Power grid dispatching scheme determination method and device, computer equipment, storage medium and program product

Abstract

The invention relates to a power grid dispatching scheme determination method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the steps that a security constraint unit combination of a power grid in the operation process is acquired, the security constraint unit combination comprises a unit combination in the operation process and security constraint conditions for the unit combination, and the unit combination comprises power grid nodes in the power grid and branches formed between the power grid nodes; according to the security constraint unit combination, determining node features corresponding to the power grid nodes; determining an association relationship among the branches based on the node characteristics, and determining an effective security constraint based on the association relationship and a power grid fault condition; determining a target security constraint unit combination in the operation process of the power grid based on the effective security constraint and the unit combination; and according to the target security constraint unit combination, obtaining a scheduling scheme for the power grid in the operation process. The method is helpful for improving the determination efficiency of the scheduling scheme.

Description

Technical Field

[0001] The present application relates to the field of power technology, and particularly to a method, device, computer device, computer-readable storage medium, and computer program product for determining a power grid scheduling scheme. Background Art

[0002] The unit commitment (UC) problem aims to minimize the total operating cost within a scheduling period and determine the start-stop states and power output plans of units while satisfying unit constraints and system constraints. Security Constrained Unit Commitment (SCUC) is UC considering the security constraints of the power system. The power grid scheduling scheme determined based on SCUC can not only ensure the stable operation of the system but also improve the economic benefits of system operation. However, currently, the efficiency of determining the power grid scheduling scheme based on SCUC is not good. Summary of the Invention

[0003] Based on this, it is necessary to provide a method, device, computer device, computer-readable storage medium, and computer program product for determining a power grid scheduling scheme to solve the above technical problems and improve the efficiency of determining the power grid scheduling scheme based on Security Constrained Unit Commitment.

[0004] In a first aspect, the present application provides a method for determining a power grid scheduling scheme, including:

[0005] Obtain the security-constrained unit commitment during the operation of the power grid, where the security-constrained unit commitment includes the unit commitment during the operation and the security constraint conditions for the unit commitment, and the unit commitment includes the power grid nodes in the power grid and the branches formed between the power grid nodes;

[0006] Determine the node characteristics corresponding to each power grid node according to the security-constrained unit commitment;

[0007] Determine the association relationship between the branches based on the node characteristics, and determine the effective security constraints based on the association relationship and the power grid fault conditions; the effective security constraints include the security constraints satisfied by the non-fault branches in the unit commitment under the power grid fault conditions;

[0008] Determine the target security-constrained unit commitment during the operation of the power grid based on the effective security constraints and the unit commitment;

[0009] Obtain the scheduling scheme for the power grid during the operation according to the target security-constrained unit commitment.

[0010] In one of the embodiments, determining the node characteristics corresponding to each power grid node according to the security-constrained unit commitment includes:

[0011] Determine the topological structure of the power grid according to the security-constrained unit commitment.

[0012] Construct a graph model corresponding to the power grid according to the topological structure.

[0013] Determine the node features based on the power information corresponding to each power grid node in the graph model.

[0014] In one embodiment, constructing a graph model corresponding to the power grid based on the topological structure includes:

[0015] Determine the vertices of the graph model corresponding to the power grid based on each power grid node in the topological structure.

[0016] Determine the edges of the graph model corresponding to the power grid based on each branch in the topological structure.

[0017] Construct a graph model corresponding to the power grid according to the vertices and edges in accordance with the topological structure.

[0018] In one embodiment, determining the node features based on the power information corresponding to each power grid node in the graph model includes:

[0019] Obtain the power information corresponding to each power grid node.

[0020] Determine the node type to which each power grid node in the graph model belongs according to the power information; the node types include generation nodes, load nodes, and generation-load hybrid nodes.

[0021] Determine the node features according to the node type and the power information.

[0022] In one embodiment, the power grid fault conditions include power grid single-fault conditions.

[0023] Determine the correlation relationship between each branch based on the node features, and determine the effective security constraints based on the correlation relationship and the power grid fault conditions, including:

[0024] Input the node features into a pre-trained security constraint identification model, and the security constraint identification model determines the correlation relationship between each branch based on the node features, and determines and outputs the effective security constraints based on the correlation relationship and the power grid fault conditions.

[0025] The effective security constraints include the security constraints satisfied by the non-fault branches in the unit commitment under the power grid single-fault conditions.

[0026] In one embodiment, obtaining the security-constrained unit commitment during the operation of the power grid includes:

[0027] Determine the pre-configured power grid scheduling period for the power grid.

[0028] Under the condition of meeting the power grid dispatching cycle, perform the step of obtaining the security-constrained unit combination of the power grid during operation.

[0029] In a second aspect, the present application also provides a power grid dispatching scheme determination device, including:

[0030] An acquisition module, configured to acquire the security-constrained unit combination of the power grid during operation, where the security-constrained unit combination includes the unit combination during operation and the security constraint conditions for the unit combination, and the unit combination includes each power grid node in the power grid and the branches formed between each power grid node;

[0031] A constraint determination module, configured to determine the node characteristics corresponding to each power grid node according to the security-constrained unit combination; determine the association relationship between each branch based on the node characteristics, and determine the effective security constraints based on the association relationship and the power grid fault conditions; the effective security constraints include the security constraints satisfied by the non-fault branches in the unit combination under the power grid fault conditions;

[0032] A scheme determination module, configured to determine the target security-constrained unit combination of the power grid during operation based on the effective security constraints and the unit combination; obtain the dispatching scheme for the power grid during operation according to the target security-constrained unit combination.

[0033] In a third aspect, the present application also provides a computer device, including a memory and a processor, where the memory stores a computer program, and when the processor executes the computer program, the steps of the power grid dispatching scheme determination method in the first aspect are implemented.

[0034] In a fourth aspect, the present application also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the power grid dispatching scheme determination method in the first aspect are implemented.

[0035] In a fifth aspect, the present application also provides a computer program product, including a computer program, and when the computer program is executed by a processor, the steps of the power grid dispatching scheme determination method in the first aspect are implemented.

[0036] The above method, device, computer equipment, computer-readable storage medium, and computer program product for determining a power grid dispatching scheme determine the node characteristics corresponding to each power grid node for the security-constrained unit commitment in the power grid operation process, determine the association relationship between each branch based on the node characteristics, and determine the effective security constraints based on the association relationship and power grid fault conditions. This realizes the screening of security constraints, proposes redundant or invalid security constraints, and thus obtains the target security-constrained unit commitment. Since the target security-constrained unit commitment contains effective security constraints, and the effective security constraints are obtained based on the screening of security constraints, therefore, compared with determining the dispatching scheme according to the security-constrained unit commitment, determining the dispatching scheme according to the target security-constrained unit commitment requires less computational effort and better efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following will briefly introduce the drawings required for the description in the embodiments of the present application or related technologies. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other related drawings can be obtained based on these drawings.

[0038] Figure 1 It is a schematic flowchart of a method for determining a power grid dispatching scheme in an embodiment;

[0039] Figure 2 It is another schematic flowchart of a method for determining a power grid dispatching scheme in an embodiment;

[0040] Figure 3 It is a schematic diagram of node types in an embodiment;

[0041] Figure 4 It is a schematic diagram of the IEEE 118-node system in an embodiment;

[0042] Figure 5 It is yet another schematic flowchart of a method for determining a power grid dispatching scheme in an embodiment;

[0043] Figure 6 It is a structural block diagram of a device for determining a power grid dispatching scheme in an embodiment;

[0044] Figure 7 It is an internal structural diagram of a computer device in an embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0045] In order to make the objectives, technical solutions, and advantages of the present application clearer, the following further describes the present application in detail with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0046] Before introducing the technical solution of this application, the following explanations are made for relevant technical terms:

[0047] Security-constrained unit commitment refers to formulating a multi-period unit start-stop plan with the lowest system power purchase cost, etc. as the optimization goal under the condition of meeting the security constraints of the power system.

[0048] Power Flow refers to the distribution and flow of electric power in the power grid. For example, the voltage, phase angle of each node, and active and reactive power of each branch, etc.

[0049] Unit Commitment (abbreviated as "UC") refers to a solution that optimizes the start-stop and output of generating units to minimize the system operation cost or maximize the economic benefit during the operation of the power system.

[0050] Security-constrained Unit Commitment (abbreviated as "SCUC") is the UC under the condition of meeting the security constraints of the power system.

[0051] Base state power flow usually refers to the power flow distribution of the power system under normal operating conditions, that is, the steady-state operating point when no faults or abnormal conditions occur. In this state, the power generation and load of the system are balanced, and all electrical parameters such as voltage, current, power, etc. are within their rated values or allowable ranges.

[0052] The undirected graphical model, also known as the Markov Random Field (MRF) or Markov network, is a statistical model used to represent the probabilistic dependence relationship between variables. Different from the directed graphical model (such as the Bayesian network), the edges in the undirected graphical model have no direction, which means they do not represent a causal relationship, but represent the direct correlation or dependence relationship between variables.

[0053] The heterogeneous graph model refers to a graph model with diverse node or edge types, that is, there are at least two different types of nodes or edges in the graph. Compared with the homogeneous graph, the heterogeneous graph can better reflect the diversity of complex systems in the real world because it allows the existence of different types of entities and their interaction relationships.

[0054] The N-1 principle, also known as the single-fault safety inspection rule, means that in a power system under normal operating conditions, any component (such as a line, generator, transformer, DC monopole, etc.) is free of faults or disconnected due to faults, and the power system should be able to maintain stable operation and normal power supply, with other components not overloaded, and the voltage and frequency within the allowable range.

[0055] After the above description of technical terms, the technical solutions provided by this application will be described below:

[0056] With the technological progress and cost reduction of renewable energy, renewable energy has become an important source of power production. However, renewable energy is uncertain, which poses challenges to traditional power systems. New power systems usually have higher flexibility and adjustability, which helps to address the above challenges.

[0057] However, the large-scale access of renewable energy and the emergence of various new loads have made the optimal dispatching and operation analysis of power systems face severe challenges. For the unit commitment (SCUC) considering N-1 power flow security constraints, the optimal solution is generally obtained through an iterative solution framework. The number of variables and constraints that traditional optimization methods need to consider has increased significantly, especially the scale of N-1 constraints is huge. In addition, with the development of intelligent technologies, the quality and quantity of data that power systems can obtain are constantly improving. Traditional optimization methods may not be able to make full use of this large-scale and high-quality data to achieve more accurate prediction and decision-making.

[0058] On the contrary, traditional optimization methods can address these challenges by combining data-driven methods. Data-driven methods, such as machine learning and deep learning, can learn the complex patterns and rules of power systems from a large amount of historical data and generate more intelligent, flexible, and adaptable unit commitment schemes to meet the requirements of modern power systems for efficient, reliable, and sustainable operation. Therefore, to ensure more rapid and effective solution of complex SCUC, it is necessary to pre-eliminate redundant N-1 constraints using historical massive big data before solving the physical-driven model. By predicting the effective N-1 constraints that come into play, the performance of the original physical-driven model on the solver can be optimized, significantly reducing the solution time.

[0059] Based on this, this application provides a method for determining an effective safety constraint set for a power grid. This method is aimed at the security-constrained unit commitment during the operation of the power grid. By determining the node characteristics corresponding to each power grid node and determining the effective safety constraints based on the node characteristics, the target security-constrained unit commitment is obtained, and the dispatching scheme is determined according to the target security-constrained unit commitment, reducing the corresponding computational amount and having better efficiency. The following will be further described by way of examples:

[0060] In one embodiment, as Figure 1As shown, a method for determining a power grid dispatching scheme is provided. In this embodiment, an example is given where this method is applied to a power grid dispatching platform. It can be understood that this method can also be applied to servers, computer devices, etc. involved in power grid dispatching, such as servers of a power grid dispatching system, and can also be applied to systems including those involved in power grid dispatching. The system can include various computer devices and servers, and the computer devices and servers can interact with each other. In this embodiment, the method includes steps S101 to S105:

[0061] Step S101, obtain the security-constrained unit commitment during the operation of the power grid. The security-constrained unit commitment includes the unit commitment during the operation and the security constraint conditions for the unit commitment. The unit commitment includes the power grid nodes in the power grid and the branches formed between the power grid nodes.

[0062] Among them, the power grid can include the entire power system. Exemplarily, it can include, but is not limited to, power generation equipment, power transmission equipment, power distribution equipment, load equipment, etc.

[0063] Among them, the security constraint conditions can include the security constraints for the unit commitment. In some embodiments, the security constraint conditions can be the security constraint conditions determined based on the N-1 principle.

[0064] Among them, the power grid node can include the electrical connection points of power equipment, such as generators, transformers, loads, etc.

[0065] Among them, the branch can include the transmission line or distribution line connecting two power grid nodes, etc.

[0066] In some embodiments, the computer device corresponding to the power grid dispatching platform can obtain the security-constrained unit commitment during the operation of the power grid. Exemplarily, based on the operation situation of the power grid during the operation, the operation situation includes the specific composition structure of the power grid, the operation data of each component, etc. Through the analysis of the operation situation, the security-constrained unit commitment during the operation of the power grid can be obtained. Of course, when the power grid is operating, the corresponding security-constrained unit commitment can be automatically generated based on a pre-set program, so that the computer device can directly obtain it. Regarding the acquisition of the security-constrained unit commitment, there is no limitation here.

[0067] Step S102, determine the node characteristics corresponding to each power grid node according to the security-constrained unit commitment.

[0068] Among them, the node characteristics can be the characteristics related to the operation situation of the power grid node.

[0069] Exemplarily, for the power grid node undertaking the power generation task, the determined corresponding node characteristics can involve the active power situation of the generator, etc.

[0070] Step S103: Determine the association relationships between branches based on node features, and determine effective security constraints based on the association relationships and grid fault conditions; the effective security constraints include the security constraints satisfied by non-fault branches in the unit commitment under grid fault conditions.

[0071] Among them, the association relationship can represent the direct or indirect influence relationship between branches. For example, branch B starts depending on the start of branch A, and the start of branch B affects the current magnitude of branch C, etc.

[0072] In some embodiments, the computer device corresponding to the power grid dispatching platform can determine the association relationships between branches according to the N-1 principle and in combination with node features, so as to determine effective security constraints.

[0073] Exemplarily, assume that the power grid nodes include a, b, and c, the corresponding node features include A, B, and C, and the corresponding branches include a-b, a-c, and b-c. According to the N-1 principle, assume that branch a-b fails. According to the node features including A, B, and C, determine whether the remaining branches a-c and b-c are affected. If branches a-c and b-c are affected and change, for example, the current of branch a-c increases to 88A, then the corresponding security constraint can be found, such as the current limit value of branch a-c is 100A. Since the current of branch a-c, 88A, is less than 100A, therefore, this security constraint is satisfied, and the corresponding effective security constraint is obtained.

[0074] Step S104: Determine the target security constraint unit commitment during the operation of the power grid based on the effective security constraints and the unit commitment.

[0075] Exemplarily, through the foregoing steps, the effective security constraints can be determined. The computer device corresponding to the power grid dispatching platform can combine the effective security constraints with the unit commitment to obtain a new security constraint unit commitment - that is, the target security constraint unit commitment.

[0076] Step S105: Obtain the dispatching plan for the power grid during operation according to the target security constraint unit commitment.

[0077] In some embodiments, the Lagrangian relaxation method can be used to solve the target security constraint unit commitment, so as to obtain the dispatching plan. Of course, other solution methods can also be used, such as heuristic algorithms, specifically, the priority order method, etc.

[0078] In some embodiments, by solving the target security constraint unit commitment, the corresponding optimal solution can be obtained. Then, based on the N-1 principle, the optimal solution can be verified to determine whether there are any missing effective security constraints. If it is determined that there are no omissions, the dispatching plan can be obtained according to the optimal solution.

[0079] In some embodiments, if there are missing valid security constraints, the missing valid security constraints can be added to the security-constrained unit commitment, and then the node characteristics, valid security constraints, target security-constrained unit commitment, optimal solution, etc. are re-determined, and then the optimal solution is verified again until there are no missing valid security constraints. This can ensure the accuracy of the target security-constrained unit commitment, thereby helping to improve the accuracy of the scheduling scheme.

[0080] The above method for determining the power grid scheduling scheme is directed to the security-constrained unit commitment in the process of power grid operation, determines the node characteristics corresponding to each power grid node, determines the association relationship between each branch based on the node characteristics, and determines the valid security constraints based on the association relationship and the power grid fault conditions, which realizes the screening of security constraints, proposes redundant or invalid security constraints, and thus obtains the target security-constrained unit commitment. Since the target security-constrained unit commitment contains valid security constraints, and the valid security constraints are obtained based on the screening of security constraints, therefore, compared with determining the scheduling scheme according to the security-constrained unit commitment, the amount of calculation required to determine the scheduling scheme according to the target security-constrained unit commitment is less and the efficiency is better.

[0081] In one of the embodiments, as Figure 2 shown, the aforementioned "determining the node characteristics corresponding to each power grid node according to the security-constrained unit commitment" may include steps S201 to S203:

[0082] Step S201, determining the topological structure of the power grid according to the security-constrained unit commitment.

[0083] In some embodiments, the computer device determines the topological structure of the power grid according to the security-constrained unit commitment to ensure that there is a correspondence relationship between the topological structure and the security-constrained unit commitment. In other words, the obtained topological structure is not an arbitrary topological structure, but the topological structure corresponding to the security-constrained unit commitment.

[0084] In some embodiments, when the topological structure of the power grid changes, the security-constrained unit commitment changes accordingly.

[0085] Step S202, constructing a graph model corresponding to the power grid according to the topological structure.

[0086] In some embodiments, the graph model corresponding to the power grid constructed by the computer device according to the topological structure can be an undirected graph model.

[0087] Step S203, determining the node characteristics based on the power information corresponding to each power grid node in the graph model.

[0088] In some embodiments, the power information corresponding to each power grid node in the graph model may include the power generation situation and / or the load situation of each power grid node. Exemplarily, for a power grid node involving a power generation device, the corresponding power information may include the power generation situation.

[0089] Based on the topological structure of the power grid, a corresponding graph model is constructed. Through the power information corresponding to each power grid node in the graph model, the node features can be determined more quickly and accurately.

[0090] In one embodiment, based on the topological structure, a graph model corresponding to the power grid is constructed, including: determining the vertices of the graph model corresponding to the power grid based on each power grid node in the topological structure; determining the edges of the graph model corresponding to the power grid based on each branch in the topological structure; and constructing the graph model corresponding to the power grid according to the vertices and edges in accordance with the topological structure.

[0091] In some embodiments, the graph model may be a heterogeneous graph model. Each vertex in the graph model may belong to the same or different types. Similarly, each edge in the graph model may also belong to the same or different types. Since the graph model can be a heterogeneous graph model, and each power grid node and each branch in the power grid topological structure may belong to the same or different types, therefore, the graph model can more accurately reflect the real situation of the power grid, thus providing a more accurate basis for determining the node features.

[0092] In the above technical solution, by corresponding the power grid nodes to the vertices of the graph model and corresponding each branch to the edges of the graph model, and constructing the graph model according to the vertices and edges in accordance with the topological structure. This enables the constructed graph model to more accurately reflect the actual situation of the power grid and provides a more accurate basis for determining the node features.

[0093] In one embodiment, based on the power information corresponding to each power grid node in the graph model, determining the node features includes: obtaining the power information corresponding to each power grid node; determining the node type to which each power grid node in the graph model belongs according to the power information; the node types include power generation nodes, load nodes, and power generation - load hybrid nodes; and determining the node features according to the node type and the power information.

[0094] Among them, the power information can be understood in a broad sense, and it may include information related to the power grid node. For example, the actual computer device information corresponding to the power grid node, etc.

[0095] In some embodiments, different node types can be divided according to the characteristics or functional roles of the power grid nodes themselves. Exemplarily, the node types may include power generation nodes, load nodes, and power generation - load hybrid nodes.

[0096] Among them, the power generation node can be a type of power grid node representing those responsible for power generation; similarly, the load node can be a type of power grid node representing those with loads, and the power generation and load mixed node can be a type of power grid node representing those with loads while undertaking power generation.

[0097] In some embodiments, the node characteristics can be determined based on the different node types to which each power grid node belongs and according to the power information corresponding to each power grid node. For example, the computer device corresponding to power grid node A is a generator, and power grid node A belongs to a power generation node. According to the power information corresponding to power grid node A, the corresponding node characteristics can be obtained. For example, the node characteristics can include the maximum output active power of the generator, etc. Another example is that power grid node B belongs to a load node. According to the power information corresponding to power grid node B, the corresponding node characteristics can be obtained. For example, the node characteristics can include the active load condition of the node within a certain time interval.

[0098] By determining the node types to which each power grid node in the graph model belongs, where the node types include power generation nodes, load nodes, and power generation and load mixed nodes, and according to the node types and power information, the node characteristics are determined. This realizes the distinction of different power grid nodes, and different types of power grid nodes can correspond to different node characteristics, so that the node characteristics can more accurately reflect the actual situation of the power grid.

[0099] In one embodiment, the power grid fault condition includes a power grid single fault condition; based on the node characteristics, the correlation relationship between each branch is determined, and based on the correlation relationship and the power grid fault condition, effective security constraints are determined, including: inputting the node characteristics into a pre-trained security constraint identification model, and the security constraint identification model determines the correlation relationship between each branch based on the node characteristics, and based on the correlation relationship and the power grid fault condition, determines and outputs the effective security constraints; the effective security constraints include the security constraints satisfied by the non-fault branches in the unit commitment under the power grid single fault condition.

[0100] In some embodiments, the power grid single fault condition can be determined based on the N - 1 principle.

[0101] In some embodiments, the security constraint identification model can be a neural network model based on a heterogeneous graph.

[0102] Exemplarily, the correlation relationship can be a related influence relationship between each branch. For example, when a fault occurs in branch A, the corresponding influences on branches B, C, etc., such as whether a fault occurs and the current change situation, etc.

[0103] By using a pre-trained security constraint identification model to analyze node features, the correlation relationships between branches can be determined. Based on the correlation relationships and grid fault conditions, effective security constraints can be determined. This helps improve the determination efficiency of effective security constraints and can also obtain more accurate effective security constraints.

[0104] In one embodiment, obtaining the security-constrained unit commitment during the operation of the power grid includes: determining the pre-configured grid scheduling period for the power grid; and when the grid scheduling period is satisfied, performing the step of obtaining the security-constrained unit commitment during the operation of the power grid.

[0105] In some embodiments, the grid scheduling period can be determined and adjusted according to actual needs. Exemplarily, one natural day can be used as a grid scheduling period.

[0106] In some embodiments, when the grid scheduling period is satisfied, it means the period has expired and a new period has started. The security-constrained unit commitment of the power grid can be different in different grid scheduling periods. Since the security-constrained unit commitment can change with the grid scheduling period, when the grid scheduling period is satisfied, the security-constrained unit commitment during the operation of the power grid can be re-obtained, the node features can be re-determined, and thus the scheduling plan can be re-determined.

[0107] By re-obtaining the security-constrained unit commitment during the operation of the power grid according to the change of the grid scheduling period and then re-determining the scheduling plan, this helps ensure the accuracy of the scheduling plan.

[0108] In one embodiment, a method for determining an effective security constraint set of a power grid is provided, and a flow schematic of this method is given. In some embodiments, this method may include steps S301 to S306, specifically as follows:

[0109] Step S301, use a commercial solver to generate unit commitment (UC) plans for the historical n days, and combine the newly generated UC through perturbations on the basis of the historical UC, where the considered UC model (security-constrained unit commitment) only contains the security constraints of the base-state power flow.

[0110] Step S302, for each historical sample, obtain the AC power flows of each branch at each moment under all N - 1 faults. According to the branch power flow results, construct sample classification labels. The specific steps are as follows:

[0111] Step S302.1, for each historical sample, calculate the AC power flow of branch l at time t after branch c is disconnected for the AC power flow calculation. Then calculate the sample classification label x through Equation (1) c,l :

[0112]

[0113] In the formula, is the AC power flow limit of branch l.

[0114] Step S302.2: For all historical samples, represent their labels as a two-dimensional matrix. Specifically, for the a-th sample, as shown in Equation (2), X a :

[0115]

[0116] In the formula, a represents the sample number, and L is the number of branches in the system.

[0117] Step S303: Construct an undirected graph model. The vertices of the graph correspond to the nodes of the power grid, and the edges of the graph correspond to the transmission lines (branches) of the power grid. An heterogeneous graph model is adopted, and different input features (i.e., node features) are defined for nodes with different attributes. The specific steps are as follows:

[0118] Step S303.1: For a new energy power system with N nodes and L branches, construct the corresponding undirected graph model (i.e., the graph model). The vertex set of the graph is V = {v1, v2,..., v N}, corresponding to the N nodes in the power system; the edge set of the graph is E = {e ij |i ∈ [1, N], j ∈ [1, N], i ≠ j}, corresponding to the connection relationship between the i-th node and the j-th node in the power system. In this embodiment, N = 118;

[0119] Step S303.2: Use Equation (3) to obtain the adjacency matrix A of the undirected graph model G of the new energy power system, defined as A = {a ij}, where a ij :

[0120]

[0121] Step S303.3: In the heterogeneous graph neural network, as Figure 3 shown, according to the power generation and load conditions of the node connections, the nodes are divided into three categories: generator nodes, load nodes, and nodes with both power generation and load. Design their node features respectively. Among them, the input feature u g of the generator node is:

[0122] u g = [P max , P min (4)

[0123] In the formula, P max is the maximum output active power of the generator, and P minis the minimum active power output of the generator.

[0124] The input feature u of the load node d is:

[0125] u d = [P1, P2, …, P t , …, P T (5)

[0126] In the formula, P t is the active power load of the node within a certain time interval. In this embodiment, T = 24.

[0127] The input feature u of the node with both a generator and a load is:

[0128] u = u g ||u d (6)

[0129] Step S304: Design a power grid effective security constraint set identification model based on a heterogeneous graph neural network. Use a heterogeneous graph self-attention network to process different node input features (i.e., node features), and then input the processed features into a fully connected layer to predict key constraints (i.e., effective security constraints). The heterogeneous graph neural network model in step S304 includes a heterogeneous node feature encoding module, a graph attention convolution module, and a branch security constraint identification module.

[0130] Step S304.1, heterogeneous node feature encoding module: This module receives the original features of generator nodes, load nodes, and hybrid nodes in the power system and processes the features through a dedicated encoding network. Specifically, the input feature X p,q is encoded through the following formula:

[0131] X p,q = Dropout(GELU(LayerNorm(W p,q X p,q + b p,q ))) (7)

[0132] Among them, W p,q is a learnable weight matrix, and b p,q is a bias term. After processing, the node features are mapped to a unified high-dimensional feature space for subsequent processing.

[0133] Step S304.2, graph attention convolution module: This module uses a multi-head graph attention convolution layer (GAT) to perform message passing and feature aggregation on heterogeneous node features. The interaction weights between nodes are calculated by the following formula:

[0134]

[0135] Among them, Q is the query matrix, K is the key matrix, and the attention weight α ij By applying the softmax normalization to e ij to obtain:

[0136] α ij = softmax(e ij ) (9)

[0137] Finally, the new feature of each node is represented as:

[0138]

[0139] where V is the value matrix. This process realizes the non-linear propagation of node features and effectively enhances the correlation between nodes.

[0140] Step S304.3, branch influence prediction module: After graph convolution processing, the node features generate an influence prediction matrix with a specific dimension through a prediction network. The specific processing process is as follows:

[0141] X out = GELU(BatchNorm(h)) + X (11)

[0142] where h is the node feature after aggregation, and X is the original feature. This module generates an influence prediction matrix by integrating the features of the nodes at both ends of the branch, and then predicts the mutual relationship of each branch in the power grid topology.

[0143] Exemplarily, after inputting the node features, a two-dimensional matrix label of 0-1 is predicted. Each element in the two-dimensional matrix corresponds to a constraint. For example, the element in the 2nd row and 3rd column represents whether the power flow of the 3rd branch exceeds the corresponding threshold after the 2nd branch fails.

[0144] Exemplarily, the prediction matrix represents whether the remaining branches will exceed the corresponding threshold due to the disconnection of a certain branch. If it exceeds, it is 1, invalid; if not, it is 0, valid.

[0145] Step S305, training the designed model, and through end-to-end training, enabling the model to identify the effective safety constraint set of the power system grid. The training of the heterogeneous graph neural network in Step S305 includes data partitioning, data preprocessing, loss function design, and model optimization steps;

[0146] Step S305.1, data partitioning: Divide the power system operation dataset according to the ratio of 80% for training and 20% for validation, providing a data basis for model training and performance evaluation.

[0147] Step S305.2, Data Preprocessing: Normalize the input features to improve the training effect of the model, including standardization of generator node features, load node features, and hybrid node features. This process can be expressed as:

[0148] pred ∈ [∈, 1 - ∈] (12)

[0149] where ∈ = 10 -7 , which is used to prevent numerical instability in logarithmic operations.

[0150] Step S305.3, Loss Function Design: Adopt a custom loss function, comprehensively consider the weights of positive and negative samples, cross-entropy loss, and introduce a zero prediction penalty term to effectively handle the data imbalance problem. This loss calculation mechanism can be described as follows:

[0151] First, calculate the number of positive and negative samples:

[0152]

[0153] where N is the total number of samples, pos_samples is the number of positive samples in the dataset, calculated based on the target values, and neg_samples is the number of negative samples in the dataset, obtained by subtracting the number of positive samples from the total number of samples.

[0154] Second, calculate the loss:

[0155] final loss = total loss + zero prediction penalty (14)

[0156] where zero prediction penalty is the penalty term, activated when all prediction values are less than 0.1 to prevent the model from making zero predictions, and total loss is the total loss calculated based on the losses and weights of positive and negative samples.

[0157] Step S305.4, Model Optimization: Use the Adam optimizer to update the model parameters, thereby dynamically adjusting the learning process. Apply the gradient clipping technique to limit the norm of the gradient to prevent gradient explosion. The parameter update formula is:

[0158] θ t = θ t-1 - η t g t (15)

[0159] where θ t is the parameter value at time t, η t is the learning rate, and g t is the current gradient.

[0160] Step S306: Validate the effectiveness of the constructed effective constraint identification strategy and analyze the results, with a focus on analyzing the computational performance of SCUC for identifying the effective security constraint set of the power grid based on the heterogeneous graph neural network. A possible implementation is provided below:

[0161] To validate the performance of the power grid effective security constraint set identification model based on the heterogeneous graph neural network, as Figure 4 shown, an improved IEEE 118 - node system is selected. In the figure, G represents the nodes where generators are located, and the numbers represent the node numbers. Each node (or bus) in the system has a unique number, which is used to identify different positions in the power network.

[0162] New load scenarios are constructed by applying normal - distribution perturbations to the historical daily load curve, and a deep comparative analysis is conducted on the key constraint sets predicted by the model and the actual key constraint sets. In this application, the compression rate is defined as: the ratio of the number of constraints successfully screened by the model in the total number of original constraints to the total number of original constraints, that is, compression rate=(total number of original constraints - number of constraints predicted by the model) / total number of original constraints×100%. The focus is on examining the compression ability of the model and the coverage degree (recall rate) of the actual key constraints, aiming to comprehensively evaluate the effectiveness and reliability of the proposed method in the intelligent identification of power system security constraints.

[0163] After the above - mentioned model is verified, a trained security constraint identification model can be obtained. When in use, the node features can be input into the trained security constraint identification model, and the effective security constraints are output by the security constraint identification model. According to the effective constraints, the target security - constrained unit commitment is determined, and then the scheduling plan is determined.

[0164] In Table 1, the implementation results based on the aforementioned method in the improved IEEE 118 - node system are given. It can be seen that the above - mentioned method can accurately determine the actual key constraints (i.e., effective security constraints), reduce the computational amount required to determine the power grid scheduling plan, and can more quickly and accurately determine the power grid scheduling plan.

[0165] Table 1

[0166] Operating scenario Consider the total number of constraints Actual critical constraints Model prediction constraints Compression ratio (%) Recall rate (%) 1 34596 46 193 99.44 100 2 34596 64 195 99.44 100 3 34596 27 196 99.43 100 4 34596 58 198 99.43 100 5 34596 52 214 99.38 100 6 34596 49 197 99.43 100 7 34596 58 220 99.36 100 8 34596 66 211 99.39 100 9 34596 61 229 99.34 100 10 34596 57 206 99.40 100

[0167] In some embodiments, such as Figure 5As shown, the operating data of the power system can be obtained, and the corresponding SCUC can be obtained therefrom. Based on the SCUC, a graph model is constructed to obtain node features. The node features are input into the trained model for analysis to output effective security constraints. These effective security constraints are added to the UC corresponding to the SCUC. Among them, the UC does not contain effective security constraints, that is, UC = SCUC - SC. The UC and these effective security constraints form a new SCUC. By solving this new SCUC optimization problem, a power grid scheduling plan is obtained. In some possible embodiments, after obtaining the power grid scheduling plan, it is identified whether the power flow constraints at each moment and each branch under all N-1 faults on the new day are effective constraints, and the AC power flow involving each branch of the power grid can be analyzed based on the N-1 principle to determine whether the new security constraints are violated.

[0168] In some embodiments, the UC and the SCUC can be updated on a natural day basis or cycle, that is, the UC and SCUC corresponding to the next day of the power grid can be different from the UC and SCUC corresponding to the current day. Therefore, when the current cycle expires and enters the next new cycle, the SCUC corresponding to the next new cycle can be obtained. Based on this SCUC, the above steps are repeated, such as constructing a graph model to obtain node features; inputting the node features into the trained model for analysis to output effective security constraints, etc., so as to re-determine the power grid scheduling plan corresponding to the next new cycle.

[0169] In some embodiments, the training of the above model can also be iterated on a natural day basis or cycle to improve the accuracy of the model.

[0170] In the above technical solution, by combining the SCUC and the structural topology of the power grid, node features are extracted, and based on the node features, an effective security constraint is identified by using a neural network model. According to the effective security constraint and the unit commitment, the power grid scheduling plan is determined, rather than determining the scheduling plan according to all security constraints and unit commitment. This can reduce the corresponding calculation amount and improve the efficiency of determining the scheduling plan.

[0171] It should be understood that although the steps in the flowcharts involved in the above-described embodiments are shown in sequence according to the indications of the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear indication in this article, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or steps or stages in other steps.

[0172] Based on the same inventive concept, an embodiment of the present application further provides a power grid scheduling scheme determination device for implementing the power grid scheduling scheme determination method described above. The solution provided by this device for solving problems is similar to the solution described in the above method. Therefore, the specific limitations in one or more embodiments of the power grid scheduling scheme determination device provided below can refer to the limitations on the power grid scheduling scheme determination method in the above text, and will not be repeated here.

[0173] In an exemplary embodiment, as Figure 6 shown, a power grid scheduling scheme determination device 600 is provided, including:

[0174] An acquisition module 601, configured to acquire a security-constrained unit combination during the operation of the power grid. The security-constrained unit combination includes the unit combination during the operation and the security constraint conditions for the unit combination. The unit combination includes each power grid node in the power grid and the branches formed between each power grid node;

[0175] A constraint determination module 602, configured to determine the node characteristics corresponding to each power grid node according to the security-constrained unit combination; determine the association relationship between each branch based on the node characteristics, and determine an effective security constraint based on the association relationship and the power grid fault conditions; the effective security constraint includes the security constraint satisfied by the non-fault branches in the unit combination under the power grid fault conditions;

[0176] A scheme determination module 603, configured to determine a target security-constrained unit combination during the operation of the power grid based on the effective security constraint and the unit combination; obtain a scheduling scheme for the power grid during the operation according to the target security-constrained unit combination.

[0177] In one embodiment, the constraint determination module 602 is further configured to determine the node features corresponding to each power grid node according to the security-constrained unit commitment, including: determining the topological structure of the power grid according to the security-constrained unit commitment; constructing a graph model corresponding to the power grid according to the topological structure; and determining the node features based on the power information corresponding to each power grid node in the graph model.

[0178] In one embodiment, the constraint determination module 602 is further configured to construct a graph model corresponding to the power grid based on the topological structure, including: determining the vertices of the graph model corresponding to the power grid based on each power grid node in the topological structure; determining the edges of the graph model corresponding to the power grid based on each branch in the topological structure; and constructing the graph model corresponding to the power grid according to the vertices and edges according to the topological structure.

[0179] In one embodiment, the constraint determination module 602 is further configured to determine the node features based on the power information corresponding to each power grid node in the graph model, including: obtaining the power information corresponding to each power grid node; determining the node type to which each power grid node in the graph model belongs according to the power information; the node types include generation nodes, load nodes, and generation-load hybrid nodes; and determining the node features according to the node type and the power information.

[0180] In one embodiment, the power grid fault condition includes a power grid single fault condition; the constraint determination module 602 is further configured to determine the association relationship between each branch based on the node features, and determine the effective security constraints based on the association relationship and the power grid fault condition, including: inputting the node features into a pre-trained security constraint identification model, and the security constraint identification model determines the association relationship between each branch based on the node features, and determines and outputs the effective security constraints based on the association relationship and the power grid fault condition; the effective security constraints include the security constraints satisfied by the non-fault branches in the unit commitment under the power grid single fault condition.

[0181] In one embodiment, the acquisition module 601 is further configured to acquire the security-constrained unit commitment during the operation of the power grid, including: determining the pre-configured power grid scheduling period for the power grid; and performing the step of acquiring the security-constrained unit commitment during the operation of the power grid when the power grid scheduling period is satisfied.

[0182] Each module in the above power grid scheduling scheme determination device can be implemented in whole or in part by software, hardware, and their combination. The above modules can be embedded in the processor in the computer device in the form of hardware or independent of the processor, or stored in the memory in the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0183] In an exemplary embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be asFigure 7 As shown in the figure. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O), and a communication interface. Among them, the processor, the memory, and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the data required for the power grid scheduling scheme determination method, such as unit commitment. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals through a network connection. When the computer program is executed by the processor, it implements a power grid scheduling scheme determination method.

[0184] Those skilled in the art can understand that Figure 7 the structure shown in the figure is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0185] In an exemplary embodiment, a computer device is provided, including a memory and a processor. A computer program is stored in the memory, and when the processor executes the computer program, the steps in the above method embodiments are implemented.

[0186] In an embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by the processor, the steps in the above method embodiments are implemented.

[0187] In an embodiment, a computer program product is provided, including a computer program. When the computer program is executed by the processor, the steps in the above method embodiments are implemented.

[0188] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in the present application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in the present application can be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, data processing logics based on quantum computing, artificial intelligence (AI) processors, etc., without limitation.

[0189] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, 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, it should be considered as within the scope recorded in the present application.

[0190] The above-described embodiments merely represent several implementation manners of the present application. The description thereof is relatively specific and detailed, but it should not be construed as a limitation on the patent scope of the present application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present application, several modifications and improvements can still be made, and these all fall within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the appended claims.

Claims

1. A method for determining a power grid dispatching scheme, characterized in that: The method comprises: Acquire a safety-constrained unit combination during operation of a power grid, wherein the safety-constrained unit combination includes the unit combination during operation and safety constraints for the unit combination, and the unit combination includes each grid node in the power grid and branches formed between each grid node; Determining node characteristics corresponding to each of the power grid nodes according to the safety constraint unit combination; Determine the association relationship between the branches based on the node characteristics, and determine the effective safety constraints based on the association relationship and the power grid fault condition; the effective safety constraints include the safety constraints satisfied by the non-fault branches in the unit combination under the power grid fault condition; Determining a target safety constraint unit combination of the power grid during the operation process based on the effective safety constraint and the unit combination; According to the target safety-constrained unit combination, a dispatching plan for the power grid during the operation process is obtained.

2. The method according to claim 1, characterized in that Determining the node characteristics corresponding to each of the grid nodes according to the safety constraint unit combination includes: Determining the topology of the power grid according to the safety constraint unit combination; According to the topological structure, construct a graph model corresponding to the power grid; The node characteristics are determined based on the power information corresponding to each of the power grid nodes in the graph model.

3. The method according to claim 2, characterized in that The step of constructing a graph model corresponding to the power grid based on the topological structure includes: Based on each of the power grid nodes in the topological structure, determining a vertex of a graph model corresponding to the power grid; Based on each of the branches in the topological structure, determining an edge of a graph model corresponding to the power grid; A graph model corresponding to the power grid is constructed according to the vertices and the edges in accordance with the topological structure.

4. The method according to claim 2, characterized in that: The determining the node characteristics based on the power information corresponding to each of the power grid nodes in the graph model includes: Obtaining power information corresponding to each of the power grid nodes; Determine the node type to which each of the power grid nodes in the graph model belongs according to the power information; the node types include power generation nodes, load nodes, and power generation and load mixed nodes; The node characteristics are determined according to the node type and the power information.

5. The method according to claim 1, characterized in that The power grid fault condition includes a power grid single fault condition; The determining of the association relationship between the branches based on the node characteristics, and determining the effective safety constraint based on the association relationship and the power grid fault condition, includes: Inputting the node features into a pre-trained safety constraint identification model, the safety constraint identification model determines the association relationship between the branches based on the node features, and determines and outputs effective safety constraints based on the association relationship and power grid fault conditions; The effective safety constraints include safety constraints satisfied by non-fault branches in the unit combination under a single fault condition of the power grid.

6. The method according to any one of claims 1 to 5, characterized in that: The obtaining of the safety constraint unit combination during the operation of the power grid includes: Determining a preconfigured grid dispatch cycle for the grid; In the case where the power grid dispatching cycle is met, the step of obtaining the safety constraint unit combination of the power grid during operation is performed.

7. A device for determining a power grid dispatching plan, characterized in that: The device comprises: an acquisition module, used for acquiring a safety-constrained unit combination during operation of a power grid, wherein the safety-constrained unit combination includes the unit combination during operation and safety constraints for the unit combination, and the unit combination includes each grid node in the power grid and branches formed between each grid node; A constraint determination module is used to determine the node characteristics corresponding to each of the grid nodes according to the safety constraint unit combination; determine the association relationship between the branches based on the node characteristics, and determine the effective safety constraints based on the association relationship and the grid fault condition; the effective safety constraints include the safety constraints satisfied by the non-fault branches in the unit combination under the grid fault condition; A scheme determination module is used to determine the target safety constraint unit combination of the power grid during the operation process based on the effective safety constraint and the unit combination; and obtain a scheduling scheme for the power grid during the operation process according to the target safety constraint unit combination.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images