CN-121984013-A - Power grid operation risk control method for coping with extreme scenes

CN121984013ACN 121984013 ACN121984013 ACN 121984013ACN-121984013-A

Abstract

The application discloses a power grid operation risk control method for coping with extreme scenes, and relates to the technical field of power system operation decision-making, wherein the method comprises the steps of clustering all nodes in a power grid by using a k-means clustering algorithm according to a power fluctuation characteristic measurement value of each node in the power grid to generate the extreme scenes; embedding a physical calculation process of the power grid operation risk in an extreme scene into a training frame of a neural network, carrying out piecewise linearization characterization on the power grid operation risk in the extreme scene by using a large M method, constructing a scheduling model, and solving the scheduling model by adopting an optical growth optimization algorithm to obtain an optimal operation strategy of the power grid when the power grid operation risk in the extreme scene is considered. According to the method, the scheduling model of the power grid is constructed when the running risk of the power grid in the extreme scene is considered, and the running risk resistance of the power grid in the extreme scene is improved.

Inventors

JIANG KAI
LIU NIAN
HAN ZHONGLIANG

Assignees

华北电力大学

Dates

Publication Date: 20260505
Application Date: 20260129

Claims (10)

1. A power grid operation risk control method for coping with extreme scenes, characterized by comprising: According to the historical data of the power grid, determining the power fluctuation characteristic measurement value of each node in the power grid respectively; According to the power fluctuation characteristic measurement value, clustering all nodes in the power grid by using a k-means clustering algorithm to generate an extreme scene; Quantifying a physical calculation process of the power grid operation risk in a polar scene; Embedding a physical calculation process of the power grid operation risk in the extreme scene into a training frame of the neural network, and carrying out piecewise linearization characterization on the power grid operation risk in the extreme scene by using a large M method; Performing piecewise linearization characterization based on the power grid operation risk in an extreme scene, and constructing a scheduling model; solving the scheduling model by adopting an optical growth optimization algorithm to obtain an optimal operation strategy of the power grid when the operation risk of the power grid in an extreme scene is considered; and executing the optimal operation strategy.
2. The method for controlling risk of operation of a power grid in response to an extreme scenario of claim 1, wherein the historical data of the power grid comprises wind power data, photovoltaic power data and load power data of each node in the power grid.
3. The method for controlling the running risk of a power grid coping with an extreme scenario according to claim 1, wherein the power fluctuation feature measurement value is: ; Wherein, the ; ; ; ; ; In the formula, The power fluctuation characteristic measurement value of the ith node; Weighting factors for fluctuation intensity of the maximum fluctuation rate; the maximum fluctuation rate of the ith node; Weighting factors for fluctuation intensity of fluctuation standard deviation; the fluctuation standard deviation of the ith node; Is a natural exponential function; The method comprises the steps of (1) balancing a weight coefficient of the maximum fluctuation rate, wherein I is a power grid node set; The T is the number of time periods and the value is 24; The net load power at the t time of the ith node; The net load power at the t-1 time of the ith node; a power reference value for the i-th node; the daily average value of the net load power of the ith node; the load power at the t moment of the i node; The wind power at the t moment of the ith node; The photovoltaic power at the t moment of the i node.
4. The method for controlling the running risk of the power grid for coping with the extreme scene according to claim 1, wherein the clustering processing is performed on all nodes in the power grid by using a k-means clustering algorithm according to the power fluctuation characteristic measurement value, and the method specifically comprises the following steps: according to the power fluctuation characteristic measurement value, clustering all nodes in the power grid by using a k-means clustering algorithm to obtain a plurality of clusters; Determining an average characteristic value of an extreme scene corresponding to each cluster respectively, wherein the average characteristic value of the extreme scene is an average value of power fluctuation characteristic measurement values of all nodes in the corresponding cluster; all clusters are arranged in descending order according to the average characteristic value of the extreme scene; And determining the clustering center of the first cluster after descending order as an extreme scene.
5. The method for controlling the running risk of the power grid for coping with an extreme scene according to claim 1, wherein a calculation formula of the running risk of the power grid in the extreme scene is: ; Wherein, the ; ; ; ; In the formula, The total operation risk brought by the boundary adjustment of the kth device at the t moment; Direct operation risk brought by boundary adjustment of kth equipment at the t moment; the diffusion operation risk brought by the boundary adjustment of the kth device at the t moment; forced outage rate for the kth device; the power of the kth device at the t moment; A set of lines affected by the kth device boundary adjustment; Forced outage rate for the jth line affected by the kth device boundary adjustment; Power after affected by boundary adjustment for the jth line; is a self-defined step function; the power before the j-th line is affected by the boundary adjustment; the method comprises the steps of collecting power generating sets; The adjustment quantity of the kth generator set at the t moment; A power flow transfer distribution factor of the kth generator set to the jth line is used; is a power grid line set; the adjustment quantity of the first line at the t moment; A line break distribution factor for the first line to the j line; the limit transmission power for the j-th line.
6. The method for controlling the running risk of the power grid for coping with the extreme scene according to claim 5, wherein the physical calculation process of the running risk of the power grid in the extreme scene is embedded into a training frame of the neural network, and the large M method is used for carrying out piecewise linearization characterization on the running risk of the power grid in the extreme scene, and specifically comprises the following steps: establishing a main network model and a main network loss function; The method comprises the steps of integrating physics of the power grid operation risk in an extreme scene into a main network model, and constructing a physical information fusion neural network by adding a physical loss item between a predicted value and a calculated value, wherein the predicted value is a predicted value of the power grid operation risk in the extreme scene, which is output by inputting a feature vector into the main network model, and the feature vector comprises equipment power, equipment power adjustment quantity, line power and line power adjustment quantity; Setting an activation function as a ReLU function, and training a physical information fusion neural network; And the representation of the segmented ReLU function is equivalent to the maximum problem, the maximum problem is converted into a constraint form by using a large M method, so that the main network can be represented as a piecewise linear function, and the piecewise linear representation of the running risk of the power grid in the extreme scene is completed.
7. The method for controlling the running risk of a power grid coping with an extreme scenario according to claim 6, wherein the predicted value is: ; In the formula, In order to be able to predict the value, 、、 And Respectively a weight matrix of a layer 1, a layer 2, a layer 3 and a layer 4 of the main network model; 、、 And Respectively the bias vectors of the 1 st layer, the 2 nd layer, the 3 rd layer and the 4 th layer of the main network model; is an input feature vector; The primary network loss function is: ; Wherein, the ; ; In the formula, Loss for the primary network model; Data loss; Is a physical loss; The weight of the physical loss of the main network model; The number of samples is a single batch; The method comprises the steps that a power grid operation risk prediction value is obtained under an extreme scene of a b sample of a main network; is the true value of the running risk of the power grid in the extreme scene of the b sample; is a calculated value of the running risk of the power grid in an extreme scene.
8. The method for controlling the running risk of a power grid coping with an extreme scene as recited in claim 1, wherein the scheduling model includes an objective function and a constraint condition; the constraint conditions comprise power balance constraint, line tide constraint and unit output upper and lower limit constraint; The objective function is: ; Wherein T is the number of time periods; the method comprises the steps of collecting power generating sets; the generation cost coefficient of the ith generator set; the power at the t moment of the ith generating set; penalty coefficients for cut load; The load shedding power at the t moment; is a power grid line set; As a factor of the cost of the risk, The total operation risk brought by the boundary adjustment of the kth device at the t moment; The power balance constraint is: ; In the formula, Is a load node set; the net load power at the t moment of the ith node in the extreme scene; The line tide constraint is as follows: ; In the formula, Limit transmission power for the first line; the adjustment quantity of the first line at the t moment; The power of the first line at the t moment; a power flow transfer distribution factor of the ith node to the first line; The upper and lower limit constraints of the output of the unit are as follows: ; In the formula, The maximum technical output of the ith generating set at the t moment; the minimum technical output of the ith generating set at the t moment is obtained.
9. The method for controlling the running risk of the power grid for coping with the extreme scene according to claim 1, wherein the optimal running strategy of the power grid in consideration of the running risk of the power grid in the extreme scene is obtained by solving the scheduling model by adopting an optical growth optimization algorithm, and specifically comprises the following steps: initializing a cell population according to the preset total number of cells Na and the upper and lower boundaries of variables in the scheduling model; the method comprises the steps of determining a parent population of the current iteration, wherein the initialized cell population is taken as the parent population when the current iteration number is equal to 1, and the optimal population at the last iteration is taken as the parent population of the current iteration when the current iteration number is greater than 1; Dividing all cells in the parent population of the iteration into cells in an illumination area and cells in a shadow area; determining an initial population of daughter cells using a cell division strategy, comprising dividing each illuminated region cell into two daughter cells using a cell division strategy; updating the initial sub-cell population by using a cell elongation strategy to obtain an updated sub-cell population of the iteration; Combining the sub-cell pairs with corresponding relation of the updated sub-cell population into one cell to obtain a sub-population of the iteration; determining the combined set of the parent population of the iteration and the child population of the iteration as a total population; respectively calculating the objective function value corresponding to each cell in the total population; And (3) arranging all cells in the total population in ascending order according to the objective function value, determining the previous Na cells as the optimal population in the current iteration, entering the next iteration, stopping the cycle until the maximum iteration number is reached, and determining the cell corresponding to the minimum objective function value in the optimal population as the optimal operation strategy of the power grid in consideration of the power grid operation risk in the extreme scene.
10. The method for controlling risk of operation of a power grid in response to an extreme scenario according to claim 9, The cell division strategy is: ; In the formula, A first subcellular generated for the ith illumination zone cell of the s-th iteration; cells randomly selected for the whole region; Is a growth limiting factor that decays with iteration number; Is a random mutation direction factor; 、、、、 And Are random numbers between-1 and 1; The cells in the illumination area are the ith iteration; Cells that are optimal for the target value of the illumination zone; a second subcellular generated for the ith illumination zone cell of the s-th iteration; a first subcellular generated for the ith iteration number i shadow region cell; the cells are the shadow cells of iteration number i; a second subcellular generated for the ith iteration number i shadow region cell; Randomly selected cells for the illumination zone; ; In the formula, The ith cell is the ith iteration updated in the cell elongation strategy; the ith cell in the s-th iteration of the cell elongation strategy; is the tortuosity coefficient of the cell; The i+1th cell in the s-th iteration of the cell elongation strategy; Cells that are optimal for the whole area target; Is a random number between-1 and 1.

Description

Power grid operation risk control method for coping with extreme scenes Technical Field The application relates to the technical field of operation decision-making of power systems, in particular to a power grid operation risk control method for coping with extreme scenes. Background Under the background of large-scale grid connection of high-proportion new energy sources and frequent extreme weather events in an electric power system, the electric power system is facing serious operation safety challenges. On one hand, the high-proportion renewable energy source represented by wind power and photovoltaic is connected in a large scale and the power load spike degree is deepened, so that the power balance difficulty of a power system is increased sharply, on the other hand, extreme events such as typhoons, ice disasters and the like are frequent due to global climate change, and the power supply reliability is seriously threatened. Under the double pressures of difficult balance and difficult supply protection, the safety boundary of the power grid is continuously tightened, and the running risk of the power grid presents the characteristics of dynamic diffusivity and the like. Therefore, it is necessary to expand a common safety constraint economic dispatch model, construct a dispatch model considering the running risk of the power grid, and strengthen the anti-risk capability of the running of the power grid so as to solve the problem of the prominent running risk of the power grid in extreme scenes. Disclosure of Invention The application aims to provide a power grid operation risk control method for coping with an extreme scene, which can improve the risk resistance of power grid operation in the extreme scene. In order to achieve the above object, the present application provides the following solutions: the application provides a power grid operation risk control method for coping with extreme scenes, which comprises the following steps: According to the historical data of the power grid, determining the power fluctuation characteristic measurement value of each node in the power grid respectively; According to the power fluctuation characteristic measurement value, clustering all nodes in the power grid by using a k-means clustering algorithm to generate an extreme scene; Quantifying a physical calculation process of the power grid operation risk in a polar scene; Embedding a physical calculation process of the power grid operation risk in the extreme scene into a training frame of the neural network, and carrying out piecewise linearization characterization on the power grid operation risk in the extreme scene by using a large M method; Performing piecewise linearization characterization based on the power grid operation risk in an extreme scene, and constructing a scheduling model; solving the scheduling model by adopting an optical growth optimization algorithm to obtain an optimal operation strategy of the power grid when the operation risk of the power grid in an extreme scene is considered; and executing the optimal operation strategy. Optionally, the grid history data comprises wind power data, photovoltaic power data and load power data of each node in the grid. Optionally, the power fluctuation characteristic measurement value is: ; Wherein, the ; ; ; ; ; In the formula,The power fluctuation characteristic measurement value of the ith node; Weighting factors for fluctuation intensity of the maximum fluctuation rate; the maximum fluctuation rate of the ith node; Weighting factors for fluctuation intensity of fluctuation standard deviation; the fluctuation standard deviation of the ith node; Is a natural exponential function; The method comprises the steps of (1) balancing a weight coefficient of the maximum fluctuation rate, wherein I is a power grid node set; The T is the number of time periods and the value is 24; The net load power at the t time of the ith node; The net load power at the t-1 time of the ith node; a power reference value for the i-th node; the daily average value of the net load power of the ith node; the load power at the t moment of the i node; The wind power at the t moment of the ith node; The photovoltaic power at the t moment of the i node. Optionally, according to the power fluctuation characteristic measurement value, clustering all nodes in the power grid by using a k-means clustering algorithm to generate an extreme scene, which specifically comprises the following steps: according to the power fluctuation characteristic measurement value, clustering all nodes in the power grid by using a k-means clustering algorithm to obtain a plurality of clusters; Determining an average characteristic value of an extreme scene corresponding to each cluster respectively, wherein the average characteristic value of the extreme scene is an average value of power fluctuation characteristic measurement values of all nodes in the corresponding cluster; all clusters are