CN-121981329-A - Power system planning method, device, medium and equipment

CN121981329ACN 121981329 ACN121981329 ACN 121981329ACN-121981329-A

Abstract

The invention discloses a power system planning method, a device, a medium and equipment, and relates to the technical field of power system planning. The invention decomposes the complex large-scale planning problem into two aspects of capacity optimization problem and position optimization problem, wherein the number of capacity optimization problem variables is far less than that of the original large-scale planning problem, so that the solving speed is faster, namely, the invention splits the single-time complex planning problem into multiple-time simple planning problem solving, thereby reducing the solving difficulty. The method eliminates internal variables of all subsystems in a large-scale power system, thereby reducing the scale of the planning problem. And the parallel Monte Carlo tree search algorithm is adopted to process the position optimization problem, so that enumeration of all installation positions is avoided, the solving efficiency is improved, and the planning effect of the electric power system is improved.

Inventors

BIE CHAOHONG
HUANG BINGKAI
HUANG YUXIONG
LI GENGFENG
DAI ZICHENG

Assignees

西安交通大学

Dates

Publication Date: 20260505
Application Date: 20260129

Claims (8)

1. A method of power system planning, comprising: Acquiring historical data of power grid parameters, user loads and new energy output, and constructing a planning model of each element to be planned in the power system for installation positions and installation capacities by taking the minimum planning cost and the minimum running cost of the power system as targets; The method comprises the steps that tree nodes of each layer in a Monte Carlo tree are used for representing the installation positions of elements to be planned, and the installation positions of the elements to be planned are determined to be valued through a parallel Monte Carlo tree searching algorithm; according to the installation position values and the planning model, constructing and splitting an installation capacity optimization problem of each element to be planned, and obtaining an upper-layer capacity solving problem based on each subsystem operation domain in the power system and a lower-layer operation optimization problem representing each subsystem under different operation scenes at different moments; According to the lower-layer operation optimization problem, the projection of each subsystem in the operation domain under all the operation scenes at all the moments is solved in parallel, the upper-layer capacity solving problem is solved according to the projection of each subsystem operation domain, the optimal value of the objective function of the planning model under the value of the installation position is determined according to the solving result, and the Monte Carlo tree is updated according to the optimal value of the objective function; and (3) carrying out iterative planning until reaching the search termination condition of the Monte Carlo tree, and determining the installation position and the installation capacity of each element to be planned according to the optimal value of the objective function corresponding to each leaf node in the Monte Carlo tree.
2. The power system planning method according to claim 1, wherein the determining the installation position value of the element to be planned by the parallel monte carlo tree searching algorithm specifically includes: Initializing a Monte Carlo tree root node, and setting the access times and return of the root node to 0; Determining an idle process, searching from a root node of the Monte Carlo tree through the idle process, determining a target child node with the maximum upper confidence limit according to the access times and returns of each child node of the currently traversed tree node, and taking the target child node as a tree node of the next traversal until traversing to the bottommost layer of the current Monte Carlo tree; after the search is finished, if the currently traversed tree node is a leaf node, determining the installation position value of each element to be planned corresponding to the leaf node; If the currently traversed tree node is an expandable tree node, updating all the child nodes of the tree node to a Monte Carlo tree, setting the access times and return of all the child nodes to be 0, and randomly selecting one child node from all the child nodes as the next traversed tree node until traversing to the leaf node to determine the installation position value of each element to be planned corresponding to the leaf node.
3. The power system planning method according to claim 1, wherein the determining the target child node with the maximum upper confidence bound according to the access times and returns of each child node of the currently traversed tree node specifically comprises: determining an upper confidence bound according to the access times and returns of each child node of the currently traversed tree node by the following steps: ; determining a target child node with the maximum upper confidence limit according to the upper confidence limit of each child node of the currently traversed tree node; Wherein, the Representing tree nodes Is set in the upper confidence limit of (1), Representing tree nodes In return for (a), Representing tree nodes Access times of (C) tree node Representing tree nodes Is provided with a node (a) which is a parent node of the (c), Representing tree nodes Is a return of (a).
4. The power system planning method of claim 1 wherein the upper layer capacity solution problem is: ; ; ; ; The lower-layer operation optimization problem is as follows: ; ; ; Wherein, the 、 And (3) with Respectively a subsystem, a planning scene and a running time, Is that The equipment to be planned in the subsystem is in Under the operation scene, Vectors of continuous running variables at time, For all of The resulting vectors are stacked in a column-expanded fashion, Is that The subsystem has other elements except the element to be planned Under the operation scene, Vectors of continuous running variables at time, In order to plan the cost of the device, For the running costs of the elements to be planned, Representative of Subsystem is at Under the operation scene, The running cost at the moment of time is high, Is that The weight of the running scene is calculated, Vectors constructed for the left-hand terms of all configuration constraints, The vector of left-hand terms of the constraints is run for all elements to be planned, Is that Subsystem is at Under the operation scene, Vectors formed by the left-hand terms of all running constraints at a time instant, Is that Is used for the upper mirror image of the lens, Is that Given the estimated value of the upper limit, Takes a value for a group of installation positions, Equivalent projection of the feasible region of the optimization problem for the underlying run.
5. The power system planning method according to claim 1, wherein the parallel solving of the projections of the operation domains of each subsystem under all time points of all operation scenes according to the lower operation optimization problem specifically comprises: And adopting a progressive vertex enumeration algorithm to parallelly solve projections of the operation domains of all subsystems under all operation scenes at all moments according to the lower operation optimization problem.
6. An electrical power system planning apparatus, comprising: The centralized server acquires historical data of power grid parameters, user loads and new energy output, and builds a planning model of each element to be planned in the power system for installation positions and installation capacities by taking the minimum planning cost and the minimum running cost of the power system as targets; the method comprises the steps of representing the installation position of each element to be planned by tree nodes of each layer in a Monte Carlo tree, determining the value of the installation position of each element to be planned by a parallel Monte Carlo tree search algorithm, constructing an installation capacity optimization problem of each element to be planned according to the value of the installation position and a planning model, splitting the installation capacity optimization problem to obtain an upper-layer capacity solving problem based on each subsystem operation domain in an electric power system and a lower-layer operation optimization problem representing each subsystem at different moments in different operation scenes; and each distributed computer solves the projection of the operation domain of the corresponding subsystem under all operation scenes at all moments in parallel according to the lower-layer operation optimization problem.
7. A computer readable storage medium, characterized in that the storage medium stores a computer program which, when executed by a processor, implements the method according to any one of claims 1-5.
8. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the method of any one of claims 1 to 5 when the computer program is executed.

Description

Power system planning method, device, medium and equipment Technical Field The present invention relates to the field of power system planning technologies, and in particular, to a power system planning method, apparatus, medium, and device. Background Modern power systems are currently evolving towards multi-regional, multi-operator interconnects. Through large power grid interconnection, the operation efficiency of the power grid can be effectively improved, the reliability is enhanced, low carbon is promoted, and the like. Among these, large-scale power system interconnection planning is key to achieving the above advantages. In particular, power system planning focuses on solving the most preferred addressing and sizing problem of newly added elements in the system, i.e. where the best node it installs in the network for a certain element, what the best capacity it needs to install. In the prior art, aiming at the power system planning problem, the most common method at present is to model the power system planning problem as a mathematical optimization problem, and determine the optimal configuration result by using the conditions of minimum sum of the running cost and the investment cost of the power system, meeting the load flow constraint, the power and electricity balance constraint and the like. However, when the problem of planning a large-scale power system is faced, the solving efficiency is too low to meet the large-scale requirements. In order to solve the problem of large-scale power system planning, further common methods can be summarized into two types of model simplification and model splitting. For the former, the concept of layering and grouping is often adopted to simplify a large planning power system. The term "layering" refers to splitting a large-scale power system through voltage levels, for example, in a power transmission network-power distribution network planning, a power distribution network with a low voltage level is often considered as a load node, a specific structure inside the power distribution network is omitted, and similarly, in the power distribution network planning, a power transmission network with a high voltage level is considered as a power supply node. The term "grouping" refers to dividing a network into a plurality of subsystems according to the distribution characteristics of users in a network with the same voltage level, especially in a power distribution network, and planning each subsystem respectively. From the perspective of solving the optimization problem, the model splitting adopts a specific mathematical method to split the mathematical optimization problem obtained by modeling into a plurality of small-scale sub-problems, and solves the original large-scale optimization problem by an iterative convergence method. Common resolution methods can be classified into original decomposition and dual decomposition 2 types. The original decomposition includes Benders decomposition, which includes lagrangian relaxation, and the like, and the methods can be further combined with an alternate direction multiplier (ADMM) method, a cut plane method, and the like. However, the large-scale power system has the characteristics of interconnection of multiple operators (multiple subsystems), compared with the traditional power system, the number of planning nodes and feeder lines is obviously increased, the power system planning problem needs to consider the operation characteristics of time sequence coupling elements (such as generators, energy storage and the like), the problem scale is further increased due to multi-period coupling, in addition, the new energy and novel load proportion in the power system are increased, the operation uncertainty of the power system is enhanced, and the uncertainty fluctuation of novel source load needs to be described by considering multiple operation scenes. For this reason, the constraint condition and the number of decision variables of the large-scale power system planning problem are extremely large, which brings great challenges to the solution of the problem, i.e. the solver cannot find a feasible solution (i.e. any planning scheme satisfying the constraint is not generated) or can find a poor suboptimal solution (i.e. some planning schemes satisfying the constraint are found, but the effect of these planning schemes is not good) for the problem within a certain time. In summary, the current power system planning method has poor planning effect and planning efficiency. Disclosure of Invention Based on the foregoing, it is necessary to provide a power system planning method, apparatus, medium and device for solving the above technical problems. The invention adopts the following technical scheme: the invention provides a power system planning method, which comprises the following steps: Acquiring historical data of power grid parameters, user loads and new energy output, and constructing a planning model of each element