K represents the number of boundary nodes, Z represents the number of inter-cluster connections on the edge side, and H represents the number of time segments. Let i be the voltage at the lower boundary node i of the h-th time section. These represent the active power and reactive power flowing through tie line j at the h-th time section, respectively.

[0028] 3) Construct a controllability assessment model for the distribution network edge-side cluster based on the Koopman operator, including: randomly initializing the basis vector matrix, performing dimensionality increase processing on the input variables, and training the controllability assessment model for the distribution network edge-side cluster using the training set, that is, using the least squares method to obtain the assessment matrix, and obtaining the completed training model for the controllability assessment model for the distribution network edge-side cluster.

[0029] The aforementioned assessment model for the controllability of distribution network edge-side clusters is expressed as follows:

[0030]

[0031] In the formula, Y is the output variable of the controllability assessment model for the distribution network edge-side cluster, X is the input variable of the controllability assessment model for the distribution network edge-side cluster, and M is the controllability assessment matrix. lift ψ(X) is the input variable after the dimensionality upgrade of the controllability assessment model of the distribution network edge-side cluster;

[0032] The aforementioned dimensionality-upgrading process for input variables is represented as follows:

[0033]

[0034]

[0035]

[0036] In the formula, ψ d (x) is the d-th variable that increases in dimension; f lift It is a non-linear, dimension-increasing function; x h Let C be the input variable corresponding to the h-th time segment; C is the basis vector matrix; c d Let c be the basis vector corresponding to the d-th variable in increased dimensions. d The value of ψ varies with d The value changes with the input variable (x), and its order of magnitude is the same as that of the input variable;r,h c is the r-th input variable corresponding to the h-th time segment; r,d The basis corresponding to the r-th dimension of the input variable corresponding to the d-th dimension-upgraded variable; D is the dimension of the input variable upgrade; H is the number of time sections; G is the number of nodes in the power distribution system;

[0037] The process of solving the controllability assessment matrix M is expressed as follows:

[0038]

[0039] In the formula, [·] T Represents the transpose of a matrix. Represents the Moore-Penrose inverse of the matrix.

[0040] 4) Based on the completed training of the distribution network edge-side cluster controllability assessment model obtained in step 3), perform quantitative analysis on the boundary interaction power and voltage of the distribution network cluster and output the analysis results.

[0041] The quantitative analysis of the interaction power and voltage at the boundary of the distribution network cluster is expressed as follows:

[0042]

[0043] In the formula, x t Inject power into the node at time t, y t Let x be the voltage and interaction power of the cluster boundary node at time t, M be the controllability assessment matrix, and x be the voltage and interaction power of the cluster boundary node at time t. lift As the input variable after the dimensionality upgrade of the controllability assessment model for the distribution network edge-side clusters, ψ(x) t ) is the variable that increases in dimension at time t.

[0044] The following are specific examples:

[0045] To verify the effectiveness of the data-driven quantitative evaluation method for the controllability of distribution network edge-side clusters proposed in this invention, an improved IEEE 33-node example was selected for testing. The system was divided into three distribution network edge-side clusters, such as... Figure 2 As shown. The system voltage level is 12.66kV, and it comprises a total of 32 branches, operating in a radial pattern. The total active power of the load is 3.715MW, and the total reactive power is 2.3Mvar.

[0046] A sample set was generated using historical operating data and tested. With a sampling period of 5 minutes, 365 days of photovoltaic output curves and load fluctuation curves were obtained, as shown below. Figure 3 , Figure 4As shown, a dataset containing 105,120 time segments was generated. Using the annual node-injected active power and node-injected reactive power as input variables, a quantitative analysis of the voltage at the boundary nodes of the distribution network cluster and the power of the inter-cluster tie lines was performed.

[0047] The computer hardware environment for performing the optimized calculations was an Intel(R) Core(TM) i7-9700 CPU with a clock speed of 3.00GHz and 24GB of memory; the software environment was a Windows 10 operating system.

[0048] Table 1 Training accuracy of the controllability assessment model for distribution network edge-side clusters

[0049]

[0050] The performance of the proposed method was evaluated using mean absolute error, maximum absolute error, and maximum relative error. The training results of the distribution network edge-side cluster controllability assessment model are shown in Table 1. The boundary interaction power of the distribution network edge-side cluster under different source-load outputs is shown in Table 1. Figure 5 , Figure 6 As shown, the upper and lower bounds of the boundary exchange power between cluster 1 and cluster 2 are 2.362MW and -1.322MW, respectively, and the upper and lower bounds of the boundary exchange power between cluster 1 and cluster 2 are 0.893MW and -0.865MW, respectively. In summary, the quantitative evaluation method for the controllability of distribution network edge-side clusters proposed in this invention can accurately describe the voltage and interaction power of cluster boundary nodes over long time scales, supporting the global optimization and control requirements of the system.

Claims

1. A data-driven method for quantitatively evaluating the controllability of distribution network edge-side clusters, characterized in that, Includes the following steps: 1) Based on the selected power distribution system, input the historical operating data of the power distribution system, including: node injected active power, node injected reactive power, cluster boundary node voltage, and inter-cluster tie line power; 2) Based on the historical operation data of the distribution system in step 1), construct a training set for the controllability assessment model of the distribution network edge side cluster, including: setting the node injected active power and node injected reactive power as input variables, and setting the cluster boundary node voltage and inter-cluster tie line power as output variables to form a training set; 3) Constructing a controllable cluster assessment model for the edge side of the distribution network based on the Koopman operator, including: randomly initializing the basis vector matrix, performing dimensionality increase processing on the input variables, and training the controllable cluster assessment model for the edge side of the distribution network using the training set, i.e., using the least squares method to obtain the assessment matrix, thus obtaining the completed controllable cluster assessment model for the edge side of the distribution network; the controllable cluster assessment model for the edge side of the distribution network is expressed as: ； In the formula, For the output variables of the controllability assessment model of the distribution network edge-side cluster, As input variables for the controllability assessment model of the distribution network edge-side cluster, This is a controllability assessment matrix. The input variables for the upgraded dimensionality assessment model of the controllability assessment of distribution network edge-side clusters. As a variable for increasing dimensionality; The aforementioned dimensionality-upgrading process for input variables is represented as follows: ； In the formula, For the first One variable that increases in dimensionality; It is a nonlinear dimension-increasing function; For the first Input variables corresponding to each time segment; The basis vector matrix; For the first The basis vectors corresponding to each variable in the increased dimension The value of varies It changes with the input variable, and its order of magnitude is the same as that of the input variable; For the first The first time segment corresponding to the first One input variable; In order to be with the first The first variable of the upgraded dimension The basis corresponding to the input variables; Increase the dimensionality of input variables; This refers to the number of time sections; This refers to the number of nodes in the power distribution system. Controllability Assessment Matrix The solution process is expressed as follows: ； In the formula, Represents the transpose of a matrix. Represents the Moore-Penrose inverse of the matrix; 4) Based on the completed training of the distribution network edge-side cluster controllability assessment model obtained in step 3), perform quantitative analysis on the cross-sectional power and voltage of the distribution network cluster boundary, and output the analysis results; the quantitative analysis of the cross-sectional power and voltage of the distribution network cluster boundary is expressed as follows: ； In the formula, for Power injection at time nodes for Voltage and interaction power of cluster boundary nodes at all times. This is a controllability assessment matrix. The input variables for the upgraded dimensionality assessment model of the controllability assessment of distribution network edge-side clusters. for The variable that increases in dimensionality at any given moment.

2. The data-driven quantitative evaluation method for the controllability of distribution network edge-side clusters according to claim 1, characterized in that, Step 2) sets the node injected active power and node injected reactive power as input variables as follows: ； In the formula, As input variables for the controllability assessment model of the distribution network edge-side cluster, For the first The input variables corresponding to each time segment This refers to the number of nodes in the power distribution system. The number of time sections, , The first The first time section The active and reactive power injected into each node; Step 2) sets the cluster boundary node voltage and inter-cluster tie-line power as output variables: ； In the formula, For the output variables of the controllability assessment model of the distribution network edge-side cluster, For the first Output variables corresponding to each time segment The number of boundary nodes, This represents the number of inter-cluster communication lines on the edge side. The number of time sections, For the first Lower boundary node of time section voltage, , The first The connecting line at each time section The active and reactive power flowing through.

Images