Scheduling method and device of power system and electronic equipment

Abstract

The invention discloses a scheduling method and device of a power system and electronic equipment. The method relates to the technical field of artificial intelligence, and comprises the following steps: solving a mathematical model of a security constraint unit commitment problem of a power system to obtain an initial solution of the mathematical model, the mathematical model being constructed according to parameter information of a plurality of generator sets of the power system; the initial solution and the mathematical model are processed through a graph convolutional neural network, a feasible solution of the security constraint unit commitment problem is obtained, and the feasible solution comprises starting time information and shutdown time information of the generator units in the multiple generator units in a future period of time; the output power of the generator sets in the plurality of generator sets at the preset time point is obtained; and scheduling the power system according to the feasible solution. According to the method and the device, the technical problem of relatively low scheduling efficiency of a power system caused by low efficiency of solving a security constraint unit commitment problem by adopting a branch and bound algorithm in related technologies is solved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and more specifically, to a dispatching method, device and electronic equipment for a power system. Background Art

[0002] The Security-Constrained Unit Commitment (SCUC) problem is a basic optimization problem faced by power system operators. Its goal is to determine the optimal plan for generating units to meet the forecast demand within the planning scope while ensuring the safety and reliability of power system operation.

[0003] At present, the branch and bound algorithm is usually used in related technologies to solve the SCUC problem and find a feasible solution to dispatch the power system. The branch and bound algorithm is a tree search algorithm that needs to traverse each leaf node from the root node. When encountering an infeasible solution, it often directly discards the infeasible solution and prunes by updating the upper and lower bounds to reduce the search space. However, the algorithm needs to traverse all leaf nodes. As the scale of the SCUC problem increases, the search space will become larger, and there are disadvantages of low solution efficiency and long solution time, which leads to low dispatch efficiency of the power system.

[0004] Currently, no effective solution has been proposed to solve the problem that the efficiency of solving the safety-constrained unit combination problem using the branch-and-bound algorithm in the above-mentioned related technologies is low, resulting in relatively low dispatching efficiency of the power system. Summary of the invention

[0005] The embodiments of the present application provide a dispatching method, device and electronic device for an electric power system, so as to at least solve the technical problem that the efficiency of solving the safety constraint unit combination problem using the branch and bound algorithm in the related technology is low, resulting in relatively low dispatching efficiency of the electric power system.

[0006] According to one aspect of an embodiment of the present application, a scheduling method for an electric power system is provided, comprising: solving a mathematical model of a safety-constrained unit combination problem of the electric power system to obtain an initial solution of the mathematical model, wherein the mathematical model is constructed based on parameter information of multiple generator sets of the electric power system; processing the initial solution and the mathematical model through a graph convolutional neural network to obtain a feasible solution to the safety-constrained unit combination problem, wherein the feasible solution includes startup time information and shutdown time information of generator sets among the multiple generator sets within a future period of time, and output power of generator sets among the multiple generator sets at a preset time point; and scheduling the electric power system based on the feasible solution.

[0007] Furthermore, solving the mathematical model of the safety constraint unit combination problem of the power system to obtain the initial solution of the mathematical model includes: solving the linear relaxation of the mathematical model to obtain a relaxed solution; judging whether there are non-integer decision variables in the decision variables of the relaxed solution; if there are non-integer decision variables in the decision variables of the relaxed solution, rounding the non-integer decision variables to obtain the initial solution.

[0008] Furthermore, the initial solution and the mathematical model are processed by a graph convolutional neural network to obtain a feasible solution to the safety constrained unit combination problem, including: constructing a target bipartite graph based on the initial solution and the mathematical model; processing the target bipartite graph by the graph convolutional neural network to obtain a first eigenvector of a generator set among the multiple generator sets; determining a feasible solution corresponding to an integer variable of a generator set among the multiple generator sets based on the first eigenvector by the graph convolutional neural network, wherein the integer variable is the on / off time information of the multiple generator sets; determining a feasible solution to the safety constrained unit combination problem based on the feasible solution corresponding to the integer variable of the generator set among the multiple generator sets.

[0009] Furthermore, based on the initial solution and the mathematical model, constructing a target bipartite graph includes: constructing variable nodes based on the initial solution and the variables in the mathematical model; constructing constraint nodes based on the constraints in the mathematical model; constructing edges between the variable nodes and the constraint nodes based on the coefficients of the variables in the constraints; and constructing the target bipartite graph based on the variable nodes, the constraint nodes, and the edges between the variable nodes and the constraint nodes.

[0010] Furthermore, the target bipartite graph is processed by the graph convolutional neural network to obtain a first eigenvector of a generator group among the multiple generator groups, including: processing information contained in variable nodes and constraint nodes in the target bipartite graph by the graph convolutional neural network to obtain a second eigenvector corresponding to the variable nodes in the target bipartite graph; determining a correspondence between the variable nodes in the target bipartite graph and the generator groups among the multiple generator groups; and based on the correspondence, aggregating the second eigenvector to obtain the first eigenvector.

[0011] Furthermore, the information contained in the variable nodes and constraint nodes in the target bipartite graph is processed by the graph convolutional neural network to obtain a second feature vector corresponding to the variable nodes in the target bipartite graph, including: for the target variable node, determining multiple constraint nodes having an edge relationship with the target variable node, wherein the target variable node is any one of the variable nodes in the target bipartite graph; and performing feature extraction on the multiple constraint nodes, the target variable node, and the edges between the target variable node and the multiple constraint nodes by the graph convolutional neural network to obtain the second feature vector.

[0012] Furthermore, determining the feasible solutions corresponding to the integer variables of the generator groups among the multiple generator groups based on the first eigenvector through the graph convolutional neural network includes: processing the first eigenvector through the graph convolutional neural network to obtain a target probability value corresponding to the generator groups among the multiple generator groups, wherein the target probability value is used to characterize the probability that the initial solution of the generator groups among the multiple generator groups violates the constraints in the mathematical model; judging whether the target probability value is less than a preset threshold; if the target probability value is less than the preset threshold, determining the initial solution corresponding to the integer variables of the generator groups among the multiple generator groups as the feasible solution corresponding to the integer variables.

[0013] Furthermore, determining a feasible solution to the safety constrained unit combination problem based on feasible solutions corresponding to integer variables of the generator sets among the multiple generator sets includes: correcting the mathematical model based on the feasible solutions corresponding to integer variables of the generator sets among the multiple generator sets to obtain a corrected mathematical model; and solving the corrected mathematical model to obtain a feasible solution to the safety constrained unit combination problem.

[0014] Furthermore, after determining whether the target probability value is less than a preset threshold, the method also includes: if there is a target probability value corresponding to a target generator group that is greater than or equal to the preset threshold, determining the initial solution corresponding to the integer variable of the target generator group based on the initial solution; performing a neighborhood operation on the initial solution corresponding to the integer variable of the target generator group through a variable neighborhood search algorithm to obtain a feasible solution corresponding to the integer variable of the target generator group.

[0015] According to another aspect of an embodiment of the present application, a scheduling method for an electric power system is also provided, comprising: obtaining a mathematical model of a safety-constrained unit combination problem of the electric power system uploaded by a client, wherein the mathematical model is constructed based on parameter information of multiple generator sets of the electric power system; solving the mathematical model of the safety-constrained unit combination problem of the electric power system in a cloud server to obtain an initial solution of the mathematical model; processing the initial solution and the mathematical model through a graph convolutional neural network to obtain a feasible solution to the safety-constrained unit combination problem, wherein the feasible solution includes startup time information and shutdown time information of generator sets among the multiple generator sets within a future period of time, and output power of generator sets among the multiple generator sets at a preset time point; and feeding back the feasible solution to the client to schedule the electric power system based on the feasible solution.

[0016] According to another aspect of an embodiment of the present application, a dispatching device for an electric power system is also provided, comprising: a solving unit, used to solve a mathematical model of a safety-constrained unit combination problem of the electric power system to obtain an initial solution of the mathematical model, wherein the mathematical model is constructed based on parameter information of multiple generator sets of the electric power system; a processing unit, used to process the initial solution and the mathematical model through a graph convolutional neural network to obtain a feasible solution to the safety-constrained unit combination problem, wherein the feasible solution includes startup time information and shutdown time information of generator sets among the multiple generator sets within a future period of time, as well as the output power of generator sets among the multiple generator sets at a preset time point; a dispatching unit, used to dispatch the electric power system based on the feasible solution.

[0017] Furthermore, the solution unit includes: a solution subunit, used to solve the linear relaxation of the mathematical model to obtain a relaxed solution; a judgment subunit, used to judge whether there are non-integer decision variables in the decision variables of the relaxed solution; and a rounding subunit, used to round the non-integer decision variables to obtain the initial solution if there are non-integer decision variables in the decision variables of the relaxed solution.

[0018] Furthermore, the processing unit includes: a construction subunit, used to construct a target bipartite graph based on the initial solution and the mathematical model; a processing subunit, used to process the target bipartite graph through the graph convolutional neural network to obtain a first eigenvector of a generator group among the multiple generator groups; a first determination subunit, used to determine a feasible solution corresponding to an integer variable of a generator group among the multiple generator groups based on the first eigenvector through the graph convolutional neural network, wherein the integer variable is the on / off time information of the multiple generator groups; a second determination subunit, used to determine a feasible solution to the safety constrained unit combination problem based on the feasible solution corresponding to the integer variable of the generator group among the multiple generator groups.

[0019] Furthermore, the construction subunit includes: a first construction module, used to construct a variable node based on the initial solution and the variables in the mathematical model; a second construction module, used to construct a constraint node based on the constraints in the mathematical model; a third construction module, used to construct an edge between the variable node and the constraint node based on the coefficient of the variable in the constraint; and a fourth construction module, used to construct the target bipartite graph based on the variable node, the constraint node and the edge between the variable node and the constraint node.

[0020] Furthermore, the processing subunit includes: a first processing module, used to process the information contained in the variable nodes and constraint nodes in the target bipartite graph through the graph convolutional neural network to obtain a second eigenvector corresponding to the variable nodes in the target bipartite graph; a first determination module, used to determine the correspondence between the variable nodes in the target bipartite graph and the generator groups among the multiple generator groups; an aggregation module, used to aggregate the second eigenvector based on the correspondence to obtain the first eigenvector.

[0021] Furthermore, the processing module includes: a determination submodule, used to determine, for a target variable node, multiple constraint nodes having an edge relationship with the target variable node, wherein the target variable node is any one of the variable nodes in the target bipartite graph; an extraction submodule, used to perform feature extraction on the multiple constraint nodes, the target variable node, and the edges between the target variable node and the multiple constraint nodes through the graph convolutional neural network to obtain the second feature vector.

[0022] Furthermore, the first determination subunit includes: a second processing module, used to process the first eigenvector through the graph convolutional neural network to obtain a target probability value corresponding to a generator group among the multiple generator groups, wherein the target probability value is used to characterize the probability that an initial solution of a generator group among the multiple generator groups violates the constraints in the mathematical model; a judgment module, used to judge whether the target probability value is less than a preset threshold; and a second determination module, used to determine the initial solution corresponding to the integer variable of the generator group among the multiple generator groups as a feasible solution corresponding to the integer variable if the target probability value is less than the preset threshold.

[0023] Furthermore, the second determination subunit includes: a correction module, used to correct the mathematical model according to the feasible solutions corresponding to the integer variables of the generator sets among the multiple generator sets to obtain a corrected mathematical model; a solution module, used to solve the corrected mathematical model to obtain a feasible solution to the safety constrained unit combination problem.

[0024] Furthermore, the device also includes: a determination unit, which is used to determine the initial solution corresponding to the integer variables of the target generator group based on the initial solution after judging whether the target probability value is less than a preset threshold value, if there is a target probability value corresponding to the target generator group that is greater than or equal to the preset threshold value; an operation unit, which is used to perform neighborhood operations on the initial solution corresponding to the integer variables of the target generator group through a variable neighborhood search algorithm to obtain a feasible solution corresponding to the integer variables of the target generator group.

[0025] According to another aspect of an embodiment of the present invention, there is further provided an electronic device, comprising: a memory storing an executable program; and a processor for running the program, wherein when the program is run, any one of the above-mentioned methods for dispatching a power system is executed.

[0026] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is provided, wherein the storage medium stores a program, wherein when the program is executed, the device where the storage medium is located is controlled to execute any one of the above-mentioned methods for dispatching a power system.

[0027] According to another aspect of an embodiment of the present invention, a computer program product is provided, including a computer program or instructions, and when the computer program or instructions are executed by a processor, any one of the above-mentioned methods for dispatching a power system is implemented.

[0028] In an embodiment of the present application, the following steps are adopted: solving a mathematical model of the safety constraint unit combination problem of the power system to obtain an initial solution of the mathematical model, wherein the mathematical model is constructed based on parameter information of multiple generator sets of the power system; processing the initial solution and the mathematical model through a graph convolutional neural network to obtain a feasible solution to the safety constraint unit combination problem, wherein the feasible solution includes startup time information and shutdown time information of generator sets among multiple generator sets in a future period of time, and the output power of generator sets among multiple generator sets at a preset time point; scheduling the power system based on the feasible solution, solving the technical problem that the branch and bound algorithm used in the related art to solve the safety constraint unit combination problem is inefficient, resulting in a relatively low scheduling efficiency of the power system. In this scheme, the mathematical model is first solved to obtain an initial solution that may not be feasible, and then the graph convolutional neural network is used to make full use of the initial solution that may not be feasible to solve the feasible solution of the safety constraint unit combination problem, effectively reducing the time consumption for solving the solution, and achieving the purpose of making full use of the information of the infeasible solution, thereby achieving the technical effect of improving the efficiency of solving the safety constraint unit combination problem and improving the scheduling efficiency of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0030] Figure 1 is a hardware structure block diagram of a computer terminal provided according to Embodiment 1 of the present application;

[0031] Figure 2 The process of the power system dispatching method provided in the first embodiment of the present application is as follows Figure 1 ;

[0032] Figure 3 is a schematic diagram of a bipartite graph provided according to Embodiment 1 of the present application;

[0033] Figure 4 The process of the power system dispatching method provided in the first embodiment of the present application is as follows Figure 2 ;

[0034] Figure 5 is a flow chart of a dispatching method for a power system provided according to Embodiment 2 of the present application;

[0035] Figure 6 is a schematic diagram of a dispatching device for a power system provided according to Embodiment 3 of the present application;

[0036] Figure 7 It is a structural block diagram of an electronic device provided according to Embodiment 4 of the present application. DETAILED DESCRIPTION

[0037] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0038] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0039] First, some nouns or terms that appear in the description of the embodiments of the present application are subject to the following explanations:

[0040] Security-Constrained Unit Commitment (SCUC) problem: In order to meet the electricity demand of users in the power grid, a plan is formulated to stipulate when each generator set will be turned on and off in the future and the output power at each time point. The plan needs to meet a series of physical requirements and keep the total power generation cost small. This problem is usually modeled as a mixed 0-1 linear programming model.

[0041] Graph convolutional network (GCN) is a type of neural network used to process graph structured data. It uses the topological structure of the graph to extract the feature information of nodes through convolution operations.

[0042] Variable neighborhood search (VNS): VNS systematically changes the neighborhood structure in the search space to avoid falling into the local optimal solution, thereby finding the global optimal solution or an approximate global optimal solution. Its basic idea is to dynamically adjust the neighborhood range, perturb and improve the current solution through a variety of different neighborhood operations (such as insertion, exchange, inversion, etc.), and gradually explore better solutions.

[0043] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards in the relevant regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0044] Example 1

[0045] According to an embodiment of the present application, a dispatching method for an electric power system is also provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0046] The method embodiment provided in the first embodiment of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Figure 1 FIG. 1 shows a hardware structure block diagram of a computer terminal (or mobile device) for implementing a dispatching method for a power system. Figure 1 As shown, the computer terminal (or mobile device) 10 may include a processor set 102 (the processor set 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA, and the processor set 102 may include a processor set, Figure 1 102a, 102b, ..., 102n are used to illustrate), a memory 104 for storing data, and a transmission module 106 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB, UniversalSerial Bus) port (which may be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1More or fewer components as shown, or with Figure 1 Different configurations shown.

[0047] It should be noted that the one or more processors 102 and / or other data processing circuits described above may generally be referred to herein as "data processing circuits". The data processing circuits may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuit may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the computer terminal 10 (or mobile device). As described in the embodiments of the present application, the data processing circuit acts as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0048] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the dispatching method of the power system in the embodiment of the present application. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, realizing the dispatching method of the power system mentioned above. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the computer terminal 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0049] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the computer terminal 10. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0050] The display may be, for example, a touch screen liquid crystal display that enables a user to interact with a user interface of the computer terminal 10 (or mobile device).

[0051] The SCUC problem is a basic optimization problem faced by power system operators, whose goal is to determine the optimal plan for generating units to meet the forecast demand within the planning horizon while ensuring the safety and reliability of power system operation.

[0052] At present, the branch and bound algorithm is usually used in the relevant technology to solve the SCUC problem and find a feasible solution to dispatch the power system. The branch and bound algorithm ignores the integer constraints of the model, treats the corresponding variables as ordinary continuous variables, and obtains a linear programming (LP) model, which is called the linear programming relaxation (LPR) of the original model. Solving LP can obtain a solution, which is called the relaxed solution. The corresponding objective function is the lower bound of the original problem, that is, the objective function value of the original problem is greater than or equal to the current value. If the corresponding variable in the relaxed solution just meets the integer requirement, the relaxed solution is the better solution of the original problem, and the solution ends; otherwise, branching is performed, that is, a fractional variable is selected, and the value is fixed to 0 and 1 respectively, two LP sub-problems are constructed, and the two LP sub-problems are solved respectively. If the obtained solution meets the integer requirement, a feasible solution to the original problem is obtained, and the objective function value corresponding to the solution is recorded as the upper bound, that is, the objective function value of the original problem cannot be greater than the current value. If there are variables that still cannot meet the integer constraint requirements, the subproblem will continue to be branched and solved, and the upper and lower bounds will be continuously updated. When the upper and lower bounds are equal, it indicates that a better solution has been obtained.

[0053] It can be seen that the branch and bound algorithm is a tree search algorithm that needs to traverse each leaf node from the root node. When encountering an infeasible solution, it often directly discards the infeasible solution and prunes by updating the upper and lower bounds to reduce the search space. However, this algorithm needs to traverse all leaf nodes. As the scale of the SCUC problem increases, the search space will become larger, and there are disadvantages of low solution efficiency and long solution time, which leads to low dispatch efficiency of the power system.

[0054] Under the above operating environment, this application provides Figure 2 The dispatch method of the power system shown. Figure 2 The process of the power system dispatching method according to the first embodiment of the present application is as follows Figure 1 The scheduling method includes:

[0055] Step S201, solving a mathematical model of a safety-constrained unit commitment problem of a power system to obtain an initial solution of the mathematical model, wherein the mathematical model is constructed based on parameter information of a plurality of generator sets of the power system.

[0056] The goal of the mathematical model of the SCUC problem is usually to make the total power generation cost (including startup and shutdown costs, operating costs, etc.) smaller. The constraint level mainly includes power balance constraints, reserve constraints, up / down limit constraints, minimum downtime / startup time constraints, and network security constraints. Among them, the power balance constraint ensures that the total power generation of all generators in each time interval meets the system demand (i.e., load); the reserve constraint requires that the power grid has sufficient reserve capacity to handle unexpected fluctuations in demand or power generation; the up / down limit constraint reflects the ability of the generator to increase (up) or decrease (down) the output power in a single time period; the minimum downtime / startup time constraint means that once a generator is turned off (on), it must remain offline (online) for a certain minimum time before it is started (turned off) again; the network security constraint limits the current through the transmission line to prevent the line from overheating.

[0057] The mathematical model of the SCUC problem can be constructed based on the parameter data of multiple generator sets in the power system (such as the upper and lower limits of the generator power, etc.). For example, with the on / off state and the output power of each generator set in each time interval as decision variables, the SCUC problem can be abstracted into a mixed 0-1 linear programming model (i.e., the mathematical model mentioned above), expressed as follows:

[0058] min c T x

[0059] ste≤Ax≤f

[0060] l≤x≤u

[0061]

[0062] Among them, c T represents the power generation cost matrix, x represents the decision variable, st represents the constraint, A represents the coefficient matrix composed of the constraint conditions, e represents the lower bound of Ax, f represents the upper bound of Ax, l represents the lower bound of x, u represents the upper bound of x, x∈R n , A∈R m×n ,e,f∈R m ,l,u∈R n ,

[0063] For example, the mathematical model of the SCUC problem is obtained through the dispatching system of the power system.

[0064] For example, by solving the linear relaxation of the mathematical model of the SCUC problem, a relaxed solution is obtained, and the variables in the relaxed solution that are not integers (i.e., variables whose values ​​are fractions) are rounded to obtain an initial solution. It should be noted that the initial solution may not be feasible.

[0065] Step S202, the initial solution and the mathematical model are processed by a graph convolutional neural network to obtain a feasible solution to the safety constrained unit combination problem, wherein the feasible solution includes the startup time information and shutdown time information of the generator sets among the multiple generator sets in a future period of time, and the output power of the generator sets among the multiple generator sets at a preset time point.

[0066] Optionally, in order to make full use of the information of the current infeasible solution, the initial solution and the mathematical model are processed by a graph convolutional neural network to obtain a feasible solution to the safety constrained unit combination problem. For example, the initial solution and the mathematical model can be feature extracted by a graph convolutional neural network, and then the feasible solution to the safety constrained unit combination problem is determined based on the extracted feature information.

[0067] Step S203: dispatching the power system according to the feasible solution.

[0068] Optionally, the operation of the generator sets of the power system may be scheduled according to the startup time information and shutdown time information of the generator sets included in the feasible solution, as well as the output power of the generator sets at a preset time point.

[0069] To sum up, the mathematical model is first solved to obtain an initial solution that may not be feasible, and then the graph convolutional neural network is used to make full use of the initial solution that may not be feasible to solve the feasible solution of the safety constrained unit combination problem, which effectively reduces the time consumption of the solution and achieves the purpose of making full use of the information of the infeasible solution, thereby achieving the technical effect of improving the efficiency of solving the safety constrained unit combination problem and improving the dispatching efficiency of the power system.

[0070] In order to obtain an initial solution of the mathematical model, in the dispatching method of the power system provided in Example 1 of the present application, the mathematical model of the safety constraint unit combination problem of the power system is solved to obtain the initial solution of the mathematical model, including: solving the linear relaxation of the mathematical model to obtain a relaxed solution; judging whether there are non-integer decision variables in the decision variables of the relaxed solution; if there are non-integer decision variables in the decision variables of the relaxed solution, rounding the non-integer decision variables to obtain an initial solution.

[0071] Optionally, a relaxed solution can be obtained by solving the linear relaxation of the mathematical model of the SCUC problem, and then it is determined whether there are non-integer decision variables in the decision variables of the relaxed solution. If the decision variables in the relaxed solution just meet the integer requirements, the relaxed solution is the optimal solution to the original problem, and the solution is completed; if there are non-integer decision variables (i.e., variables with fractional values) in the decision variables of the relaxed solution, the non-integer decision variables are rounded to obtain an initial solution, which may not be feasible.

[0072] By judging whether there are non-integer decision variables in the decision variables of the relaxed solution, the initial solution of the mathematical model is accurately determined.

[0073] In order to make full use of the information of the current infeasible solution, in the dispatching method of the power system provided in the first embodiment of the present application, the initial solution and the mathematical model are processed by a graph convolutional neural network to obtain a feasible solution to the safety constrained unit combination problem, including: constructing a target bipartite graph based on the initial solution and the mathematical model; processing the target bipartite graph by a graph convolutional neural network to obtain a first eigenvector of a generator set among multiple generator sets; determining a feasible solution corresponding to an integer variable of a generator set among multiple generator sets based on the first eigenvector by a graph convolutional neural network, wherein the integer variable is the on / off time information of multiple generator sets; determining a feasible solution to the safety constrained unit combination problem based on the feasible solution corresponding to the integer variable of the generator set among the multiple generator sets.

[0074] Optionally, a target bipartite graph is constructed based on the initial solution and the mathematical model. For example, the nodes in the bipartite graph are constructed by extracting the variables and constraints in the mathematical model. For example, the variable nodes of the target bipartite graph include the variables in the objective function (c T x), the coefficients in the constraint, the variable type, the value in the relaxed circular solution, etc. The constraint node includes the average coefficient of the variable in the constraint, the right-hand side term, etc. The edge is characterized by the coefficient of the variable in the constraint. If a variable does not appear in a constraint, there is no edge connecting the two nodes.

[0075] After obtaining the target bipartite graph, the target bipartite graph is subjected to feature extraction and feature aggregation by using a graph convolutional neural network to obtain the first feature vector of the generator set in the multiple generator sets. For example, the variables contained in the variable node of the target bipartite graph are in the objective function (c T x), the coefficients in the constraint node, the variable type, the values ​​in the relaxed circular solution, and the average coefficients and right-hand terms of the variables in the constraint included in the constraint node are processed by feature extraction and feature aggregation to obtain the first eigenvector of the generator set in the multiple generator sets.

[0076] Then, the first eigenvector is used through the graph convolutional neural network to determine the feasible solutions corresponding to the integer variables of the generator sets in the multiple generator sets. It should be noted that the integer variables are the startup time information and shutdown time information of the multiple generator sets. Finally, the feasible solutions to the safety constrained unit combination problem are determined based on the feasible solutions corresponding to the integer variables of the generator sets in the multiple generator sets.

[0077] For example, the graph convolutional neural network can be used to predict the probability that the initial solution corresponding to the integer variable of a generator group in multiple generator groups violates the constraint based on the first eigenvector. If the probability is relatively low, the initial solution corresponding to the integer variable of the generator group can be directly determined as the feasible solution corresponding to the integer variable.

[0078] For example, after determining the feasible solutions corresponding to the integer variables, the feasible solutions corresponding to the integer variables can be brought into the above-mentioned mathematical model, and the model can be transformed into a linear programming model. By solving the changed model, the feasible solution to the safety constrained unit combination problem can be obtained.

[0079] By predicting the probability through graph convolutional neural network, the integer variables corresponding to some units in the infeasible solution can be fixed, which can effectively improve the efficiency of solving the mathematical model of the safety constrained unit combination problem with fixed integer variables.

[0080] In the dispatching method of the power system provided in the first embodiment of the present application, constructing a target bipartite graph based on the initial solution and the mathematical model includes: constructing variable nodes based on the variables in the initial solution and the mathematical model; constructing constraint nodes based on the constraints in the mathematical model; constructing edges between variable nodes and constraint nodes based on the coefficients of the variables in the constraints; constructing a target bipartite graph based on variable nodes, constraint nodes, and edges between variable nodes and constraint nodes.

[0081] Optionally, the following steps are used to construct the target bipartite graph: variable nodes are constructed based on the initial solution and the variables in the mathematical model, and then constraint nodes are constructed based on the constraints in the mathematical model, and edges between variable nodes and constraint nodes are constructed using the coefficients of the variables in the constraints, and finally the target bipartite graph is constructed based on the variable nodes, constraint nodes, and edges between variable nodes and constraint nodes.

[0082] For example, the mathematical model of the SCUC problem is represented as a bipartite graph: Among them, the constraint node Variable Node The coefficient of a variable in a constraint is represented by an edge (e ij ∈ε), which is used to represent the relationship between variable nodes and constraint nodes. The bipartite graph can be represented as Figure 3 As shown in the schematic diagram, Figure 3 v1, v2…v n For multiple variables, a1, a m Characterization constraints, e 11 v1+…+e 1n v n ≤b1 and e m1 v1+…+e mn v n ≤ bm is the constraint condition corresponding to the constraint node, e 11 , e 1n Characterize the coefficient of the variable in the constraint.

[0083] Through the bipartite graph, each node can learn the characteristics of its adjacent nodes, improving the comprehensiveness of the subsequent determination of the characteristics of the generator set.

[0084] In order to obtain the first eigenvector, in the dispatching method of the power system provided in the first embodiment of the present application, the target bipartite graph is processed by a graph convolutional neural network to obtain the first eigenvector of a generator group among multiple generator groups, including: processing the information contained in the variable nodes and constraint nodes in the target bipartite graph by a graph convolutional neural network to obtain a second eigenvector corresponding to the variable nodes in the target bipartite graph; determining the correspondence between the variable nodes in the target bipartite graph and the generator groups among the multiple generator groups; and aggregating the second eigenvector based on the correspondence to obtain the first eigenvector.

[0085] Optionally, the graph convolutional neural network processes the information contained in the variable nodes and constraint nodes in the target bipartite graph to obtain a second eigenvector corresponding to the variable node in the target bipartite graph. For example, the information contained in the constraint node is passed to the variable node through message passing to obtain the second eigenvector corresponding to the variable node.

[0086] Then, the association between the variable and the corresponding generator is identified, and the corresponding relationship between the variable node and the generator group in the multiple generator groups is determined. Finally, according to the corresponding relationship, the second eigenvectors of the multiple variable nodes corresponding to the generator group in the multiple generator groups are aggregated to obtain the first eigenvector corresponding to the generator group in the multiple generator groups.

[0087] For example, a generator node can be represented as The embedding vector of the generator set (i.e. the first eigenvector mentioned above) is represented as w h , the aggregation operation can be expressed as follows:

[0088]

[0089] Where η(·) is the value that will be connected to the generator node w h Functions for the aggregation of associated integer variables.

[0090] It should be noted that aggregating the second eigenvectors of multiple variable nodes corresponding to the generator groups in the multiple generator groups to obtain the first eigenvector corresponding to the generator groups in the multiple generator groups may be aggregating the second eigenvectors of multiple integer variable nodes corresponding to the generator groups in the multiple generator groups to obtain the first eigenvector corresponding to the generator groups in the multiple generator groups.

[0091] The similarity between the relaxed circular solution and the optimal solution can be accurately evaluated by using the first eigenvector corresponding to the generator set in the plurality of generator sets, thereby improving the accuracy of determining the feasible solution corresponding to the integer variable.

[0092] In order to improve the accuracy of determining the second eigenvector, in the dispatching method of the power system provided in Example 1 of the present application, the information contained in the variable nodes and constraint nodes in the target bipartite graph is processed by a graph convolutional neural network to obtain the second eigenvector corresponding to the variable nodes in the target bipartite graph, including: for the target variable node, determining multiple constraint nodes having an edge relationship with the target variable node, wherein the target variable node is any one of the variable nodes in the target bipartite graph; performing feature extraction on multiple constraint nodes, the target variable node, and the edges between the target variable node and the multiple constraint nodes by a graph convolutional neural network to obtain the second eigenvector.

[0093] Optionally, for any one of the variable nodes in the target bipartite graph (i.e., the target variable node mentioned above), multiple constraint nodes having an edge relationship with the target variable node are determined according to the target bipartite graph, and then features of the multiple constraint nodes, the target variable node, and the edges between the target variable node and the multiple constraint nodes are extracted through a graph convolutional neural network to obtain a second feature vector.

[0094] For example, message passing between variable nodes and constraint nodes can be represented as follows:

[0095]

[0096] Among them, c i and v j Represents the constraint node c i The embedding vector and variable node v j The embedding vector (i.e., the second eigenvector mentioned above, [;] represents the aggregation operation, f1, f2, g1, and g2 represent multi-layer perceptrons (MLPs) with hidden layers and activation functions.

[0097] After representing the SCUC problem model as a bipartite graph, the message passing method is used to enable each node to learn the characteristics of its adjacent nodes, and aggregate all integer variable features to their respective generator nodes. For each generator, the probability that its corresponding integer variable violates the constraint is predicted, thereby improving the accuracy of determining the feasible solution corresponding to the integer variable.

[0098] In order to further improve the accuracy of determining feasible solutions corresponding to integer variables, in the dispatching method of the power system provided in Example 1 of the present application, determining the feasible solutions corresponding to the integer variables of a generator group among multiple generator groups based on the first eigenvector through a graph convolutional neural network includes: processing the first eigenvector through a graph convolutional neural network to obtain a target probability value corresponding to the generator group among the multiple generator groups, wherein the target probability value is used to characterize the probability that the initial solution of the generator group among the multiple generator groups violates the constraints in the mathematical model; judging whether the target probability value is less than a preset threshold; if the target probability value is less than the preset threshold, determining the initial solution corresponding to the integer variable of the generator group among the multiple generator groups as the feasible solution corresponding to the integer variable.

[0099] In an optional embodiment, after determining whether the target probability value is less than a preset threshold, the method further includes: if there is a target probability value corresponding to the target generator group that is greater than or equal to the preset threshold, then determining the initial solution corresponding to the integer variable of the target generator group based on the initial solution; performing a neighborhood operation on the initial solution corresponding to the integer variable of the target generator group through a variable neighborhood search algorithm to obtain a feasible solution corresponding to the integer variable of the target generator group.

[0100] Optionally, the following steps are used to determine the feasible solution corresponding to the integer variable: the first feature information is processed by a graph convolutional neural network to obtain a target probability value corresponding to a generator group in a plurality of generator groups. It should be noted that the target probability value is used to characterize the probability that the initial solution of a generator group in a plurality of generator groups violates the constraints in the mathematical model. After obtaining the target probability value, determine whether the target probability value is less than a preset threshold value (for example, about 0.2 to 0.3). If the target probability value is less than the preset threshold value, the initial solution of the integer variable corresponding to the generator group is consistent with the optimal solution, that is, the probability that the initial solution of the generator group violates the constraints in the mathematical model is very low, and the initial solution corresponding to the integer variable of the generator group in the plurality of generator groups can be directly determined as the feasible solution corresponding to the integer variable.

[0101] For example, after the aggregation operation obtains the embedding vector of the generator set (i.e., the first eigenvector mentioned above), the multi-layer perceptron of the graph convolutional neural network is used to process these embedding vectors, and finally the softmax function is used to map the output value to probability:

[0102]

[0103] Here, f3 represents a multi-layer perceptron (MLPs) with hidden layers and activation functions.

[0104] Optionally, if there is a target probability value corresponding to a target generator group that is greater than or equal to a preset threshold, it indicates that the initial solution of the integer variable of the target generator group has a high probability of violating the constraints in the mathematical model and cannot be used as a feasible solution. At this time, the initial solution corresponding to the integer variable of the target generator group can be subjected to a neighborhood operation through a variable neighborhood search algorithm to obtain a feasible solution corresponding to the integer variable of the target generator group.

[0105] For example, the on / off state vector of the target generator set (i.e., the initial solution corresponding to the integer variable of the target generator set) is used as the input of the VNS algorithm, and then a variety of neighborhood operations are set, such as single variable flipping (randomly selecting an integer variable for value flipping), time period exchange (for a certain generator, randomly), two-machine exchange, etc. A maximum number of attempts is set for each neighborhood operation. If the maximum number of attempts is reached using the current neighborhood operation and the partial solution obtained is still not feasible, then jump to the next neighborhood operation and continue to try until the partial solution obtained is feasible or the maximum number of iterations is reached. It should be noted that the condition for judging whether a partial solution is feasible is whether the partial solution satisfies the constraints of the original SCUC problem.

[0106] Finally, by combining the partial feasible solutions obtained by the VNS algorithm with the feasible solutions of the integer variables of the fixed generator, we can get a feasible solution of the original SCUC problem with an integer variable.

[0107] Through the graph convolutional neural network and variable neighborhood search algorithm, the information of the relaxed circular solution is fully utilized to obtain a feasible solution for the complete integer variables. Compared with the prior art where branch solving is required when variables do not meet the integer requirements, it can effectively improve the efficiency of subsequent determination of the feasible solution to the safety constrained unit combination problem.

[0108] In order to improve the efficiency of determining feasible solutions to the safety constrained unit combination problem, in the dispatching method of the power system provided in Example 1 of the present application, determining the feasible solution to the safety constrained unit combination problem based on the feasible solutions corresponding to the integer variables of the generator sets among the multiple generator sets includes: correcting the mathematical model based on the feasible solutions corresponding to the integer variables of the generator sets among the multiple generator sets to obtain a corrected mathematical model; solving the corrected mathematical model to obtain a feasible solution to the safety constrained unit combination problem.

[0109] Optionally, after obtaining feasible solutions corresponding to the integer variables of all generator sets, the feasible solutions corresponding to the integer variables of the generator sets among the multiple generator sets are brought into the mathematical model, and the mathematical model is converted into a linear programming model, that is, the above-mentioned revised mathematical model is obtained. Finally, by solving the revised mathematical model, a feasible solution to the safety constrained unit combination problem is obtained.

[0110] For example, according to feasible solutions corresponding to integer variables of multiple generator sets, the upper and lower bounds of the integer variables in the mathematical model are modified to fix their values, thereby obtaining the above-mentioned modified mathematical model.

[0111] GCN aggregates the integer variable features to the generator node to predict the probability of constraint violation, and sets a probability threshold. The integer variables corresponding to the generators that are smaller than the threshold are fixed, and the integer variables corresponding to the generators that are larger than the threshold are solved using the VNS algorithm to obtain partial feasible solutions, and finally obtain a complete initial integer feasible solution. Compared with the traditional branch and bound method for solving the SCUC problem, it can fully consider and utilize the information of the relaxed circular solution, thereby achieving the effect of improving the efficiency of solving the SCUC problem.

[0112] In an optional embodiment, the Figure 4 The flowchart shown implements the solution to the SCUC problem, specifically including: determining the mathematical model of the SCUC problem, obtaining a relaxed solution by solving the linear relaxation (LP) of the mathematical model of the SCUC problem, and rounding the variables with fractional values ​​to obtain a relaxed circular solution, which is likely to be infeasible in the original SCUC model. The mathematical model of the SCUC problem is represented as a bipartite graph, and the relaxed circular solution is input into the GCN (i.e., graph convolutional neural network), and the probability of the relaxed circular solution of the integer variable corresponding to the generator set violating the constraint is obtained through the GCN. If the probability is greater than or equal to the threshold, the relaxed rounding of the integer variables corresponding to the generator set is input into the VNS algorithm, and the integer variables corresponding to the generator set are iteratively solved by the VNS algorithm to obtain a feasible solution. For generator sets with probabilities less than the threshold, the integer variables corresponding to these generator sets are directly fixed, and the partial feasible solutions obtained by the VNS algorithm are combined with the fixed integer variables of the generator to obtain an initial integer feasible solution to the original SCUC problem. The solution is brought into the model, and the model becomes a linear programming model. Solving the model can obtain a complete feasible solution.

[0113] In the dispatching method of the power system provided in the first embodiment of the present application, the mathematical model of the safety constraint unit combination problem of the power system is solved to obtain the initial solution of the mathematical model, wherein the mathematical model is constructed based on the parameter information of multiple generator sets of the power system; the initial solution and the mathematical model are processed by a graph convolutional neural network to obtain a feasible solution to the safety constraint unit combination problem, wherein the feasible solution includes the start-up time information and shutdown time information of the generator sets among the multiple generator sets in a future period of time, and the output power of the generator sets among the multiple generator sets at a preset time point; according to the feasible solution, the power system is dispatched, which solves the technical problem that the branch and bound algorithm used in the related technology to solve the safety constraint unit combination problem is inefficient, resulting in relatively low dispatching efficiency of the power system. In this scheme, the mathematical model is first solved to obtain an initial solution that may not be feasible, and then the graph convolutional neural network is used to make full use of the initial solution that may not be feasible to solve the feasible solution of the safety constraint unit combination problem, which effectively reduces the time consumption for solution and achieves the purpose of making full use of the information of the infeasible solution, thereby achieving the technical effect of improving the efficiency of solving the safety constraint unit combination problem and improving the dispatching efficiency of the power system.

[0114] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0115] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods of each embodiment of the present application.

[0116] Example 2

[0117] According to an embodiment of the present application, a scheduling method for a power system is also provided, such as Figure 5 As shown, the method includes:

[0118] Step S501, obtaining a mathematical model of a safety constraint unit commitment problem of a power system uploaded by a client, wherein the mathematical model is constructed based on parameter information of multiple generator sets of the power system;

[0119] Step S502, solving a mathematical model of the safety-constrained unit combination problem of the power system in the cloud server to obtain an initial solution of the mathematical model; processing the initial solution and the mathematical model through a graph convolutional neural network to obtain a feasible solution to the safety-constrained unit combination problem, wherein the feasible solution includes startup time information and shutdown time information of a generator set among multiple generator sets within a period of time in the future, and output power of a generator set among multiple generator sets at a preset time point;

[0120] Step S503, feeding back the feasible solution to the client, so as to dispatch the power system according to the feasible solution.

[0121] Through the above scheme, the mathematical model is first solved to obtain an initial solution that may not be feasible, and then the graph convolutional neural network is used to make full use of the initial solution that may not be feasible to solve the feasible solution of the safety constrained unit combination problem, which effectively reduces the time consumption of the solution and achieves the purpose of making full use of the information of the infeasible solution, thereby achieving the technical effect of improving the solution efficiency of the safety constrained unit combination problem and improving the dispatch efficiency of the power system.

[0122] In the cloud server, the specific method for dispatching the power system is the same as the method in Example 1, and will not be repeated here.

[0123] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0124] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods of each embodiment of the present application.

[0125] Example 3

[0126] According to an embodiment of the present application, a dispatching device for a power system for implementing the above-mentioned dispatching method for a power system is also provided. Figure 6 As shown, the device includes: a solving unit 601, a processing unit 602 and a scheduling unit 603.

[0127] A solving unit 601 is used to solve a mathematical model of the safety constraint unit commitment problem of the power system to obtain an initial solution of the mathematical model, wherein the mathematical model is constructed based on parameter information of multiple generator sets of the power system;

[0128] The processing unit 602 is used to process the initial solution and the mathematical model through a graph convolutional neural network to obtain a feasible solution to the safety constrained unit commitment problem, wherein the feasible solution includes startup time information and shutdown time information of a generator set among the multiple generator sets within a future period of time, and output power of a generator set among the multiple generator sets at a preset time point;

[0129] The dispatching unit 603 is used to dispatch the power system according to the feasible solution.

[0130] In the dispatching device of the electric power system provided in the third embodiment of the present application, the mathematical model of the safety constraint unit combination problem of the electric power system is solved by the solving unit 601 to obtain an initial solution of the mathematical model, wherein the mathematical model is constructed based on the parameter information of multiple generator sets of the electric power system; the processing unit 602 processes the initial solution and the mathematical model through a graph convolutional neural network to obtain a feasible solution to the safety constraint unit combination problem, wherein the feasible solution includes the start-up time information and the shutdown time information of the generator sets among the multiple generator sets within a period of time in the future, and the output power of the generator sets among the multiple generator sets at a preset time point; the dispatching unit Element 603 dispatches the power system based on the feasible solution, which solves the technical problem that the branch and bound algorithm used in the related technology to solve the safety constrained unit combination problem is inefficient, resulting in relatively low dispatch efficiency of the power system. In this scheme, the mathematical model is first solved to obtain an initial solution that may not be feasible, and then the graph convolutional neural network is used to make full use of the initial solution that may not be feasible to solve the feasible solution of the safety constrained unit combination problem, which effectively reduces the time consumption of the solution and achieves the purpose of making full use of the information of the infeasible solution, thereby achieving the technical effect of improving the efficiency of solving the safety constrained unit combination problem and improving the dispatch efficiency of the power system.

[0131] Optionally, in the dispatching device of the power system provided in Example 3 of the present application, the solving unit 601 includes: a solving subunit, used to solve the linear relaxation of the mathematical model to obtain a relaxed solution; a judging subunit, used to judge whether there are non-integer decision variables in the decision variables of the relaxed solution; and a rounding subunit, used to round the non-integer decision variables to obtain an initial solution if there are non-integer decision variables in the decision variables of the relaxed solution.

[0132] Optionally, in the dispatching device of the power system provided in Example 3 of the present application, the processing unit 602 includes: a construction subunit, used to construct a target bipartite graph based on an initial solution and a mathematical model; a processing subunit, used to process the target bipartite graph through a graph convolutional neural network to obtain a first eigenvector of a generator group among multiple generator groups; a first determination subunit, used to determine a feasible solution corresponding to an integer variable of a generator group among multiple generator groups based on the first eigenvector through a graph convolutional neural network, wherein the integer variables are on and off time information of multiple generator groups; a second determination subunit, used to determine a feasible solution to the safety constraint unit combination problem based on the feasible solution corresponding to the integer variables of the generator groups among multiple generator groups.

[0133] Optionally, in the dispatching device of the power system provided in Example 3 of the present application, the construction subunit includes: a first construction module, used to construct variable nodes based on the initial solution and the variables in the mathematical model; a second construction module, used to construct constraint nodes based on the constraints in the mathematical model; a third construction module, used to construct edges between variable nodes and constraint nodes based on the coefficients of the variables in the constraints; and a fourth construction module, used to construct a target bipartite graph based on variable nodes, constraint nodes, and edges between variable nodes and constraint nodes.

[0134] Optionally, in the dispatching device of the power system provided in Example 3 of the present application, the processing subunit includes: a first processing module, used to process the information contained in the variable nodes and constraint nodes in the target bipartite graph through a graph convolutional neural network to obtain a second eigenvector corresponding to the variable nodes in the target bipartite graph; a first determination module, used to determine the correspondence between the variable nodes in the target bipartite graph and the generator groups among multiple generator groups; and an aggregation module, used to aggregate the second eigenvector based on the correspondence to obtain the first eigenvector.

[0135] Optionally, in the dispatching device of the power system provided in Example 3 of the present application, the processing module includes: a determination submodule, used to determine, for a target variable node, multiple constraint nodes having an edge relationship with the target variable node, wherein the target variable node is any one of the variable nodes in the target bipartite graph; an extraction submodule, used to perform feature extraction on multiple constraint nodes, target variable nodes, and edges between the target variable node and the multiple constraint nodes through a graph convolutional neural network to obtain a second feature vector.

[0136] Optionally, in the dispatching device of the power system provided in Example 3 of the present application, the first determination subunit includes: a second processing module, used to process the first eigenvector through a graph convolutional neural network to obtain a target probability value corresponding to a generator group among the multiple generator groups, wherein the target probability value is used to characterize the probability that the initial solution of the generator group among the multiple generator groups violates the constraints in the mathematical model; a judgment module, used to judge whether the target probability value is less than a preset threshold; and a second determination module, used to determine the initial solution corresponding to the integer variable of the generator group among the multiple generator groups as a feasible solution corresponding to the integer variable if the target probability value is less than the preset threshold.

[0137] Optionally, in the dispatching device of the power system provided in Example 3 of the present application, the second determination subunit includes: a correction module, used to correct the mathematical model according to the feasible solutions corresponding to the integer variables of the generator groups among the multiple generator groups to obtain a corrected mathematical model; and a solution module, used to solve the corrected mathematical model to obtain a feasible solution to the safety constrained unit combination problem.

[0138] Optionally, in the dispatching device of the power system provided in Example 3 of the present application, the device also includes: a determination unit, which is used to determine, based on the initial solution, an initial solution corresponding to the integer variables of the target generator group after judging whether the target probability value is less than a preset threshold value, if there is a target probability value corresponding to the target generator group that is greater than or equal to the preset threshold value; and an operation unit, which is used to perform neighborhood operations on the initial solution corresponding to the integer variables of the target generator group through a variable neighborhood search algorithm to obtain a feasible solution corresponding to the integer variables of the target generator group.

[0139] It should be noted that the above-mentioned solving unit 601, processing unit 602 and scheduling unit 603 correspond to steps S201 to S203 in the first embodiment, and the three units and the corresponding steps implement the same examples and application scenarios, but are not limited to the contents disclosed in the first embodiment. It should be noted that the above-mentioned modules as part of the device can be run in the computer terminal 10 provided in the first embodiment.

[0140] It should be noted that the preferred implementation scheme involved in the above embodiments of the present application is the same as the scheme provided in Example 1, as well as the application scenario and implementation process, but is not limited to the scheme provided in Example 1.

[0141] Example 4

[0142] The embodiment of the present application may provide an electronic device, which may be any electronic device in an electronic device terminal group. Optionally, in this embodiment, the electronic device may also be replaced by a terminal device such as a mobile terminal.

[0143] Optionally, in this embodiment, the electronic device may be located in at least one network device among a plurality of network devices of a computer network.

[0144] In this embodiment, the above-mentioned electronic device can execute the program code of the following steps in the dispatching method of the power system: solving the mathematical model of the safety constraint unit combination problem of the power system to obtain the initial solution of the mathematical model, wherein the mathematical model is constructed based on the parameter information of multiple generator sets in the power system; processing the initial solution and the mathematical model through a graph convolutional neural network to obtain a feasible solution to the safety constraint unit combination problem, wherein the feasible solution includes the start-up time information and shutdown time information of the generator sets among the multiple generator sets in a future period of time, and the output power of the generator sets among the multiple generator sets at a preset time point; and dispatching the power system based on the feasible solution.

[0145] The above-mentioned electronic device can execute the program code of the following steps in the dispatching method of the power system: solving the mathematical model of the safety constraint unit combination problem of the power system to obtain the initial solution of the mathematical model includes: solving the linear relaxation of the mathematical model to obtain a relaxed solution; judging whether there are non-integer decision variables in the decision variables of the relaxed solution; if there are non-integer decision variables in the decision variables of the relaxed solution, rounding the non-integer decision variables to obtain the initial solution.

[0146] The above-mentioned electronic device can execute the program code of the following steps in the dispatching method of the power system: the initial solution and the mathematical model are processed by a graph convolutional neural network to obtain a feasible solution to the safety constrained unit combination problem, including: constructing a target bipartite graph based on the initial solution and the mathematical model; processing the target bipartite graph by a graph convolutional neural network to obtain a first eigenvector of a generator set among multiple generator sets; determining a feasible solution corresponding to an integer variable of a generator set among multiple generator sets based on the first eigenvector by a graph convolutional neural network, wherein the integer variable is the on / off time information of multiple generator sets; determining a feasible solution to the safety constrained unit combination problem based on the feasible solution corresponding to the integer variable of the generator set among multiple generator sets.

[0147] The above-mentioned electronic device can execute the program code of the following steps in the dispatching method of the power system: constructing a target bipartite graph based on the initial solution and the mathematical model, including: constructing variable nodes based on the variables in the initial solution and the mathematical model; constructing constraint nodes based on the constraints in the mathematical model; constructing edges between variable nodes and constraint nodes based on the coefficients of the variables in the constraints; constructing a target bipartite graph based on variable nodes, constraint nodes and edges between variable nodes and constraint nodes.

[0148] The above-mentioned electronic device can execute the program code of the following steps in the dispatching method of the power system: processing the target bipartite graph through a graph convolutional neural network to obtain the first eigenvector of a generator group among multiple generator groups, including: processing the information contained in the variable nodes and constraint nodes in the target bipartite graph through a graph convolutional neural network to obtain the second eigenvector corresponding to the variable nodes in the target bipartite graph; determining the correspondence between the variable nodes in the target bipartite graph and the generator groups among the multiple generator groups; and aggregating the second eigenvector based on the correspondence to obtain the first eigenvector.

[0149] The above-mentioned electronic device can execute the program code of the following steps in the dispatching method of the power system: processing the information contained in the variable nodes and constraint nodes in the target bipartite graph through a graph convolutional neural network to obtain a second feature vector corresponding to the variable node in the target bipartite graph, including: for the target variable node, determining multiple constraint nodes that have an edge relationship with the target variable node, wherein the target variable node is any one of the variable nodes in the target bipartite graph; extracting features of multiple constraint nodes, target variable nodes, and edges between the target variable node and multiple constraint nodes through a graph convolutional neural network to obtain a second feature vector.

[0150] The above-mentioned electronic device can execute the program code of the following steps in the dispatching method of the power system: determining the feasible solution corresponding to the integer variable of the generator group among the multiple generator groups based on the first eigenvector through the graph convolutional neural network, including: processing the first eigenvector through the graph convolutional neural network to obtain the target probability value corresponding to the generator group among the multiple generator groups, wherein the target probability value is used to characterize the probability that the initial solution of the generator group among the multiple generator groups violates the constraints in the mathematical model; judging whether the target probability value is less than a preset threshold; if the target probability value is less than the preset threshold, determining the initial solution corresponding to the integer variable of the generator group among the multiple generator groups as the feasible solution corresponding to the integer variable.

[0151] The above-mentioned electronic device can execute the program code of the following steps in the dispatching method of the power system: determining the feasible solution of the safety constrained unit combination problem based on the feasible solutions corresponding to the integer variables of the generator sets among the multiple generator sets includes: correcting the mathematical model based on the feasible solutions corresponding to the integer variables of the generator sets among the multiple generator sets to obtain a corrected mathematical model; solving the corrected mathematical model to obtain a feasible solution to the safety constrained unit combination problem.

[0152] The above-mentioned electronic device can execute the program code of the following steps in the dispatching method of the power system: after determining whether the target probability value is less than a preset threshold, the method also includes: if there is a target probability value corresponding to the target generator group that is greater than or equal to the preset threshold, then based on the initial solution, determine the initial solution corresponding to the integer variable of the target generator group; perform neighborhood operations on the initial solution corresponding to the integer variable of the target generator group through a variable neighborhood search algorithm to obtain a feasible solution corresponding to the integer variable of the target generator group.

[0153] Optionally, Figure 7 is a structural block diagram of an electronic device according to an embodiment of the present application. Figure 7 As shown, the electronic device 70 may include: one or more ( Figure 7 Only one is shown in the figure) processor 702 and memory 704. The electronic device 70 may also include a storage controller, through which the memory 704 is controlled and managed; the electronic device 70 may also include a peripheral interface, through which the radio frequency module, audio module and display screen are connected.

[0154] Among them, the memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the dispatching method and device of the power system in the embodiment of the present application. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, that is, realizing the above-mentioned dispatching method of the power system. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include a memory remotely arranged relative to the processor, and these remote memories can be connected to the electronic device 20 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0155] The processor can call the information and application programs stored in the memory through the transmission device to perform the following steps: solve the mathematical model of the safety constraint unit combination problem of the power system to obtain an initial solution of the mathematical model, wherein the mathematical model is constructed based on the parameter information of multiple generator sets in the power system; process the initial solution and the mathematical model through a graph convolutional neural network to obtain a feasible solution to the safety constraint unit combination problem, wherein the feasible solution includes the start-up time information and shutdown time information of the generator sets among the multiple generator sets within a period of time in the future, and the output power of the generator sets among the multiple generator sets at a preset time point; and dispatch the power system based on the feasible solution.

[0156] Optionally, the processor may also execute program code for the following steps: solving a mathematical model of the safety-constrained unit combination problem of the power system to obtain an initial solution of the mathematical model, including: solving the linear relaxation of the mathematical model to obtain a relaxed solution; determining whether there are non-integer decision variables in the decision variables of the relaxed solution; if there are non-integer decision variables in the decision variables of the relaxed solution, rounding the non-integer decision variables to obtain an initial solution.

[0157] Optionally, the processor may also execute the program code of the following steps: processing the initial solution and the mathematical model through a graph convolutional neural network to obtain a feasible solution to the safety constrained unit combination problem, including: constructing a target bipartite graph based on the initial solution and the mathematical model; processing the target bipartite graph through a graph convolutional neural network to obtain a first eigenvector of a generator set among multiple generator sets; determining a feasible solution corresponding to an integer variable of a generator set among multiple generator sets based on the first eigenvector through a graph convolutional neural network, wherein the integer variables are the on / off time information of multiple generator sets; determining a feasible solution to the safety constrained unit combination problem based on the feasible solution corresponding to the integer variables of the generator set among multiple generator sets.

[0158] Optionally, the processor may also execute the program code of the following steps: constructing a target bipartite graph based on the initial solution and the mathematical model, including: constructing variable nodes based on the variables in the initial solution and the mathematical model; constructing constraint nodes based on the constraints in the mathematical model; constructing edges between variable nodes and constraint nodes based on the coefficients of the variables in the constraints; constructing a target bipartite graph based on variable nodes, constraint nodes, and edges between variable nodes and constraint nodes.

[0159] Optionally, the processor may also execute the program code of the following steps: processing the target bipartite graph through a graph convolutional neural network to obtain a first eigenvector of a generator group among multiple generator groups, including: processing the information contained in the variable nodes and constraint nodes in the target bipartite graph through a graph convolutional neural network to obtain a second eigenvector corresponding to the variable nodes in the target bipartite graph; determining the correspondence between the variable nodes in the target bipartite graph and the generator groups among multiple generator groups; and aggregating the second eigenvector based on the correspondence to obtain the first eigenvector.

[0160] Optionally, the processor may also execute the program code of the following steps: processing the information contained in the variable nodes and constraint nodes in the target bipartite graph through a graph convolutional neural network to obtain a second feature vector corresponding to the variable nodes in the target bipartite graph, including: for the target variable node, determining multiple constraint nodes having an edge relationship with the target variable node, wherein the target variable node is any one of the variable nodes in the target bipartite graph; performing feature extraction on multiple constraint nodes, target variable nodes, and edges between the target variable node and multiple constraint nodes through a graph convolutional neural network to obtain a second feature vector.

[0161] Optionally, the processor may also execute program code of the following steps: determining a feasible solution corresponding to an integer variable of a generator group among multiple generator groups based on a first eigenvector through a graph convolutional neural network, including: processing the first eigenvector through a graph convolutional neural network to obtain a target probability value corresponding to a generator group among multiple generator groups, wherein the target probability value is used to characterize the probability that an initial solution of a generator group among the multiple generator groups violates the constraints in the mathematical model; determining whether the target probability value is less than a preset threshold; if the target probability value is less than the preset threshold, determining the initial solution corresponding to the integer variable of the generator group among the multiple generator groups as a feasible solution corresponding to the integer variable.

[0162] Optionally, the processor may also execute program code of the following steps: determining a feasible solution to the safety constrained unit combination problem based on feasible solutions corresponding to integer variables of generator sets among multiple generator sets, including: correcting a mathematical model based on feasible solutions corresponding to integer variables of generator sets among multiple generator sets to obtain a corrected mathematical model; solving the corrected mathematical model to obtain a feasible solution to the safety constrained unit combination problem.

[0163] Optionally, the processor may also execute the program code of the following steps: after determining whether the target probability value is less than a preset threshold, the method further includes: if there is a target probability value corresponding to the target generator group that is greater than or equal to the preset threshold, then, based on the initial solution, determining the initial solution corresponding to the integer variable of the target generator group; performing a neighborhood operation on the initial solution corresponding to the integer variable of the target generator group through a variable neighborhood search algorithm to obtain a feasible solution corresponding to the integer variable of the target generator group.

[0164] It can be understood by those skilled in the art that Figure 7 The structure shown is for illustration only, and the electronic device 70 may also be a smart phone (such as an Android phone, an iOS phone, etc.), a tablet computer, a PDA, a mobile Internet device (Mobile Internet Devices, MID), a PAD, or other terminal devices. Figure 7 The structure of the electronic device is not limited. Figure 7 More or fewer components (such as network interfaces, display devices, etc.) shown in, or having Figure 7 Different configurations shown.

[0165] A person of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the hardware related to the terminal device through a program, and the program can be stored in a computer-readable storage medium, and the storage medium may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0166] Example 5

[0167] The embodiment of the present application further provides a computer program product. Optionally, in this embodiment, the computer program product can be used to store the program code executed by the power system scheduling method provided in the first embodiment.

[0168] Optionally, in this embodiment, the computer program product may be located in any computer terminal in a computer terminal group in a computer network, or in any mobile terminal in a mobile terminal group.

[0169] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0170] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0171] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0172] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0173] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0174] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or CD-ROM and other media that can store program codes.

[0175] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A dispatching method for a power system, characterized in that: include: Solving a mathematical model of the safety constraint unit commitment problem of the power system to obtain an initial solution of the mathematical model, wherein the mathematical model is constructed based on parameter information of multiple generator sets of the power system; The initial solution and the mathematical model are processed by a graph convolutional neural network to obtain a feasible solution to the safety-constrained unit commitment problem, wherein the feasible solution includes startup time information and shutdown time information of the generator sets among the multiple generator sets within a future period of time, and output power of the generator sets among the multiple generator sets at a preset time point; The power system is dispatched according to the feasible solution.

2. The method according to claim 1, characterized in that The mathematical model of the safety constraint unit commitment problem of the power system is solved, and the initial solution of the mathematical model is obtained, including: Solving the linear relaxation of the mathematical model to obtain a relaxed solution; Determine whether there is a non-integer decision variable among the decision variables of the relaxation solution; If the non-integer decision variables exist in the decision variables of the relaxed solution, the non-integer decision variables are rounded to obtain the initial solution.

3. The method according to claim 1, characterized in that The initial solution and the mathematical model are processed by a graph convolutional neural network to obtain a feasible solution to the safety-constrained unit commitment problem, including: Constructing a target bipartite graph according to the initial solution and the mathematical model; Processing the target bipartite graph by using the graph convolutional neural network to obtain a first eigenvector of a generator set among the multiple generator sets; Determine, by means of the graph convolutional neural network based on the first feature vector, a feasible solution corresponding to an integer variable of a generator set among the multiple generator sets, wherein the integer variable is on / off time information of the multiple generator sets; A feasible solution to the safety constrained unit commitment problem is determined based on feasible solutions corresponding to integer variables of the generator sets among the plurality of generator sets.

4. The method according to claim 3, characterized in that According to the initial solution and the mathematical model, constructing a target bipartite graph includes: Constructing variable nodes according to the initial solution and the variables in the mathematical model; Constructing constraint nodes according to the constraints in the mathematical model; constructing an edge between the variable node and the constraint node according to the coefficient of the variable in the constraint; The target bipartite graph is constructed according to the variable nodes, the constraint nodes, and the edges between the variable nodes and the constraint nodes.

5. The method according to claim 3, characterized in that: Processing the target bipartite graph by the graph convolutional neural network to obtain a first eigenvector of a generator set among the multiple generator sets includes: Processing information contained in the variable nodes and constraint nodes in the target bipartite graph by the graph convolutional neural network to obtain a second eigenvector corresponding to the variable nodes in the target bipartite graph; Determining a correspondence between a variable node in the target bipartite graph and a generator set in the plurality of generator sets; According to the corresponding relationship, the second feature vector is aggregated to obtain the first feature vector.

6. The method according to claim 5, characterized in that The information contained in the variable nodes and constraint nodes in the target bipartite graph is processed by the graph convolutional neural network to obtain the second eigenvector corresponding to the variable nodes in the target bipartite graph, including: For a target variable node, determining a plurality of constraint nodes having an edge relationship with the target variable node, wherein the target variable node is any one of the variable nodes in the target bipartite graph; The second feature vector is obtained by performing feature extraction on the multiple constraint nodes, the target variable node, and the edges between the target variable node and the multiple constraint nodes through the graph convolutional neural network.

7. The method according to claim 3, characterized in that Determining a feasible solution corresponding to an integer variable of a generator set among the plurality of generator sets based on the first feature vector by using the graph convolutional neural network includes: Processing the first feature vector by the graph convolutional neural network to obtain a target probability value corresponding to a generator set among the multiple generator sets, wherein the target probability value is used to characterize the probability that an initial solution of a generator set among the multiple generator sets violates a constraint in the mathematical model; Determine whether the target probability value is less than a preset threshold; If the target probability value is less than the preset threshold, the initial solution corresponding to the integer variable of the generator set among the multiple generator sets is determined as the feasible solution corresponding to the integer variable.

8. The method according to claim 3, characterized in that Determining a feasible solution to the safety constrained unit commitment problem according to feasible solutions corresponding to integer variables of the generator sets among the plurality of generator sets includes: Modifying the mathematical model according to feasible solutions corresponding to integer variables of the generator sets among the multiple generator sets to obtain a modified mathematical model; The modified mathematical model is solved to obtain a feasible solution to the safety-constrained unit commitment problem.

9. The method according to claim 7, characterized in that: After determining whether the target probability value is less than a preset threshold, the method further includes: If there is a target probability value corresponding to a target generator set that is greater than or equal to the preset threshold, determining an initial solution corresponding to the integer variable of the target generator set according to the initial solution; A neighborhood operation is performed on an initial solution corresponding to the integer variable of the target generator set by a variable neighborhood search algorithm to obtain a feasible solution corresponding to the integer variable of the target generator set.

10. A dispatching method for a power system, characterized in that: include: Acquire a mathematical model of a safety-constrained unit commitment problem of a power system uploaded by a client, wherein the mathematical model is constructed based on parameter information of a plurality of generator sets of the power system; A mathematical model of the safety-constrained unit combination problem of the power system is solved in a cloud server to obtain an initial solution of the mathematical model; the initial solution and the mathematical model are processed by a graph convolutional neural network to obtain a feasible solution to the safety-constrained unit combination problem, wherein the feasible solution includes startup time information and shutdown time information of the generator sets among the multiple generator sets within a future period of time, and output power of the generator sets among the multiple generator sets at a preset time point; The feasible solution is fed back to the client so that the power system is dispatched according to the feasible solution.

11. A dispatching device for a power system, characterized in that: include: A solving unit, used for solving a mathematical model of the safety constraint unit commitment problem of the power system to obtain an initial solution of the mathematical model, wherein the mathematical model is constructed based on parameter information of multiple generator sets of the power system; a processing unit, configured to process the initial solution and the mathematical model through a graph convolutional neural network to obtain a feasible solution to the safety-constrained unit commitment problem, wherein the feasible solution includes startup time information and shutdown time information of a generator set among the multiple generator sets within a future period of time, and output power of a generator set among the multiple generator sets at a preset time point; A dispatching unit is used to dispatch the power system according to the feasible solution.

12. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the storage medium is located is controlled to execute the dispatching method for the power system according to any one of claims 1 to 10.

13. An electronic device, characterized in that: include: A memory storing an executable program; A processor is used to run the program, wherein the program, when running, executes the dispatching method of the power system described in any one of claims 1 to 10.

14. A computer program product, characterized in that The method comprises a computer program or an instruction, which, when executed by a processor, implements the dispatching method of the power system according to any one of claims 1 to 10.

Images