A Decision Tree-Based Method and System for Decision-Making Power Supply Paths to Important Users in a Power Grid

Abstract

This invention discloses a decision tree-based method and system for determining the power supply path for critical loads in a power grid. The method includes the following steps: reading the planned power grid topology to obtain the number of load nodes, critical load nodes, branches, switchable lines, substations, and 500kV substations; constructing constraints with the objective of minimizing the sum of load shedding amounts at each fault to obtain a decision model for the power supply path of critical users in the planned power grid; transforming the decision model using a multivariate decision tree to obtain a MILP model; and using the optimization solver Cplex to quickly solve the MILP model to obtain the power supply path for critical loads. This invention avoids the problem of blind spots in risk prevention and control in planned power grids, reduces the computational scale, realizes the decision of power supply paths for critical users, and ensures the safe and reliable power supply for critical users.

Description

Technical Field

[0001] This invention belongs to the technical field of power grid power supply path decision-making, specifically relating to a decision tree-based method and system for power supply path decision-making for important users in a power grid. Background Technology

[0002] Urban power grids contain numerous critical loads, such as important government agencies and hospitals. Power outages to these critical loads would result in significant losses and even incalculable consequences. Therefore, ensuring power supply to critical loads is a key concern in power grid operation. The most basic requirement for supplying power to critical loads is that there must be at least two power supply paths within the grid to ensure uninterrupted power supply to critical users even if one path fails. In current power grid planning practices, the decision-making process for power supply paths to critical users in planned power grids is primarily based on traditional analysis methods, performed manually. This method is inefficient, and there are currently no digital analysis tools for arranging power supply paths to upstream power sources for critical users and managing related risks. Furthermore, academic attention to this issue is limited, and there is a lack of theoretical guidance for practice. With the increasing complexity of power grid wiring and the growing number of critical users, traditional manual analysis methods rely heavily on the experience of planners. This results in a lack of closed-loop management of planning quality, low overall human efficiency, and a lack of intelligent auxiliary analysis tools for optimizing planned power grid operation and managing wiring schemes. This leads to blind spots in risk control within the planned power grid, which is detrimental to the safe and reliable power supply to critical users. Therefore, it is both necessary and urgent to research and develop decision-making methods and systems for power supply paths to important users of the power grid. Summary of the Invention

[0003] The main objective of this invention is to overcome the shortcomings and deficiencies of the prior art and provide a decision tree-based method and system for determining the power supply path of important users in a power grid. The method aims to minimize the sum of load shedding amounts of load nodes under various faults, constructs constraints, establishes a planning-state power grid important user power supply path decision model, and introduces a multivariate decision tree for this model to reduce the overall model size, thereby achieving fast and efficient solution and realizing the decision on the power supply path of important users.

[0004] To achieve the above objectives, the present invention provides a power supply path decision method for important users in a power grid based on a decision tree, comprising the following steps:

[0005] Read the planned power grid topology diagram to obtain the number of load nodes, the number of important load nodes, the number of branches, the number of lines that can be switched, the number of substations, and the number of 500kV substations;

[0006] With the objective of minimizing the sum of load shedding amounts at each load node under various faults, constraints are constructed to obtain a power supply path decision model for important users in the planned power grid. The constraints include power balance constraints at each load node, line capacity constraints, load shedding amounts under anticipated faults, and safe and reliable power supply constraints for important loads.

[0007] The MILP model is obtained by transforming the power supply path decision model for important users in the planned power grid using a multivariate decision tree.

[0008] The MILP model is solved quickly using the optimization solver Cplex to obtain the power supply path for critical loads.

[0009] As a preferred technical solution, the obtained power supply path decision model for important users in the planned power grid is specifically as follows:

[0010] With the objective of minimizing the sum of load shedding amounts at each fault node, the initial objective function of the power supply path decision model for important users in the planned power grid is obtained, expressed as:

[0011]

[0012] In equation (1), x = [x1, x2, ..., x k ] T These are the control variables for initializing the objective function, representing the switching states of switchable lines in the planned-state power grid. x is a vector; when x... k When x is 0, it means that the kth available line is not in operation. k When Ω is 1, it indicates that the kth switchable line is put into operation; c For the set of faults, Ω ld For the set of load nodes, The load shedding amount of load node i under the c-th fault;

[0013] Based on the initial objective function, the constraints for constructing the power supply path decision model for important users in the planned power grid are as follows:

[0014] The power balance constraints for each load node are constructed as follows:

[0015]

[0016]

[0017] Among them, P Gi Q represents the active power output of the generator at load node i. Gi P represents the reactive power output of the generator at load node i; Li Let Q be the initial active load of load node i. Li G represents the initial reactive load of load node i;ij (x) is the mutual conductance function between load nodes i and j, B ij (x) is the mutual susceptance function between load nodes i and j; e i and e j f represents the real part of the voltage at load nodes i and j, respectively; i and f j , where i and j are the imaginary parts of the voltages at nodes i and j, respectively, and n is the number of load nodes;

[0018] The line capacity constraint is constructed as follows:

[0019] S ij ≤ S ijmax (4)

[0020] Among them, S ij S represents the apparent power transmitted along the line between load nodes i and j; ijmax This represents the maximum apparent power of the line between load nodes i and j.

[0021] Construct load shedding constraints and reliable power supply constraints for critical loads under fault conditions. This means that when all main transformers in any 500kV substation experience a power outage due to a disconnection fault, after shedding the loads of some load nodes in the planned power grid, all critical loads should still be guaranteed a safe and reliable power supply with the required voltage amplitude. This is expressed as:

[0022]

[0023] in, f is the real part of the voltage at the i-th load node after the c-th fault occurs and the load is sheared. j c This represents the imaginary part of the voltage at the i-th load node after the c-th fault occurs and the load is cut off. Let i be the active power load shedding amount of load node i after the c-th fault occurs. Let i be the reactive load shedding amount of load node i after the c-th fault occurs; Let be the amplitude of the transmission power of line ij after the c-th fault occurs, i.e., the apparent transmission power. The voltage amplitude of the critical load node after the c-th fault occurs;

[0024] By combining equations (1)-(5), we obtain the power supply path decision model for important users in the planned power grid.

[0025] As a preferred technical solution, the transformation of the power supply path decision model for important users in the planned power grid using a multivariate decision tree to obtain the MILP model is as follows:

[0026] By introducing 0-1 variables and using a multivariate decision tree, the load shedding constraints under anticipated faults and the safe and reliable power supply constraints of important loads are transformed, thereby optimizing the initial objective function of the power supply path decision model for important users in the planned power grid.

[0027] Multivariate decision trees are used to transform the power balance constraints and line capacity constraints of each load node;

[0028] The optimized objective function and constraints are combined to obtain the MILP model.

[0029] As a preferred technical solution, the initialization objective function of the optimized planning-state power grid important user power supply path decision model is specifically as follows:

[0030] The control variable x of the initial objective function of the power supply path decision model for important users in the planned power grid is used as the object attribute value of the multivariate decision tree, and the load shedding amount is used as the attribute value of the multivariate decision tree.

[0031] Assuming the number of faults in a 500kV substation is M, the number of leaf nodes in the multivariate decision tree for the m-th 500kV substation fault is T.

[0032] Introduce a series of 0-1 variables s m,1 ,s m,2 ,…,s m,T , is used to describe the position of the control variable x in the corresponding leaf node of the multivariate decision tree;

[0033] Assuming that when the m-th 500kV substation fails, the r-th partitioning hyperplane traversed by leaf node t is... Then we have:

[0034]

[0035] In the formula, s m,t For the m-th 500kV substation experiencing a fault, the 0-1 variable corresponding to the t-th leaf node is... and b m,t,r When the m-th 500kV substation fails, the coefficient of the r-th dividing hyperplane traversed by leaf node t, where θ and x are vectors;

[0036] Since the control variable x can only be located at one leaf node at a time, for the same multivariate decision tree, the corresponding 0-1 variable s m,1 ,s m,2 ,…,s m,T satisfy:

[0037]

[0038] For each 500kV substation fault, assume that the load shedding amount corresponding to each leaf node in its multivariate decision tree is p. m,t At this point, the optimized initial objective function of the power supply path decision model for important users in the planned power grid is:

[0039]

[0040] As a preferred technical solution, the method of using a multivariate decision tree to transform the power balance constraints of each load node and the line capacity constraints specifically involves:

[0041] For the power balance constraints (2) and (3) and the line capacity constraints (4) of each load node, there are two cases: satisfied and not satisfied. The satisfied data is set to 1, and the unsatisfied data is set to 0. Then, it is transformed into the following by multivariate decision tree:

[0042]

[0043]

[0044]

[0045] Where, σ t σ is a 0-1 variable used to represent the position of the control variable x in the leaf nodes of the multivariate decision tree corresponding to the power balance constraints and line capacity constraints of each load node. When x is in the t-th leaf node, σ t =1, otherwise 0; T0 is the number of leaf nodes of the multivariate decision tree corresponding to the power balance constraints and line capacity constraints of each load node; and b t,r q is the coefficient of the r-th dividing hyperplane at the t-th leaf node; t For each leaf node, q represents the attribute value. When the power balance constraint and line capacity constraint of each load node have a solution, q t =1, and when there is no solution to the power balance constraints and line capacity constraints at each load node, q t =0.

[0046] As a preferred technical solution, the optimized objective function (13) and constraints (11), (12), (14), (15), and (16) are combined to obtain the MILP model (17), which is expressed as:

[0047]

[0048]

[0049] Equation (17) is solved using the optimization solver Cplex to obtain the power supply path decision for important users in the planned power grid.

[0050] As a preferred technical solution, variable substitution is used to transform equations (11) and (15) to obtain linear constraints (18) and (19), respectively:

[0051] Use variable d t Transforming equation (11) yields the linear constraint (18), which is expressed as:

[0052]

[0053] Where, d t =σ t x is a vector, d t，k For d t The k-th element, x k The k-th element of the control variable x;

[0054] Use variable c m，t Transforming equation (15) yields linear constraint (19), which is expressed as:

[0055]

[0056] Among them, c m，t =s m，t x is a vector, c m，t，k For c m，t The kth element.

[0057] As a preferred technical solution, the transformed objective function (13) and constraints (12), (14), (16), (18), and (19) are combined to obtain the MILP model (20), which is expressed as:

[0058]

[0059]

[0060] Equation (20) is solved using the optimization solver Cplex to obtain the power supply path decision for important users in the planned power grid.

[0061] In another aspect, the present invention provides a power supply path decision system for important users of the power grid based on decision trees, which is applied to the above-mentioned power supply path decision method for important users of the power grid based on decision trees, including a data acquisition module, a model building module, a model transformation module and a model solving module;

[0062] The data acquisition module is used to read the planned power grid topology map and obtain the number of load nodes, the number of important load nodes, the number of branches, the number of lines that can be switched, the number of substations, and the number of 500kV substations;

[0063] The model building module is used to construct constraints with the objective of minimizing the sum of load shedding amounts at load nodes under various faults, thereby obtaining a power supply path decision model for important users in the planned power grid. The constraints include power balance constraints at each load node, line capacity constraints, load shedding amounts under anticipated faults, and safe and reliable power supply constraints for important loads.

[0064] The model transformation module is used to transform the power supply path decision model for important users in the planned power grid using a multivariate decision tree to obtain the MILP model;

[0065] The model solving module is used to call the optimization solver Cplex to quickly solve the MILP model and obtain the power supply path of important loads.

[0066] In another aspect, the present invention provides a computer-readable storage medium storing a program that, when executed by a processor, implements the above-described decision tree-based power supply path decision method for critical users in a power grid.

[0067] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0068] 1. This invention establishes a mixed-integer nonlinear programming model for power supply path decision-making of important users in a planned power grid. With the goal of minimizing the load shedding after a fault, it incorporates load shedding constraints under anticipated faults and safe and reliable power supply constraints for important loads. When a fault occurs, it minimizes the load shedding of other load nodes while ensuring the power supply to important users, thereby reducing the economic losses caused by power supply failures.

[0069] 2. This invention quantitatively abstracts the power supply path decision problem for important users in a planned power grid into a mathematical expression, which facilitates subsequent analysis and use, thereby realizing automated analysis of power supply path arrangement. On the one hand, it improves the quality and efficiency of planning schemes, and on the other hand, it avoids the risks brought about by manual arrangement.

[0070] 3. This invention discretizes the power supply path decision for important users in a planned power grid, and discretizes the objective function of load shedding from a continuous function into a piecewise linear function, and discretizes the reliability constraint into two cases: reliable and unreliable, so as to facilitate description by a multivariate decision tree.

[0071] 4. This invention introduces a multivariate decision tree to transform the discretized objective function and the safety and reliability power supply constraints of important loads, turning them from a large number of nonlinear constraints into a series of multivariate decision trees, thereby reducing the problem size and computational difficulty.

[0072] 5. This invention simplifies multivariate decision trees by introducing 0-1 variables to simplify the complex tree structure into a series of simple mixed integer nonlinear constraints, making it easier to input into the solver for solving.

[0073] 6. This invention addresses the mixed-integer nonlinear constraints after simplification of multivariate decision trees by transforming them into a series of mixed-integer linear constraints through variable substitution. This allows for the rapid and efficient solution of the optimization problem using the mature optimization solver Cplex. Attached Figure Description

[0074] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0075] Figure 1 This is a flowchart of the power supply path decision method for important users in the power grid based on decision tree in an embodiment of the present invention;

[0076] Figure 2(a) is a binary mapping diagram of a univariate decision tree in an embodiment of the present invention;

[0077] Figure 2(b) is a schematic diagram of the structure of a univariate decision tree in an embodiment of the present invention;

[0078] Figure 3(a) is a mapping diagram of the multivariate decision tree in an embodiment of the present invention;

[0079] Figure 3(b) is a schematic diagram of the structure of the multivariate decision tree in an embodiment of the present invention;

[0080] Figure 4 This is a schematic diagram of the power supply path before a power grid failure in a city center in China.

[0081] Figure 5 This invention provides a power supply path diagram determined by the method of this invention after a power grid failure in a central city in China.

[0082] Figure 6 This is a structural diagram of the power supply path decision system for important users of the power grid based on decision tree in an embodiment of the present invention;

[0083] Figure 7 This is a schematic diagram of the structure of a computer-readable storage medium in an embodiment of the present invention. Detailed Implementation

[0084] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of the present application, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present application without creative effort are within the scope of protection of the present application.

[0085] In this application, the reference to "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described in this application can be combined with other embodiments.

[0086] like Figure 1 As shown in the figure, this embodiment discloses a power supply path decision method for important users in a power grid based on a decision tree, including the following steps:

[0087] S1. Read the planned power grid topology diagram to obtain the number of load nodes, the number of important load nodes, the number of branches, the number of lines that can be switched, the number of substations, and the number of 500kV substations;

[0088] S2. With the objective of minimizing the sum of load shedding amounts at each load node under various faults, constraints are constructed to obtain the power supply path decision model for important users in the planned power grid. The constraints include power balance constraints at each load node, line capacity constraints, load shedding amounts under anticipated faults, and safe and reliable power supply constraints for important loads.

[0089] S3. Use a multivariate decision tree to transform the power supply path decision model for important users in the planned power grid, and obtain the MILP model;

[0090] S4. Call the optimization solver Cplex to quickly solve the MILP model and obtain the power supply path of the important load.

[0091] Furthermore, the specific steps for establishing the power supply path decision model for important users in the planned power grid in step S2 are as follows:

[0092] S201. With the objective of minimizing the sum of load shedding amounts at each fault node, the initial objective function of the power supply path decision model for important users in the planned power grid is obtained, expressed as:

[0093]

[0094] In equation (1), x = [x1, x2, ..., x k ]T These are the control variables for initializing the objective function, representing the switching states of switchable lines in the planned-state power grid. x is a vector; when x... k When x is 0, it means that the kth available line is not in operation. k When Ω is 1, it indicates that the kth switchable line is put into operation; c For the set of faults, Ω ld For the set of load nodes, The load shedding amount of load node i under the c-th fault;

[0095] S202. Based on the initial objective function, construct the constraints for the power supply path decision model for important users in the planned power grid, including:

[0096] ① Construct power balance constraints for each load node, expressed as:

[0097]

[0098]

[0099] Among them, P Gi Q represents the active power output of the generator at load node i. Gi P represents the reactive power output of the generator at load node i; Li Let Q be the initial active load of load node i. Li G represents the initial reactive load of load node i; ij (x) is the mutual conductance function between load nodes i and j, B ij (x) represents the mutual susceptance function between load nodes i and j; both the mutual conductance function and the mutual susceptance function are functions of the control variable x; e i and e j f represents the real part of the voltage at load nodes i and j, respectively; i and f j , where i and j are the imaginary parts of the voltages at nodes i and j, respectively, and n is the number of load nodes;

[0100] ② Construct line capacity constraints, expressed as:

[0101] S ij ≤S ijmax (4)

[0102] Among them, S ij S represents the apparent power transmitted along the line between load nodes i and j; ijmax This represents the maximum apparent power of the line between load nodes i and j.

[0103] ③ Construct load shedding constraints and safe and reliable power supply constraints for critical loads under fault conditions. This means that when all main transformers in any 500kV substation experience a power outage due to a disconnection fault, after shedding the loads of some load nodes in the planned power grid, all critical loads should still be guaranteed safe and reliable power supply with the required voltage amplitude. This can be expressed as:

[0104]

[0105] in, f is the real part of the voltage at the i-th load node after the c-th fault occurs and the load is sheared. j c This represents the imaginary part of the voltage at the i-th load node after the c-th fault occurs and the load is cut off. Let i be the active power load shedding amount of load node i after the c-th fault occurs. Let i be the reactive load shedding amount of load node i after the c-th fault occurs; Let be the amplitude of the transmission power of line ij after the c-th fault occurs, i.e., the apparent transmission power. The voltage amplitude of the critical load node after the c-th fault occurs;

[0106] S203, combining equations (1)-(5), yields the power supply path decision model for important users in the planned power grid.

[0107] Since the obtained model is a large-scale mixed-integer nonlinear programming (MINLP) model, its computational efficiency is very low or even unsolvable when applied to a real large power grid. Therefore, this invention introduces a multivariate decision tree to transform the intelligent decision model for power supply paths of important users in the planned power grid into a small-scale mixed-integer linear programming (MILP) model for fast solution.

[0108] A decision tree represents a mapping relationship between object attribute values ​​and object values. In a univariate decision tree, each node represents an object, each branch path represents a possible attribute value, and each leaf node corresponds to the value of the object represented by the path from the root node to that leaf node. As shown in Figure 2, a decision tree essentially describes a mapping relationship. Taking a binary mapping y = f(x1, x2) as an example, Figure 2(a) shows the binary mapping, and Figure 2(b) shows its corresponding univariate decision tree. However, a univariate decision tree selects one attribute for each branch value, so its decision boundary consists of several segments parallel to the coordinate axes. In practical applications, the boundary of the division is often not parallel to the coordinate axes. If a univariate decision tree is still used, on the one hand, it will cause too many leaf nodes, which will bring a large amount of computation to the subsequent optimization problem solution; on the other hand, univariate decision trees are prone to overfitting, which will reduce the accuracy of the problem solution. Therefore, this invention introduces a multivariate decision tree to describe the mapping relationship, thereby reducing the amount of computation and improving the computational accuracy.

[0109] Multivariate decision trees use linear combinations of multiple attribute values ​​as the splitting criterion, as shown in Figure 3. The splitting criterion for each step in a multivariate decision tree is the hyperplane in the variable space. Each split considers far more information than a univariate decision tree, therefore its ability to characterize the actual situation is significantly higher. The training method for multivariate decision trees is as follows:

[0110] Let the training set be S = {(x i ,y i If )}, then the loss function of the entire multivariate decision tree is:

[0111]

[0112] Where, x i Let y be the value of the control variable for the i-th sample. i Let N be the value of the load shedding for the i-th sample. s ω is the number of samples in the training set. it The probability that the i-th sample falls in the t-th leaf node can be obtained from the statistics of the data samples; Let t be the value of the t-th leaf node; λ represents the fitting error; γ and λ are penalty coefficients used to simplify the structure of the multivariate decision tree; T is the number of leaf nodes.

[0113] As can be seen from equation (6), there is no coupling between the leaf nodes, meaning the objective function can decouple the leaf nodes. In other words, for a single leaf node, when the loss of the entire decision tree is minimized, the loss of each leaf node is also minimized. From equation (6), the loss function is... The quadratic function can be obtained from the property of the extreme points of a quadratic function:

[0114]

[0115]

[0116] Among them, Loss t Let be the minimum loss of the t-th leaf node; the minimum loss of the entire multivariate decision tree is equal to the sum of the minimum losses of each leaf node.

[0117] The sigmoid function φ(·) is introduced to describe whether a sample lies on the dividing hyperplane. Weights on both sides:

[0118]

[0119] The score for a single split of a multivariate decision tree is:

[0120]

[0121] Its significance lies in the reduction of the loss of the entire multivariate decision tree after a single partition, i.e., the minimum loss of the parent node minus the difference between the minimum losses of the two child nodes; by maximizing the score S of each partition, the optimal partitioning hyperplane for that partition can be obtained. After training the multivariate decision tree with a series of samples, the values ​​of all partitioning hyperplanes and all leaf nodes of the multivariate decision tree can be obtained, thus obtaining a complete multivariate decision tree. This tree is then used to transform the power supply path decision model for important users in a planned power grid, resulting in the MILP model, specifically:

[0122] S301. Since the main computational load of the power supply path decision model for important users in the planned power grid comes from the load shedding constraints under anticipated faults and the safe and reliable power supply constraints of important loads (5), this embodiment introduces 0-1 variables. Based on the load shedding constraints under anticipated faults and the safe and reliable power supply constraints of important loads, the constraint (5) is transformed from a non-convex nonlinear constraint with integer variables into a linear constraint with integer variables. This optimizes the initialization objective function of the power supply path decision model for important users in the planned power grid, thereby improving the computational efficiency of model solving. Specifically:

[0123] Constraint (5) describes the mapping relationship between load shedding and control variable x under a fault in a 500kV substation. Therefore, control variable x is regarded as the object attribute value of the multivariate decision tree, and load shedding is regarded as the attribute value of the multivariate decision tree. On the other hand, for a specific 500kV substation fault, if its corresponding constraint (5) has a solution, that is, it can guarantee the power supply of important users by shedding load, the power supply reliability constraint of important users must be satisfied. Therefore, the load shedding constraint and the power supply reliability constraint of important users can be considered together. It is only necessary to set the load shedding of the data that cannot satisfy the power supply reliability constraint of important loads to a large value.

[0124] Assume that the total number of faults in all 500kV substations is M, and for the m-th 500kV substation fault, the number of leaf nodes in its corresponding multivariate decision tree is T.

[0125] Introduce a series of 0-1 variables s m,1 ,s m,2 ,…,s m,T , is used to describe the position of the control variable x in the corresponding leaf node of the multivariate decision tree;

[0126] Assuming that when the m-th 500kV substation fails, the r-th partitioning hyperplane traversed by leaf node t is... Then we have:

[0127]

[0128] In the formula, s m,t For the m-th 500kV substation experiencing a fault, the 0-1 variable corresponding to the t-th leaf node is... and b m,t,r When the m-th 500kV substation fails, the coefficient of the r-th dividing hyperplane traversed by leaf node t, where θ and x are vectors;

[0129] Since the control variable x can only be located at one leaf node at a time, for the same multivariate decision tree, the corresponding 0-1 variable s m,1 ,s m,2 ,…,s m,T satisfy:

[0130]

[0131] For each 500kV substation fault, assume that the load shedding amount corresponding to each leaf node in its multivariate decision tree is p. m,t At this point, the optimized initial objective function of the power supply path decision model for important users in the planned power grid is:

[0132]

[0133] S302. Similarly, a multivariate decision tree is used to transform the power balance constraints and line capacity constraints of each load node, specifically as follows:

[0134] For the power balance constraints (2) and (3) and the line capacity constraints (4) of each load node, there are two cases: satisfied and not satisfied. The satisfied data is set to 1, and the unsatisfied data is set to 0. Then, it is transformed into the following by multivariate decision tree:

[0135]

[0136]

[0137]

[0138] Where, σ t σ is a 0-1 variable used to represent the position of the control variable x in the leaf node of the multivariate decision tree corresponding to constraints (2)-(4). When x is in the t-th leaf node, σ t =1, otherwise 0; T0 is the number of leaf nodes of the multivariate decision tree corresponding to constraints (2)-(4); and b t,r q is the coefficient of the r-th dividing hyperplane at the t-th leaf node; t For each leaf node, q represents the attribute value. When constraints (2)-(4) have a solution, q t =1, and when constraints (2)-(4) have no solution, q t =0;

[0139] Constraint (14) is used to restrict the control variable x to fall on only one leaf node of the multivariate decision tree;

[0140] Constraint (15) is the partitioning hyperplane constraint corresponding to each leaf node;

[0141] Constraint (16) is used to restrict the control variable x to fall only on q. t =1, meaning the leaf nodes where the power balance constraints and line capacity constraints of each load node can be satisfied.

[0142] S303. Combine the optimized objective function (13) with constraints (11), (12), (14), (15), and (16) to obtain the MILP model (17), which is expressed as:

[0143]

[0144]

[0145] In model (17), the original nonlinear constraints are replaced by constraints corresponding to some multivariate decision trees, which greatly reduces the number of constraints.

[0146] However, since model (17) still contains non-convex nonlinear constraints (11) and (15) related to 0-1 variables, solving the entire model remains somewhat difficult. To address this issue, this embodiment of the invention also replaces constraints (11) and (15) with some variable substitutions, transforming them into linear constraints (18) and (19), thereby improving computational efficiency. Specifically:

[0147] Use variable d t Transforming equation (11) yields the linear constraint (18), which is expressed as:

[0148]

[0149] Where, d t =σ t x is a vector, d t,k For d t The k-th element, x k The k-th element of the control variable x;

[0150] Use variable c m,t Transforming equation (15) yields linear constraint (19), which is expressed as:

[0151]

[0152] Among them, c m,t =s m,t x is a vector, c m,t,k For c m,t The kth element.

[0153] The transformed objective function (13) and constraints (12), (14), (16), (18), and (19) are combined to obtain the MILP model (20), which is expressed as:

[0154]

[0155]

[0156] In summary, the obtained MILP models (17) and (20) are solved quickly using the mature commercial optimization solver Cplex to obtain the power supply path decisions for important users in the planned power grid.

[0157] To verify the effectiveness of the above embodiments, this embodiment uses a central power grid in a Chinese city as an example for simulation calculations. Figure 4As shown, the central power grid contains 1560 load nodes, 1668 branch lines, 53 generators, and 7 500kV substations, one of which has its 500kV side busbar designated as the balancing node of the power grid system. There are also 90 220kV substations and 553 110kV substations. 109 lines in the power grid are designated as switchable lines, and 20 load nodes are designated as critical loads. Six anticipated faults are set, each involving the failure and disconnection of all main transformers in one of the six unbalanced 500kV substations. Calculations were performed on the system, and the results are shown in Table 1. Table 1 compares the load shedding amounts corresponding to each fault under the method of this embodiment with those under normal operating conditions without considering line switching. Table 1 shows that before optimization, the total system load shedding for the six anticipated faults was 2267MW, while after optimization, the total system load shedding for the six anticipated faults decreased to 2203MW, demonstrating the effectiveness of the decision-making method proposed in this invention.

[0158] Table 1. Calculation results of anticipated fault substations and corresponding load shedding quantities

[0159]

[0160] Taking the ZIJ 500kV substation experiencing a fault where all main transformers disconnect, and the PINGA 220kV substation's 110kV side load as a critical load as another example, Figure 4-5 This shows the power supply paths for critical loads before and after the fault. Figure 4 It can be seen that before the fault, the critical load had three power supply paths, each coming from a different power source. The specific paths are as follows:

[0161] LINGAO 500kV busbar—SHENZ 500kV busbar—PENGC 500kV busbar—ZIJ 500kV busbar—ZIJ220kV busbar—PINGA 220kV busbar—PINGA 110kV side load;

[0162] BAOA 500kV busbar—PENGC 500kV busbar—ZIJ 500kV busbar—ZIJ 220kV busbar—PINGA 220kV busbar—PINGA 110kV side load;

[0163] MAW Power Plant 220kV Busbar—XIANX 220kV Busbar—ZIJ 220kV Busbar—PINGA 220kV Busbar—PINGA 110kV Side Load;

[0164] Following the fault, all three main transformers at the ZIJ substation disconnected, preventing power transmission from the 500kV bus to the 220kV bus. Therefore, the power supply path from the 500kV bus at the BAOA substation no longer existed. The power supply path from the LINGAO power plant, however, could be achieved by connecting other 500kV substations to the 220kV grid via main transformers, and then supplying power to critical loads. The path became:

[0165] LINGAO Power Plant 500kV Busbar—SHENZ 500kV Busbar—PENGC 500kV Busbar—ZIJ 500kV Busbar—XIAND 500kV Busbar—XIAND 220kV Busbar—FEIC 220kV Busbar—ZIJ 220kV Busbar—PINGA 220kV Busbar—PINGA Station 110kV Side Load;

[0166] Furthermore, the power supply path of the coal-fired unit's MAW (Medium-voltage Power Flow) only supplies power to important loads through the 220kV grid, therefore it is unaffected by this fault. Moreover, after this anticipated fault, the voltage of the PINGA 110kV bus is 0.932 pu, which is greater than 0.9 pu, meeting the safety power supply requirements. The calculation results demonstrate that the method proposed in this embodiment can effectively ensure that important users in the system still have sufficient power supply after a power outage caused by a complete main transformer disconnection fault at a 500kV substation.

[0167] It should be noted that, for the sake of simplicity, the aforementioned method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, because according to the present invention, some steps can be performed in other orders or simultaneously.

[0168] Based on the same idea as the decision tree-based power supply path decision method for important power grid users in the above embodiments, the present invention also provides a decision tree-based power supply path decision system for important power grid users. This system can be used to execute the above-described decision tree-based power supply path decision method for important power grid users. For ease of explanation, the structural diagram of the decision tree-based power supply path decision system for important power grid users only shows the parts related to the embodiments of the present invention. Those skilled in the art will understand that the illustrated structure does not constitute a limitation on the device, and may include more or fewer components than illustrated, or combine certain components, or have different component arrangements.

[0169] like Figure 6 As shown, another embodiment of the present invention provides a power supply path decision system for important users in a power grid based on a decision tree, including a data acquisition module, a model building module, a model transformation module, and a model solving module;

[0170] The data acquisition module is used to read the planned power grid topology map and obtain the number of load nodes, the number of important load nodes, the number of branches, the number of lines that can be switched, the number of substations, and the number of 500kV substations.

[0171] The model building module is used to construct constraints with the goal of minimizing the sum of load shedding amounts at load nodes under various faults, and obtain the power supply path decision model for important users in the planned power grid. The constraints include power balance constraints at each load node, line capacity constraints, load shedding amounts under anticipated faults, and safe and reliable power supply constraints for important loads.

[0172] The model transformation module is used to transform the power supply path decision model for important users in the planned power grid using a multivariate decision tree, resulting in a MILP model.

[0173] The model solving module is used to call the optimization solver Cplex to quickly solve the MILP model and obtain the power supply path of important loads.

[0174] It should be noted that the power supply path decision system for important loads in the power grid based on decision trees of the present invention corresponds one-to-one with the power supply path decision method for important loads in the power grid based on decision trees of the present invention. The technical features and beneficial effects described in the embodiments of the power supply path decision method for important loads in the power grid based on decision trees are applicable to the embodiments of the power supply path decision system for important loads in the power grid based on decision trees. For details, please refer to the description in the embodiments of the method of the present invention, which will not be repeated here.

[0175] Furthermore, in the above embodiments of the power grid critical load power supply path decision system based on decision tree, the logical division of each program module is only an example. In actual applications, the above functions can be assigned to different program modules as needed, for example, for the sake of corresponding hardware configuration requirements or software implementation convenience. That is, the internal structure of the power grid critical load power supply path decision system based on decision tree is divided into different program modules to complete all or part of the functions described above.

[0176] like Figure 7 As shown, in one embodiment, a computer-readable storage medium is provided, storing a program in a memory. When the program is executed by a processor, it implements the aforementioned decision tree-based power supply path decision method for critical loads in a power grid, specifically:

[0177] Read the planned power grid topology diagram to obtain the number of load nodes, the number of important load nodes, the number of branches, the number of lines that can be switched, the number of substations, and the number of 500kV substations;

[0178] With the objective of minimizing the sum of load shedding amounts at each load node under various faults, constraints are constructed to obtain a power supply path decision model for important users in the planned power grid. The constraints include power balance constraints at each load node, line capacity constraints, load shedding amounts under anticipated faults, and safe and reliable power supply constraints for important loads.

[0179] The MILP model is obtained by transforming the power supply path decision model for important users in the planned power grid using a multivariate decision tree.

[0180] The MILP model is solved quickly using the optimization solver Cplex to obtain the power supply path for critical loads.

[0181] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments described above. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0182] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0183] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be considered equivalent substitutions and shall be included within the protection scope of the present invention.

Claims

1. A decision tree-based method for determining power supply paths to critical users in a power grid, characterized in that, Includes the following steps: Read the planned power grid topology diagram to obtain the number of load nodes, the number of important load nodes, the number of branches, the number of lines that can be switched, the number of substations, and the number of 500kV substations; With the objective of minimizing the sum of load shedding amounts at each load node under various fault conditions, a constraint condition is constructed to obtain the power supply path decision model for important users in the planned power grid. The constraint conditions include power balance constraints at each load node, line capacity constraints, load shedding amounts under anticipated fault conditions, and safe and reliable power supply constraints for important loads. The initial objective function of the power supply path decision model for important users in the planned power grid is expressed as: In equation (1), x = [x1, x2, ..., x k ] T These are the control variables for initializing the objective function, representing the switching states of switchable lines in the planned-state power grid. x is a vector; when x... k When x is 0, it means that the kth available line is not in operation. k When Ω is 1, it indicates that the kth switchable line is put into operation; c For the set of faults, Ω ld For the set of load nodes, The load shedding amount of load node i under the c-th fault; the constraints include power balance constraints of each load node, line capacity constraints, load shedding amount constraints under fault, and safe and reliable power supply constraints of important loads. The MILP model is obtained by transforming the power supply path decision model for important users in the planned power grid using a multivariate decision tree. By introducing 0-1 variables and using a multivariate decision tree, the load shedding constraints under anticipated faults and the safe and reliable power supply constraints of important loads are transformed, thereby optimizing the initial objective function of the power supply path decision model for important users in the planned power grid. Multivariate decision trees are used to transform the power balance constraints and line capacity constraints of each load node; The optimized objective function and constraints are combined to obtain the MILP model; The MILP model is solved quickly using the optimization solver Cplex to obtain the power supply path for critical loads.

2. The power supply path decision method for critical loads in a power grid based on decision trees according to claim 1, characterized in that, Based on the initial objective function, the constraints for constructing the power supply path decision model for important users in the planned power grid are as follows: The power balance constraints for each load node are constructed as follows: Among them, P Gi Q represents the active power output of the generator at load node i. Gi P represents the reactive power output of the generator at load node i; Li Let Q be the initial active load of load node i. Li G represents the initial reactive load of load node i; ij (x) is the mutual conductance function between load nodes i and j, B ij (x) is the mutual susceptance function between load nodes i and j; e i and e j f represents the real part of the voltage at load nodes i and j, respectively; i and f j , where i and j are the imaginary parts of the voltages at nodes i and j, respectively, and n is the number of load nodes; Constructing line capacity constraints, expressed as: S ij ≤S ijmax (4) Among them, S ij S represents the apparent power transmitted along the line between load nodes i and j; ijmax This represents the maximum apparent power of the line between load nodes i and j. Construct load shedding constraints and reliable power supply constraints for critical loads under fault conditions. This means that when all main transformers in any 500kV substation experience a power outage due to a disconnection fault, after shedding the loads of some load nodes in the planned power grid, all critical loads should still be guaranteed a safe and reliable power supply with the required voltage amplitude. This is expressed as: in, f is the real part of the voltage at the i-th load node after the c-th fault occurs and the load is sheared. j c This represents the imaginary part of the voltage at the i-th load node after the c-th fault occurs and the load is cut off. Let i be the active power load shedding amount of load node i after the c-th fault occurs. Let i be the reactive load shedding amount of load node i after the c-th fault occurs; Let be the transmission power amplitude of line ij after the c-th fault occurs, i.e., the transmission apparent power. The voltage amplitude of the critical load node after the c-th fault occurs; By combining equations (1)-(5), we obtain the power supply path decision model for important users in the planned power grid.

3. The power supply path decision method for important users in a power grid based on decision trees according to claim 2, characterized in that, The initial objective function of the optimized planning-state power grid important user power supply path decision model is as follows: The control variable x of the initial objective function of the power supply path decision model for important users in the planned power grid is used as the object attribute value of the multivariate decision tree, and the load shedding amount is used as the attribute value of the multivariate decision tree. Assuming the number of faults in a 500kV substation is M, the number of leaf nodes in the multivariate decision tree for the m-th 500kV substation fault is T. Introduce a series of 0-1 variables s m,1 ,s m,2 ,…,s m,T , is used to describe the position of the control variable x in the corresponding leaf node of the multivariate decision tree; Assuming that when the m-th 500kV substation fails, the r-th partitioning hyperplane traversed by leaf node t is... Then we have: In the formula, s m,t For the m-th 500kV substation experiencing a fault, the 0-1 variable corresponding to the t-th leaf node is... and b m,t,r When the m-th 500kV substation fails, the coefficient of the r-th dividing hyperplane traversed by leaf node t, where θ and x are vectors; Since the control variable x can only be located at one leaf node at a time, for the same multivariate decision tree, the corresponding 0-1 variable s m,1 ,s m,2 ,…,s m,T satisfy: For each 500kV substation fault, assume that the load shedding amount corresponding to each leaf node in its multivariate decision tree is p. m,t At this point, the optimized initial objective function of the power supply path decision model for important users in the planned power grid is:

4. The power supply path decision method for important users in a power grid based on decision trees according to claim 3, characterized in that, The process of transforming the power balance constraints and line capacity constraints of each load node using a multivariate decision tree is as follows: For the power balance constraints (2) and (3) and the line capacity constraints (4) of each load node, there are two cases: satisfied and not satisfied. The satisfied data is set to 1, and the unsatisfied data is set to 0. Then, it is transformed into the following by multivariate decision tree: Where, σ t σ is a 0-1 variable used to represent the position of the control variable x in the leaf nodes of the multivariate decision tree corresponding to the power balance constraints and line capacity constraints of each load node. When x is in the t-th leaf node, σ t =1, otherwise 0; T0 is the number of leaf nodes of the multivariate decision tree corresponding to the power balance constraints and line capacity constraints of each load node; and b t,r q is the coefficient of the r-th dividing hyperplane at the t-th leaf node; t For each leaf node, q represents the attribute value. When the power balance constraint and line capacity constraint of each load node have a solution, q t =1, and when there is no solution to the power balance constraints and line capacity constraints at each load node, q t =0.

5. The power supply path decision method for important users in a power grid based on decision trees according to claim 4, characterized in that, The optimized objective function (13) and constraints (11), (12), (14), (15), and (16) are combined to obtain the MILP model (17), which is expressed as: Equation (17) is solved using the optimization solver Cplex to obtain the power supply path decision for important users in the planned power grid.

6. The power supply path decision method for important users in a power grid based on decision trees according to claim 4, characterized in that, By using variable substitution to transform equations (11) and (15), we obtain linear constraints (18) and (19), respectively: Use variable d t Transforming equation (11) yields the linear constraint (18), which is expressed as: Where, d t =σ t x is a vector, d t,k For d t The k-th element, x k The k-th element of the control variable x; Use variable c m,t Transforming equation (15) yields linear constraint (19), which is expressed as: Among them, c n,t =s m,t x is a vector, c m,t,k For c m,t The kth element.

7. The power supply path decision method for important users in a power grid based on decision trees according to claim 6, characterized in that, The transformed objective function (13) and constraints (12), (14), (16), (18), and (19) are combined to obtain the MILP model (20), which is expressed as: Equation (20) is solved using the optimization solver Cplex to obtain the power supply path decision for important users in the planned power grid.

8. A power supply path decision system for critical users in a power grid based on decision trees, characterized in that, The power supply path decision method for important users of the power grid based on decision tree, as described in any one of claims 1-7, includes a data acquisition module, a model building module, a model transformation module, and a model solving module; The data acquisition module is used to read the planned power grid topology map and obtain the number of load nodes, the number of important load nodes, the number of branches, the number of lines that can be switched, the number of substations, and the number of 500kV substations; The model building module is used to construct constraints with the objective of minimizing the sum of load shedding amounts at load nodes under various faults, thereby obtaining a power supply path decision model for important users in the planned power grid. The constraints include power balance constraints at each load node, line capacity constraints, load shedding amounts under anticipated faults, and safe and reliable power supply constraints for important loads. The model transformation module is used to transform the power supply path decision model for important users in the planned power grid using a multivariate decision tree to obtain the MILP model; The model solving module is used to call the optimization solver Cplex to quickly solve the MILP model and obtain the power supply path of important loads.

9. A computer-readable storage medium storing a program, characterized in that, When the program is executed by the processor, it implements the power supply path decision method for important users of the power grid based on any one of claims 1-7.

Images