A power distribution network adaptive dispatchable domain construction method and related products

Abstract

This invention discloses a method and system for constructing an adaptive dispatchable domain for a distribution network, belonging to the field of distribution network operation and control technology. The method includes: constructing a linearized power flow model and a distributed photovoltaic uncertainty set; generating an initial linearized dispatchable domain through external secant projection and calculating its Chebyshev center as the safe physical origin; receiving the optimal interactive power dispatch command from the upper-level grid; connecting the safe physical origin with the coordinate points of the dispatch command to generate a target quantity; determining a local sector-shaped correction region with the safe physical origin as its vertex based on the intersection of the target quantity and the initial domain; calculating the true feasible boundary within this region using a nonlinear model and performing directional correction; updating the local boundary chain to generate the adaptive dispatchable domain. This method can accurately lock the local boundary most relevant to the upper-level dispatch requirements with extremely low computational cost, overcoming the problems of the curse of high-dimensional nonlinear computation and the distortion of linear model boundaries, and efficiently releasing the true safe control capability of the distribution network.

Description

Technical Field

[0001] This invention relates to the field of power distribution network operation control technology, specifically to a method for constructing an adaptive dispatchable domain for a power distribution network and related products. Background Technology

[0002] The high proportion of distributed photovoltaic and other uncertain resources connected to the grid has led to increasingly frequent and uncertain power interactions between the distribution network and the upstream main grid. In order to quantify the range of active and reactive power regulation capabilities that the distribution network as a whole can provide to the upstream main grid while ensuring the safe operation of the distribution network (such as ensuring that node voltage and branch current do not exceed limits), the industry has introduced the concept of the dispatchable domain of the distribution network.

[0003] In existing technologies, the mainstream approach to constructing a dispatchable domain involves combining the nonlinear power flow equations, topological parameters, and safety operating boundaries of various underlying controllable devices within the distribution network. Mathematical projection techniques are then used to map the high-dimensional operating space onto the active-reactive power plane at the point of common connection. For example, while projection methods based on linearized approximation models (such as LinDistFlow) can significantly reduce computational complexity, they neglect strongly non-convex branch losses and the quadratic term of voltage drop, leading to distorted accuracy of the generated boundaries in actual physical operation. This can result in overly conservative or optimistic approaches in key control directions. On the other hand, directly introducing a nonlinear power flow model for global boundary search can lead to the curse of dimensionality due to the complex scale of variables and intertwined non-convex constraints within the distribution network. Furthermore, the lack of guidance from specific scheduling requirements of the upper-level power grid inevitably generates a large amount of redundant boundary calculation overhead unrelated to current actual scheduling commands, making it difficult to meet the high timeliness requirements of coordinated primary and secondary power dispatch.

[0004] Therefore, how to release and quantify the true potential security control capabilities of the distribution network with high precision and at extremely low computational cost, based on the real-time scheduling intentions of the upper-level power grid, without falling into the disaster of high-dimensional nonlinear computation, has become a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for constructing an adaptive dispatchable domain for a distribution network, so as to overcome the technical defects in the prior art, such as the distortion of boundary accuracy caused by relying on linearized approximation models, and the curse of dimensionality and excessive redundant computational overhead caused by global nonlinear correction.

[0006] The present invention solves the above-mentioned technical problems through the following technical solution: This invention provides a method for constructing an adaptive dispatchable domain for a distribution network, comprising the following steps: A linearized power flow model of the distribution network, LinDistFlow, is constructed. An uncertainty set of distributed photovoltaic power output is constructed based on the kernel density estimation method. The linearized operation constraints corresponding to all photovoltaic uncertainty vertex scenarios are stacked to obtain a linearized operation constraint set of the distribution network. Based on the linearized operation constraint set of the distribution network, the high-dimensional operation space of the distribution network is projected onto the active-reactive power two-dimensional plane at the common connection point between the distribution network and the upper-level main network using the external cut method, thereby generating an initial linearized dispatchable domain; the Chebyshev center of the initial linearized dispatchable domain is calculated as the safe physical origin of the distribution network. The system receives the optimal interactive power dispatch command at the point of common coupling from the upper-level power grid, connects the coordinate point corresponding to the safe physical origin with the optimal interactive power dispatch command, and generates a target quantity starting from the safe physical origin. Based on the intersection of the target quantity and the initial linearized dispatchable domain, a local sector correction region is determined with the safe physical origin as the vertex and the target quantity as the central axis. Within the local sector correction region, the nonlinear DistFlow power flow model is used to calculate the real feasible boundary that satisfies the nonlinear power flow constraints, and the boundary segment in the initial linearized dispatchable domain located within the local sector correction region is directionally corrected based on the real feasible boundary. Based on the results of the directional correction, the local boundary chains located within the local sector correction region in the initial linearized schedulable domain are updated to generate an adaptive schedulable domain for the distribution network that takes into account the uncertainties of distributed photovoltaics.

[0007] A further improvement of this invention lies in constructing an uncertainty set for distributed photovoltaic (PV) output based on the kernel density estimation method, and stacking the linearized operational constraints corresponding to all PV uncertainty vertex scenarios to obtain a set of linearized operational constraints for the distribution network. Specifically, this includes the following steps: Collect distributed photovoltaic (PV) power output data minute by minute within a preset historical time period; preprocess the historical power output data to remove outliers and missing values; calculate the probability density function of PV power output using a Gaussian kernel function, with the kernel bandwidth determined by cross-validation; determine the upper and lower limits of uncertainty range for PV power output based on a preset confidence level; and generate all uncertain vertex scenarios for PV power output using a vertex enumeration method based on the uncertainty range. For each photovoltaic uncertainty vertex scenario, generate corresponding distribution network linearization operation constraints; The linearization constraints of all scenarios are merged, and the intersection of all constraints is taken as the linearization constraint set of the distribution network.

[0008] A further improvement of this invention lies in using the external cut method to project the high-dimensional operating space of the distribution network onto a two-dimensional plane of active and reactive power at the common connection point between the distribution network and the upper-level main network, thereby generating an initial linearized dispatchable domain. Specifically, this includes the following steps: Initialize the external cutting polyhedron as a hyperrectangle containing all feasible points; In each iteration, the linear programming problem is solved to find an extreme point of the current polyhedron; Draw a cutting plane through this pole to remove the infeasible part of the polyhedron; Repeat the above iterative process until the difference in schedulable area between two iterations is less than a preset threshold or the maximum number of iterations is reached, thus obtaining the initial linearized schedulable domain.

[0009] A further improvement of this invention lies in determining a local sector-shaped correction region with the safe physical origin as the vertex and the target quantity as the central axis, based on the intersection of the target quantity and the initial linearized schedulable domain. This specifically includes the following steps: Calculate all intersections between the target quantity and the boundary of the initial linearized schedulable domain; When there are multiple intersection points, the intersection point farthest from the safe physical origin is selected as the boundary intersection point; With the safety physical origin as the vertex and the target quantity as the central axis, the area expands to both sides at a preset angle to form a local fan-shaped correction area; The boundary of the local fan-shaped correction region is two rays originating from the safe physical origin, and the angle between the two rays and the target amount is half of the preset angle.

[0010] A further improvement of this invention lies in using a nonlinear DistFlow power flow model to calculate the true feasible boundary satisfying the nonlinear power flow constraints, and then performing directional correction on the boundary segments located within the local sector correction region in the initial linearized schedulable domain based on the true feasible boundary. Specifically, this includes the following steps: Within the local sector correction area, discrete scan points are generated with a preset radial step size and a preset angular step size; For the power of the common connection point corresponding to each scan point, the nonlinear DistFlow power flow equation is solved by the forward-backward substitution method. Determine whether the power flow results obtained from the solution satisfy all distribution network operation constraints; Connect all the scan points that satisfy the constraints to form the real feasible boundary; Identify all boundary segments in the initial linearized schedulable domain that are entirely located within the local sector correction region; Replace the boundary segments with the calculated real feasible boundaries; When the actual feasible boundary intersects with the boundary of the local sector correction region, the intersection point is used as the endpoint of the corrected boundary.

[0011] A further improvement of this invention is that the method for constructing an adaptive dispatchable domain for a distribution network further includes the following steps: After generating the adaptive dispatchable domain of the distribution network, it is uploaded to the upper-level power grid dispatch center; Receive the final dispatch instructions issued by the superior power grid dispatch center based on the adaptive dispatchable domain; control the operation of distributed photovoltaic and adjustable loads in the distribution network according to the final dispatch instructions.

[0012] The present invention also provides a system for constructing an adaptive dispatchable domain for a distribution network, comprising: The first module is used to construct the LinDistFlow linearized power flow model of the distribution network. It constructs the uncertainty set of distributed photovoltaic power output based on the kernel density estimation method, and stacks the linearized operation constraints corresponding to all photovoltaic uncertainty vertex scenarios to obtain the linearized operation constraint set of the distribution network. The second module is used to project the high-dimensional operating space of the distribution network onto the active-reactive power two-dimensional plane at the common connection point between the distribution network and the upper-level main network based on the linearized operation constraint set of the distribution network, and to generate the initial linearized dispatchable domain; the Chebyshev center of the initial linearized dispatchable domain is calculated as the safe physical origin of the distribution network. The third module is used to receive the optimal interactive power dispatch command at the point of common coupling issued by the upper-level power grid, connect the safe physical origin with the coordinate point corresponding to the optimal interactive power dispatch command, and generate a target quantity starting from the safe physical origin; based on the intersection of the target quantity and the initial linearized dispatchable domain, a local sector correction region with the safe physical origin as the vertex and the target quantity as the central axis is determined; within the local sector correction region, the nonlinear DistFlow power flow model is used to calculate the real feasible boundary that satisfies the nonlinear power flow constraints, and the boundary segment in the initial linearized dispatchable domain located within the local sector correction region is oriented and corrected based on the real feasible boundary; The fourth module is used to update the local boundary chains located within the local sector correction region in the initial linearized schedulable domain based on the results of the directional correction, thereby generating an adaptive schedulable domain for the distribution network that takes into account the uncertainties of distributed photovoltaics.

[0013] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described method for constructing an adaptive dispatchable domain for a power distribution network.

[0014] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of the above-described method for constructing an adaptive dispatchable domain for a power distribution network.

[0015] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described method for constructing an adaptive dispatchable domain for a power distribution network.

[0016] Compared with the prior art, the positive and progressive effects of the present invention are as follows: The adaptive dispatchable domain construction method for distribution networks provided by this invention adopts a decoupled progressive architecture that first performs linear dimensionality reduction projection and then performs directional nonlinear correction based on upper-level dispatch instructions. First, an initial linearized dispatchable domain is quickly generated through the LinDistFlow linearization model and the external cut method, avoiding the curse of dimensionality of directly solving high-dimensional nonlinear power flow models. Then, by determining the Chebyshev center and the safe physical origin, and generating target quantities, the local sector correction region most relevant to the actual dispatch requirements of the upper-level power grid is accurately located. Finally, a high-precision nonlinear DistFlow model is used only in this local region for boundary directional correction, which improves the real safety control capability of the distribution network with extremely low computational cost and overcomes the dual defects of low global correction efficiency and linear model boundary distortion in the prior art.

[0017] Furthermore, the kernel density estimation method is used to construct the uncertainty set of distributed photovoltaic power output. This eliminates the need for prior assumptions about the probability distribution of photovoltaic power output and can more accurately fit the distribution characteristics of actual historical data. All uncertainty vertex scenarios are generated by vertex enumeration, and the corresponding constraints are stacked and their intersection is taken, ensuring the robustness and security of the constructed linearized operating constraint set under the preset confidence level.

[0018] Furthermore, the external cut method is employed to iteratively solve for the poles of the linear programming problem and construct the cutting plane, thereby progressively projecting the high-dimensional linear feasible region onto a two-dimensional active-reactive power plane. This process avoids direct elimination of high-dimensional variables, resulting in high computational efficiency. Moreover, by controlling the iteration termination condition through the area difference threshold, a flexible balance can be achieved between computational accuracy and speed.

[0019] Furthermore, by calculating the intersection of the target quantity and the boundary of the initial linearized schedulable domain, and selecting the intersection farthest from the safe physical origin as the boundary intersection, it is ensured that the determined local sector correction area can cover the direction most urgently needed by the upper-level power grid; by expanding to both sides with a preset angle to form a sector area, the limitations of single-direction correction are avoided, and redundant calculations caused by global search are prevented. Attached Figure Description

[0020] The accompanying drawings are provided to further understand the invention and constitute a part of this invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0021] Figure 1 This is a schematic flowchart of the method of the present invention; Figure 2 This is a schematic diagram of the topology of a 5-node test system; Figure 3 A schematic diagram of the initial linearized schedulable region and Chebyshev center; Figure 4 A schematic diagram of the schedulable domain for local vertex ray correction. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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, 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.

[0023] In the description of this invention, it should be understood that the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0024] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0025] It should be understood that although terms such as first, second, third, etc., may be used in the embodiments of the present invention to describe the preset range, these preset ranges should not be limited to these terms. These terms are only used to distinguish the preset ranges from one another. For example, without departing from the scope of the embodiments of the present invention, the first preset range may also be referred to as the second preset range, and similarly, the second preset range may also be referred to as the first preset range.

[0026] Depending on the context, the word "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0027] With the large-scale integration of distributed energy sources such as distributed photovoltaics, energy storage, and flexible loads, the power interaction between the active distribution network and the upstream main grid is becoming increasingly frequent and uncertain. To quantify the range of active and reactive power regulation capabilities that the distribution network as a whole can provide to the upstream main grid while ensuring absolute operational safety within the distribution network (e.g., node voltage and branch current not exceeding limits), the industry has introduced the concept of the distribution network dispatchable domain (or feasible domain, PQ-Disp-Reg). The fundamental principle for constructing this region is to simultaneously consider the nonlinear power flow equations within the distribution network (e.g., the DistFlow model), topology parameters, and the safe operating boundaries of various underlying controllable devices (e.g., photovoltaic inverters, SVG, flexible loads). Through polyhedral mathematical projection or Monte Carlo sampling techniques, the high-dimensional operating space containing hundreds or thousands of internal node states is equivalently mapped onto the two-dimensional interactive power (PQ) plane of the point of common coupling (PCC), thus providing intuitive physical boundary constraints for the efficient coordinated scheduling of the main and distribution networks.

[0028] Existing traditional search algorithms lack engineering practicality. As the number and dimensionality of constraints in high-dimensional models increase, fixed search step sizes and preset directions lead to numerous redundant search paths and inefficiencies, failing to meet the computational power requirements of minute-level or even second-level real-time online scheduling in distribution networks. On the other hand, although the aforementioned projection theory achieves spatial dimensionality reduction, most studies treat the schedulable domain as a static offline evaluation tool. In actual distribution network continuous rolling scheduling, such as the time coupling of energy storage state of charge (SOC) and the time-varying characteristics of photovoltaic output, the schedulable domain will dynamically drift. If existing static projection models need to update the boundary, a global invalid reconstruction must be performed in each control cycle, which is extremely time-consuming and completely severs the physical connection between the security domain characterization and the real-time scheduling intention of the upper-level power grid.

[0029] Because existing technologies mainly rely on linearized approximate power flow models when constructing the dispatchable domain of distribution networks, they neglect the strong non-convex branch losses and voltage drop quadratic terms. This leads to the initial linear boundary being distorted in actual physical operation, and is prone to being conservative or optimistic in key control directions. If non-linear power flow models are directly introduced into the global boundary or all fixed angles to correct these deviations, the complex scale of variables and the intertwined non-convex constraints within the distribution network will inevitably lead to the dimensionality curse of non-convex optimization solutions. Furthermore, due to the lack of guidance from the specific dispatch requirements of the upper-level power grid, a large amount of redundant boundary calculation overhead that is unrelated to the current actual dispatch instructions will inevitably be generated, making it difficult to meet the high timeliness requirements of primary and secondary coordinated dispatch.

[0030] Based on the above background, this application provides a method for constructing an adaptive dispatchable domain for a distribution network. This method generates an initial linearized dispatchable domain by constructing a linearized operational constraint set for the distribution network and determining a safe physical origin. It then combines this with upper-level dispatch instructions to generate targeted quantities and lock a local sector correction region. Directional boundary correction of nonlinear power flow constraints is performed only within this region, ultimately generating an adaptive dispatchable domain that considers the uncertainties of distributed photovoltaic systems. This method avoids the curse of high-dimensional nonlinear computation while enabling high-precision quantification of the distribution network's true potential safety control capabilities with extremely low computational cost, based on the real-time dispatch intentions of the upper level.

[0031] The technical solutions described below in conjunction with the embodiments of this application will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0032] This embodiment provides a method for constructing an adaptive schedulable domain in a distribution network, which specifically includes the following steps: A LinDistFlow linearized power flow model for the distribution network is constructed. An uncertainty set for distributed photovoltaic (PV) output is built based on kernel density estimation. The linearized operational constraints corresponding to all PV uncertainty vertex scenarios are stacked to obtain a linearized operational constraint set for the distribution network. The LinDistFlow linearized power flow model transforms the nonlinear power flow equations of the distribution network into linear constraints, reducing subsequent computational complexity. The uncertainty set of distributed PV output is used to characterize the stochastic fluctuation characteristics of PV output. Stacking the linearized operational constraints of all vertex scenarios ensures that the generated constraint set meets the safe operation requirements of the distribution network under all PV fluctuation scenarios.

[0033] Based on the linearized operation constraint set of the distribution network, the external cut method is used to project the high-dimensional operation space of the distribution network onto a two-dimensional plane of active and reactive power at the common connection point between the distribution network and the upper-level main network, generating an initial linearized dispatchable domain. The Chebyshev center of the initial linearized dispatchable domain is calculated as the safe physical origin of the distribution network. The external cut method can efficiently realize the dimensionality reduction projection of the high-dimensional operation space onto a two-dimensional plane, avoiding the computational burden caused by direct high-dimensional projection. The Chebyshev center is the point farthest from all boundaries in the initial linearized dispatchable domain, representing the safest position of the distribution network operation, and providing a reference origin for subsequent targeted correction.

[0034] The system receives the optimal interactive power dispatch command at the point of common coupling from the upper-level power grid, connects the coordinates of the safe physical origin with the coordinates corresponding to the optimal interactive power dispatch command, and generates a target quantity starting from the safe physical origin. Based on the intersection of the target quantity and the initial linearized dispatchable domain, a local sector correction region is determined with the safe physical origin as the vertex and the target quantity as the central axis. Within the local sector correction region, the nonlinear DistFlow power flow model is used to calculate the true feasible boundary that satisfies the nonlinear power flow constraints, and the boundary segments in the initial linearized dispatchable domain located within the local sector correction region are directionally corrected based on the true feasible boundary. The target quantity can accurately characterize the direction of the urgent dispatch demand of the upper-level power grid; the local sector correction region limits the nonlinear calculation range to the local area most relevant to the current dispatch demand, greatly reducing redundant calculations; the nonlinear DistFlow power flow model can accurately reflect the actual operating characteristics of the distribution network and correct the boundary accuracy errors caused by the linearization model.

[0035] Based on the results of the directional correction, the local boundary chains within the local sector correction region in the initial linearized dispatchable domain are updated to generate an adaptive dispatchable domain for the distribution network that considers the uncertainties of distributed photovoltaic power. Updating only the boundary chains in local regions avoids the huge overhead of global nonlinear calculations while ensuring boundary accuracy in key scheduling directions, achieving a balance between computational efficiency and boundary accuracy.

[0036] The adaptive dispatchable domain construction method for distribution networks provided by this invention extracts the target quantity characterizing the urgent needs of the main grid by solving the Chebyshev center of the initial linear domain as the safe physical origin and connecting it with the optimal scheduling command of the upper level. This adaptively locks the relevant local active boundary chain. On this basis, nonlinear power flow constraints are reintroduced only in the local sector area determined by the adaptive method to implement high-precision targeted correction. This perfectly overcomes the curse of high-dimensional nonlinear computation with extremely low computational overhead, and efficiently and accurately releases and characterizes the true safe control capability of the distribution network in the direction of the urgent needs of the main grid. Because the algorithm only reintroduces an accurate nonlinear power flow model for directional correction (expansion or contraction) within the local sector high-value region determined by the target quantity, while keeping the linear boundary unchanged for other irrelevant regions, it not only completely avoids the dimensionality curse caused by the high-dimensional non-convex optimization projection of the entire distribution network at the underlying mathematical level, eliminating a large amount of redundant computational overhead of useless boundaries, and greatly improving the dynamic timeliness of main and distribution coordinated control, but also effectively overcomes the problem of conservative or distorted operating boundaries caused by neglecting the quadratic terms of network loss and voltage drop in traditional pure linear models. It realizes the release and dynamic characterization of the real potential safety control capability of the distribution network with high precision at extremely low computational cost, providing a solid physical boundary guarantee for the robust and safe operation of the active distribution network under large-scale photovoltaic fluctuations and the refined power balance scheduling of the main grid.

[0037] This invention is based on the following inventive concept: Based on the mathematical projection concept of the dispatchable domain, the external cut method is introduced to construct the dispatchable domain of the adaptive distribution network. By solving for the center of the largest inscribed circle (Chebyshev Center) of the dispatchable domain at time t, the safe physical origin is used. This center is connected to the optimal dispatch command issued by the upper-level grid to extract the target quantity that represents the urgent needs of the main grid. Nonlinear constraints are reintroduced, and the lower-level grid only expands in a high-precision targeted direction in the direction of the main grid's demand. This characterizes the range of adaptive active and reactive power regulation capabilities that the distribution network can provide to the upper-level grid under photovoltaic fluctuation conditions.

[0038] First, each feeder aggregates its internal operating constraints, linear Distflow power flow equations, and equality constraints. Inequality constraints with network security A two-dimensional mathematical projection is performed on the set consisting of voltage deviation and branch current limits, as well as the set considering the uncertainties of distributed photovoltaics. The schedulable domain at time t is generated by the external cut method. ; Secondly, the upper-level power grid performs network-wide power balance dispatching based on the dispatchable domains of each feeder, achieving efficient coordination between dispatching systems at all levels. For example, the dispatchable domain of an active distribution network aggregated from massive distributed power sources, controllable loads, and energy storage is projected and equivalently characterized and submitted to the regional power grid. The regional power grid uses the dispatchable domains of the active distribution network as constraints to perform regional power dispatching through power transmission via tie lines, and issues power allocation dispatching instructions for each feeder.

[0039] Furthermore, after the feeder layer receives the power scheduling instruction from the upper level at time t+1, it generates a Chebyshev center, establishes a targeted nonlinear correction, and then expands the search range of the region along the position of the power scheduling center, ultimately forming an adaptive schedulable domain for upper-level scheduling.

[0040] See Figure 1 Specifically, it includes the following steps: Linearized power flow model of distribution network LinDistFlow: The high-dimensional schedulable domain constructed in this invention needs to be projected onto a two-dimensional state space. However, the strongly nonconvex quadratic terms contained in a traditional Distflow power flow model are difficult to directly apply to the geometric projection calculation of multicellular structures. Therefore, this invention uses the LinDistFlow power flow model to convexify and linearize the original nonlinear DistFlow equations. This model is based on the following approximation: Ignore active and reactive power losses in branches, i.e., assume that the transmission power of branches is much greater than the line loss; assume that the per-unit voltage amplitude of each node is close to 1∠0°; use the square of the voltage amplitude. It replaces the voltage amplitude itself, eliminating the square root nonlinear term.

[0041] The linearized set of core operational constraints for the distribution network is expressed as follows: 1) Active power balance constraint

[0042] In the formula: Let be the active power flowing out of node j at time t; The active power flowing from node i to node j at time t; Inject active power into node j itself at time t; 2) Reactive power balance constraint

[0043] In the formula: Let be the active power flowing out of node j at time t; The active power flowing from node i to node j at time t; Inject active power into node j itself at time t; 3) Branch voltage drop constraint

[0044] In the formula: These are the resistance and reactance of branch ij, respectively.

[0045] After linearization, the operating constraints of the distribution network can be uniformly written as:

[0046] in: x: Distribution network state variables, such as branch active power, branch reactive power, and node voltage; y: Controllable resource adjustment variables, such as energy storage, flexible loads, reactive power of photovoltaic inverters, etc.; The interactive power that the upper-level power grid is concerned about; w: Uncertain fluctuations in distributed photovoltaic power generation; D: System constant vector.

[0047] Distributed photovoltaic power output kernel density estimation and uncertainty set construction: Accurately characterizing the stochasticity of distributed photovoltaic (PV) output is crucial for assessing the resilience of distribution networks to fluctuations and generating absolute safe operating boundaries. This invention employs kernel density estimation (KDE) from nonparametric estimation to model the marginal distribution of PV prediction errors, thereby avoiding the over-reliance on prior distributions in traditional parametric modeling. Furthermore, it introduces box assemblies to construct the uncertainty of PV output.

[0048] Let the sample set of photovoltaic power output prediction error be... The KDE estimate of its probability density function is:

[0049] In the formula: N is the total number of historical error samples; h is the smoothing bandwidth; This is the Gaussian kernel function.

[0050] To balance robustness and economy in collaborative scheduling, a confidence level is introduced. The photovoltaic power output fluctuation range at a specific confidence level is constructed by intercepting the corresponding two-sided quantiles from the obtained probability density curve. The set of box-shaped uncertainties in photovoltaic power output fluctuations. :

[0051] like As a polyhedron, its vertex scenes can be enumerated:

[0052] Establish a set of internal adjustment variables for each photovoltaic scenario. And by stacking all the scene constraints, we get: .

[0053] Power system convex set projected dispatchable domain: The initial PQ-Disp-Reg considering photovoltaic uncertainties can be defined as follows:

[0054] Where: given the P and Q interaction power If a point can be configured to have a feasible adjustment scheme in all photovoltaic peak scenarios, then the point belongs to PQ-Disp-Reg.

[0055] The aforementioned region is essentially a high-dimensional linear feasible region. Projection in space. Since a linear polyhedron remains a polyhedron after projection, the initial PQ-Disp-Reg can be written in the following explicit half-space form:

[0056] To avoid the computational burden of directly using high-dimensional projection, this invention employs a boundary iteration generation method based on the infeasibility of duality. First, an initial outer region containing the real schedulable domain is given:

[0057] in, This can be determined by the physical upper and lower bounds of P and Q. Then, the P and Q points most likely to lead to system infeasibility are searched within the current enclosing region. For any given... System feasibility is determined by The decision is made. If the system of linear inequalities is infeasible, then according to the duality theory of linear programming, there must exist a dual vector u that can represent the point where the problem is infeasible. Based on this property, a maximum infeasibility discrimination model can be constructed within the current enclosing region to find the most dangerous points P and Q.

[0058] In the k-th iteration, the current outer region is By solving a mixed-integer linear programming problem, the maximum infeasibility degree within the current region can be obtained. and the corresponding critical point .like If the condition is met, it means that there are no significantly infeasible points within the current outer region, and the algorithm terminates.

[0059] like This indicates that the current outsourced region still contains unschedulable points, requiring the generation of new boundary constraints. To obtain a valid boundary, a known feasible interior point is selected. and connect and the most dangerous point Along the line Perform a binary search to locate the critical point near the feasible region boundary. During this process, determine infeasibility using dual infeasibility criteria. The feasibility of the boundary is determined, and the dual vector ub corresponding to the infeasible side of the boundary is recorded.

[0060] Based on this dual vector, a new boundary hyperplane can be constructed:

[0061] in, This hyperplane can remove the infeasible portion of the current bounding region while preserving the real schedulable domain. After adding it to the current boundary set, we have...

[0062] Repeat the process of searching for the most dangerous point, locating the boundary, and generating the boundary hyperplane until the maximum infeasibility meets the convergence accuracy requirement. Finally, the explicit expression of the initial linearized PQ-Disp-Reg can be obtained:

[0063] The core idea of ​​this step is to gradually identify the effective boundaries of the projected region by utilizing dual infeasibility, thus avoiding direct elimination of high-dimensional variables. Through this process, the implicit feasibility model, which originally contained a large number of internal power flow state variables and regulation variables, is transformed into a two-dimensional explicit polyhedral region that only relates to PQ, providing a foundation for subsequent Chebyshev center calculations, matching of higher-level scheduling commands, and adaptive nonlinear corrections.

[0064] Establish an adaptive schedulable domain: (1) Generating target quantities based on Chebyshev centers and scheduling instructions d Get the initial Then, its Chebyshev center was further calculated:

[0065] in: The Chebyshev center represents the center point furthest from all boundaries within the dispatchable domain, and r represents the radius of the largest inscribed circle. The Chebyshev center can be understood as the safe center point for the current distribution network's PQ regulation capability.

[0066] (2) Fan-shaped targeting nonlinear correction To avoid overly localized boundary characterization due to correction along only a single direction, and to avoid introducing boundary points unrelated to the upper-level scheduling requirements through fixed-angle sector search, this invention adopts an adaptive sector-targeted nonlinear correction method based on the initial schedulable domain boundary hit.

[0067] Let the initial linearized PQ-Disp-Reg be:

[0068] in, Chebyshev Center Pointing to the superior scheduling instruction The direction is used as the main target direction:

[0069] Construct rays along this direction:

[0070] Based on the initial linearized schedulable region, calculate the distance between the intersection point of the ray and the boundary:

[0071] The corresponding linear boundary points are:

[0072] And record the active boundary code that was hit. This boundary represents the initial linear boundary that most directly constrains the P and Q regulation capabilities of the distribution network in the direction of the current superior dispatch demand.

[0073] Subsequently, according to The initial schedulable domain boundary topology is used to determine the two adjacent endpoints of this boundary. .like If a region is located exactly at a vertex, then its two adjacent boundary vertices are taken. This constructs a local sector region related to the upper-level scheduling requirements:

[0074] This sector is not a global sector with a pre-fixed angle, but is adaptively determined by the initial boundary hit by the scheduling target direction and its adjacent vertices. Therefore, it can ensure that the correction area is concentrated near the local boundary most relevant to the superior instruction.

[0075] Within a local sector, select the following set of directions:

[0076] in:

[0077] For each direction We reintroduce the nonlinear power flow model and consider all photovoltaic uncertainty vertices to solve for the true nonlinear feasible boundary in this direction:

[0078] satisfy: And for any photovoltaic uncertainty vertex scenario Each of these variables has corresponding branch power flow, node voltage, and energy storage / flexible load / photovoltaic inverter adjustment variables, ensuring that the nonlinear DistFlow power flow constraints, voltage safety constraints, line capacity constraints, and equipment adjustment upper and lower limit constraints are all valid.

[0079] The corresponding nonlinear boundary points are:

[0080] By comparison Distance from linear boundary Determine whether the initial linearization PQ-Disp-Reg is conservative or optimistic in this direction: If This indicates that the linearization boundary is conservative in this direction, allowing for directional expansion; if This indicates that the linearization boundary is too optimistic in this direction, and local contraction should be performed.

[0081] (3) Generate adaptive PQ-Disp-Reg Based on the nonlinear boundary search results in multiple directions within the local sector, the hit boundary and its neighboring boundary chains of the initial linearized PQ-Disp-Reg are locally updated to obtain the adaptive schedulable domain:

[0082] in, and From the initial boundary set and local nonlinear boundary points The process is jointly determined. During the update, only the local boundary chains related to the boundary hit in the main target direction are replaced or corrected, while the far-end boundaries that are not hit remain unchanged, thereby avoiding global miscutting or non-physical expansion of irrelevant areas by nonlinear verification results.

[0083] From this, we obtain Compared to the initial linearized PQ-Disp-Reg, it can more accurately characterize the adaptive active and reactive power regulation capabilities that the distribution network can provide to the upstream grid under photovoltaic uncertainty conditions.

[0084] To verify the effectiveness of the method of the present invention, a 5-node radial distribution network test system was constructed for simulation. (See attached diagram.) Figure 2 The baseline capacity is set to 10 MVA, the system root node (PCC) is set to node 1, the rated voltage is set to 1.00 pu, and the voltage range of the distribution network safe operation node is set to 0.95 pu to 1.05 pu.

[0085] Distributed photovoltaic (PV): Nodes 3 and 5 are connected, with baseline outputs of 0.40 pu and 0.30 pu respectively. The prediction error fluctuation range is set to... The photovoltaic inverter capacities at nodes 3 and 5 are configured to be 0.52 pu and 0.40 pu, respectively.

[0086] Controllable active and reactive resources: Node 4 is configured with a flexible load with an active power adjustment range of [-0.25, 0.35] pu and an SVG device with a reactive power adjustment range of [-0.20, 0.20] pu; Node 2 is configured with flexible resources with an active power adjustment range of [-0.15, 0.25] pu.

[0087] Initial schedulable domain construction and Chebyshev center: See Figure 3First, based on the photovoltaic uncertainty vertex scenario generated by the LinDistFlow linearized power flow model and kernel density estimation, the high-dimensional implicit feasible region containing a large number of internal power flow state variables and adjustment variables is mathematically projected onto the PQ two-dimensional space at PCC. The blue polygonal region represents the initial PQ-Disp-Reg generated by high-dimensional projection. This region characterizes the range of safe active and reactive power regulation capabilities that the underlying power grid can provide to the upper-level power grid, assuming nonlinear branch losses and strong non-convex constraints are ignored.

[0088] Furthermore, by solving for the Chebyshev center inside the initial polyhedron, the safe center point of the current distribution network's P and Q regulation capabilities is calculated. The shortest distance from this center point to the boundary of the linear polyhedron is the maximum inscribed circle radius, which intuitively reflects the system's maximum robust safety margin under all photovoltaic uncertainties. Targeted nonlinear correction and adaptive domain generation: To avoid the computational overhead introduced by fixed-angle global search that is irrelevant to the scheduling requirements of the higher-level grid, this invention uses the scheduling instructions issued by the higher-level grid as a guide to implement local sector-targeted nonlinear correction, such as... Figure 4 As shown.

[0089] The results of this example show that the method of the present invention not only avoids global non-physical expansion to irrelevant areas, but also releases the potential control capability of the distribution network with high precision in the key scheduling directions urgently needed by the main grid, thereby improving the adaptive adjustment capability of the distribution network under photovoltaic fluctuation conditions.

[0090] Based on the same inventive concept, embodiments of this application provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of a method for constructing an adaptive dispatchable domain for a power distribution network. The memory may include main memory, such as high-speed random access memory, or it may also include non-volatile memory, such as at least one disk storage device. The processor, network interface, and memory are interconnected via an internal bus, which may be an industry-standard architecture bus, a peripheral component interconnection standard bus, an extended industry-standard architecture bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. The memory is used to store the program; specifically, the program may include program code, which includes computer operation instructions. The memory may include main memory and non-volatile memory, and provides instructions and data to the processor.

[0091] Based on the same inventive concept, embodiments of this application provide a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the steps of the method for constructing an adaptive dispatchable domain in a power distribution network. Specifically, the computer-readable storage medium includes, but is not limited to, volatile memory and / or non-volatile memory. The volatile memory may include RAM (Random Access Memory) and / or cache memory, etc. The non-volatile memory may include ROM (Read-Only Memory), hard disk, flash memory, optical disk, magnetic disk, etc.

[0092] Based on the same inventive concept, this application provides a computer program product, which includes a computer program stored on a computer-readable storage medium. The computer program includes program instructions, which, when executed by a computer device, cause the computer device to perform the steps of the above-described method for constructing an adaptive dispatchable domain for a power distribution network.

[0093] Those skilled in the art will understand that embodiments of the present invention can be provided as methods or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM (Compact Disc Read-Only Memory), optical storage, etc.) containing computer-usable program code.

[0094] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer apparatus or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0095] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer device or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0096] These computer program instructions may also be loaded onto a computer device or other programmable data processing equipment to cause a series of operational steps to be performed on the computer device or other programmable equipment to produce a process implemented by the computer device, thereby providing instructions that execute on the computer device or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0097] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0098] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for constructing an adaptive dispatchable domain for a distribution network, characterized in that, Includes the following steps: A linearized power flow model of the distribution network, LinDistFlow, is constructed. An uncertainty set of distributed photovoltaic power output is constructed based on the kernel density estimation method. The linearized operation constraints corresponding to all photovoltaic uncertainty vertex scenarios are stacked to obtain a linearized operation constraint set of the distribution network. Based on the linearized operation constraint set of the distribution network, the high-dimensional operation space of the distribution network is projected onto the active-reactive power two-dimensional plane at the common connection point between the distribution network and the upper-level main network using the external cut method, thereby generating an initial linearized dispatchable domain; the Chebyshev center of the initial linearized dispatchable domain is calculated as the safe physical origin of the distribution network. The system receives the optimal interactive power dispatch command at the point of common coupling from the upper-level power grid, connects the coordinate point corresponding to the safe physical origin with the optimal interactive power dispatch command, and generates a target quantity starting from the safe physical origin. Based on the intersection of the target quantity and the initial linearized dispatchable domain, a local sector correction region is determined with the safe physical origin as the vertex and the target quantity as the central axis. Within the local sector correction region, the nonlinear DistFlow power flow model is used to calculate the real feasible boundary that satisfies the nonlinear power flow constraints, and the boundary segment in the initial linearized dispatchable domain located within the local sector correction region is directionally corrected based on the real feasible boundary. Based on the results of the directional correction, the local boundary chains located within the local sector correction region in the initial linearized schedulable domain are updated to generate an adaptive schedulable domain for the distribution network that takes into account the uncertainties of distributed photovoltaics.

2. The method for constructing an adaptive dispatchable domain for a distribution network according to claim 1, characterized in that, An uncertainty set for distributed photovoltaic (PV) output is constructed based on the kernel density estimation method. Then, the linearized operational constraints corresponding to all PV uncertainty vertex scenarios are stacked to obtain a set of linearized operational constraints for the distribution network. The specific steps include: Collect distributed photovoltaic (PV) power output data minute by minute within a preset historical time period; preprocess the historical power output data to remove outliers and missing values; calculate the probability density function of PV power output using a Gaussian kernel function, with the kernel bandwidth determined by cross-validation; determine the upper and lower limits of uncertainty range for PV power output based on a preset confidence level; and generate all uncertain vertex scenarios for PV power output using a vertex enumeration method based on the uncertainty range. For each photovoltaic uncertainty vertex scenario, generate corresponding distribution network linearization operation constraints; The linearization constraints of all scenarios are merged, and the intersection of all constraints is taken as the linearization constraint set of the distribution network.

3. The method for constructing an adaptive dispatchable domain for a distribution network according to claim 1, characterized in that, The high-dimensional operating space of the distribution network is projected onto the active-reactive power two-dimensional plane at the common connection point between the distribution network and the upper-level main network using the external cut method, generating an initial linearized dispatchable domain. The specific steps include: Initialize the external cutting polyhedron as a hyperrectangle containing all feasible points; In each iteration, the linear programming problem is solved to find an extreme point of the current polyhedron; Draw a cutting plane through this pole to remove the infeasible part of the polyhedron; Repeat the above iterative process until the difference in schedulable area between two iterations is less than a preset threshold or the maximum number of iterations is reached, thus obtaining the initial linearized schedulable domain.

4. The method for constructing an adaptive dispatchable domain for a distribution network according to claim 1, characterized in that, Based on the intersection of the target quantity and the initial linearized schedulable domain, a local sector-shaped correction region is determined with the safe physical origin as the vertex and the target quantity as the central axis. This process includes the following steps: Calculate all intersections between the target quantity and the boundary of the initial linearized schedulable domain; When there are multiple intersection points, the intersection point farthest from the safe physical origin is selected as the boundary intersection point; With the safety physical origin as the vertex and the target quantity as the central axis, the area expands to both sides at a preset angle to form a local fan-shaped correction area; The boundary of the local fan-shaped correction region is two rays originating from the safe physical origin, and the angle between the two rays and the target amount is half of the preset angle.

5. The method for constructing an adaptive dispatchable domain for a distribution network according to claim 1, characterized in that, The true feasible boundary satisfying the nonlinear power flow constraints is calculated using a nonlinear DistFlow power flow model. Based on the true feasible boundary, the boundary segment located within the local sector correction region in the initial linearized schedulable domain is oriented and corrected. Specifically, the steps include: Within the local sector correction area, discrete scan points are generated with a preset radial step size and a preset angular step size; For the power of the common connection point corresponding to each scan point, the nonlinear DistFlow power flow equation is solved by the forward-backward substitution method. Determine whether the power flow results obtained from the solution satisfy all distribution network operation constraints; Connect all the scan points that satisfy the constraints to form the real feasible boundary; Identify all boundary segments in the initial linearized schedulable domain that are entirely located within the local sector correction region; Replace the boundary segments with the calculated real feasible boundaries; When the actual feasible boundary intersects with the boundary of the local sector correction region, the intersection point is used as the endpoint of the corrected boundary.

6. The method for constructing an adaptive dispatchable domain for a distribution network according to claim 1, characterized in that, The method for constructing an adaptive dispatchable domain for a distribution network also includes the following steps: After generating the adaptive dispatchable domain of the distribution network, it is uploaded to the upper-level power grid dispatch center; Receive the final dispatch instructions issued by the superior power grid dispatch center based on the adaptive dispatchable domain; control the operation of distributed photovoltaic and adjustable loads in the distribution network according to the final dispatch instructions.

7. A system for constructing an adaptive dispatchable domain for a distribution network, characterized in that, include: The first module is used to construct the LinDistFlow linearized power flow model of the distribution network. It constructs the uncertainty set of distributed photovoltaic power output based on the kernel density estimation method, and stacks the linearized operation constraints corresponding to all photovoltaic uncertainty vertex scenarios to obtain the linearized operation constraint set of the distribution network. The second module is used to project the high-dimensional operating space of the distribution network onto the active-reactive power two-dimensional plane at the common connection point between the distribution network and the upper-level main network based on the linearized operation constraint set of the distribution network, and to generate the initial linearized dispatchable domain; the Chebyshev center of the initial linearized dispatchable domain is calculated as the safe physical origin of the distribution network. The third module is used to receive the optimal interactive power dispatch command at the point of common coupling issued by the upper-level power grid, connect the safe physical origin with the coordinate point corresponding to the optimal interactive power dispatch command, and generate a target quantity starting from the safe physical origin; based on the intersection of the target quantity and the initial linearized dispatchable domain, a local sector correction region with the safe physical origin as the vertex and the target quantity as the central axis is determined; within the local sector correction region, the nonlinear DistFlow power flow model is used to calculate the real feasible boundary that satisfies the nonlinear power flow constraints, and the boundary segment in the initial linearized dispatchable domain located within the local sector correction region is oriented and corrected based on the real feasible boundary; The fourth module is used to update the local boundary chains located within the local sector correction region in the initial linearized schedulable domain based on the results of the directional correction, thereby generating an adaptive schedulable domain for the distribution network that takes into account the uncertainties of distributed photovoltaics.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method for constructing an adaptive dispatchable domain for a distribution network as described in any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for constructing an adaptive dispatchable domain for a distribution network as described in any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method for constructing an adaptive dispatchable domain for a distribution network as described in any one of claims 1 to 6.

Images