A method and device for selecting a thermal starting point of a unit combination scheme, an electronic device, and a storage medium

Abstract

The application discloses a kind of unit combination scheme hot start point selection method, device, electronic equipment and storage medium, the method includes: according to the historical load curve data of several typical days of power grid, the load curve data of power grid operation day and power grid topology structure, generate corresponding historical typical day undirected graph and the undirected graph of power grid operation day;Using graph similarity evaluation network model, the similarity score between the undirected graph of power grid operation day and each historical typical day undirected graph is calculated, wherein, each training sample of model is composed of two historical typical day undirected graph and its similarity label, the optimal unit combination scheme corresponding to the highest similarity score of historical typical day undirected graph is selected as the target unit combination scheme of power grid operation day, and it is used as the hot start point of the total operation cost minimization model solution. Through the implementation of the present application, the solving efficiency can be improved, and the stability and reliability of the power grid are maintained.

Description

Technical Field

[0001] The present invention relates to the technical field of power system dispatching automation, and in particular to a method, device, electronic equipment and storage medium for selecting a hot start point of a unit combination scheme. Background Art

[0002] Generator unit deployment plays a crucial role in power system scheduling. Its core goal is to proactively schedule generator start / stops and output based on load demand and renewable energy forecasts to achieve power system economic efficiency and safety. A reasonable generator unit deployment plan not only optimizes resource allocation and reduces power generation costs, but also ensures stable and reliable power supply.

[0003] However, existing technologies for solving unit combination problems often suffer from slow solution speeds. While traditional methods can provide high-quality optimal solutions, they often suffer from high time complexity when dealing with large power grids or complex operating conditions. This is particularly true during peak power demand periods, where lengthy solution times can delay power dispatch decisions, impacting grid stability and reliability. Summary of the Invention

[0004] Embodiments of the present invention provide a method, device, electronic device, and storage medium for selecting a hot start point for a unit combination scheme. By implementing the present invention, the historical unit combination scheme most similar to a grid operation day can be selected as the hot start point for that grid operation day, thereby increasing the speed of solving the total operating cost minimization model and improving the stability and reliability of the grid.

[0005] An embodiment of the present invention provides a method for selecting a hot start point of a unit combination scheme, comprising:

[0006] Obtain load curve data of several typical historical days of the power grid, load curve data of power grid operation days, and topological structure of the power grid;

[0007] Generate an undirected graph of several historical typical days based on the topological structure of the power grid and the load curve data of several historical typical days of the power grid; generate an undirected graph of the power grid operation day based on the topological structure of the power grid and the load curve data of the power grid operation day;

[0008] The undirected graphs of each historical typical day and the undirected graphs of the power grid operation day are sequentially input into the graph similarity evaluation network model to generate similarity scores between the undirected graphs of the power grid operation day and the undirected graphs of each historical typical day; wherein, each training sample in training the graph similarity evaluation network model includes: undirected graphs of two different historical typical days and actual similarity labels between the two; the actual similarity of the undirected graphs of the two different historical typical days is calculated based on the optimal unit combination schemes of the two different historical typical days; the optimal unit combination scheme is the unit combination scheme with the lowest total operating cost;

[0009] The undirected graph of the historical typical day with the highest similarity score is selected as the target undirected graph; the optimal unit combination plan of the typical day corresponding to the target undirected graph is used as the target unit combination plan for the power grid operation day;

[0010] The target unit combination scheme is used as the hot starting point for solving the total operating cost minimization model.

[0011] Further, the generation methods of undirected graphs include:

[0012] Obtaining load curve data of the power grid corresponding to a date and the topology of the power grid; wherein the load curve data includes the load value of the load node at the scheduling time;

[0013] According to the topological structure of the power grid, the vertices and edges of the undirected graph corresponding to the date are generated; wherein the vertices of the undirected graph are the load nodes in the topological structure; and the edges of the undirected graph are the transmission lines in the topological structure;

[0014] According to the connection relationship of load nodes in the topological structure of the power grid, the adjacency matrix of the undirected graph is calculated and generated;

[0015] According to the load curve data of the power grid on the corresponding date, the characteristic value of each vertex in the undirected graph is generated.

[0016] Furthermore, the training process of the graph similarity evaluation network model includes:

[0017] Randomly select a preset number of training samples and repeat the training operation until the graph similarity evaluation network model converges;

[0018] The training operation includes:

[0019] Input the current training sample into the graph similarity evaluation network model to obtain the similarity score of the current training sample;

[0020] According to the mean square error function, the error between the similarity score of the current training sample and the actual similarity is calculated to generate the mean square error;

[0021] Based on the mean square error, the graph similarity evaluation network model is judged to see whether it has converged;

[0022] If the model has not converged, the Adam optimizer is used to optimize the model parameters according to the current mean square error, and the next batch of training samples is selected to update the current training samples.

[0023] Furthermore, the actual similarity is calculated using the following formula:

[0024]

[0025] Among them, η ij is the actual similarity between the undirected graphs of day i and day j in the historical typical day; is the start and stop status of unit g at time t on day i; is the start and stop status of unit g at time t on day j; is the maximum active power of unit g, that is, the rated capacity of unit g; N g is the unit set; N t A set of scheduling periods.

[0026] Furthermore, the optimal unit combination scheme is determined by the following methods:

[0027] Obtain a solver for solving the total operating cost minimization model, hardware parameters of the unit, cost parameters of the unit, hardware parameters of the transmission line, and load curve data for the corresponding date; wherein the hardware parameters of the unit include: the maximum active power generation of the unit, the upper limit of the unit's ramp-up power increment, the upper limit of the unit's ramp-down power increment, the power limit at unit startup, the power limit at unit shutdown, the ratio of spinning reserve capacity, and the minimum operating time of the unit; the cost parameters of the unit include the fuel cost coefficient of the unit; the hardware parameters of the transmission line include: the active power limit of the line, the ratio of spinning reserve capacity, the power transfer factor of the line relative to the unit, and the power transfer factor of the line relative to the load node; the load curve data include: the load value of the load node at the scheduling time;

[0028] Define the decision variables of the total operating cost minimization model in the solver; the decision variables include the start and stop status of the unit and the active power generated by the unit;

[0029] Convert the objective function and constraints into expressions and import them into the solver;

[0030] The hardware parameters of the unit, the cost parameters of the unit, the hardware parameters of the transmission line, and the load curve data of the corresponding date are set as the parameters of the solver, and the solver is called to solve the problem so that the solver calculates the start and stop status and active power generation for each scheduling period on the corresponding date as the optimal unit combination plan for the current date.

[0031] Furthermore, the total operating cost minimization model includes an objective function and constraints;

[0032] The objective function includes:

[0033]

[0034] Among them, F G represents the total operating cost of the unit combination; is the start and stop status of unit g at time t on day i; is the active power generated by unit g at time t on day i; is the start and stop status of unit g at time t-1 on day i; S UD The startup and shutdown costs of the unit; is the fuel cost of unit g at time t on day i;

[0035] Fuel cost is calculated by the following formula

[0036]

[0037] Among them, a g 、b g and c g is the fuel cost coefficient of unit g;

[0038] The constraints include: system power balance constraint, generator set output constraint, generator set ramp constraint, spinning reserve capacity constraint, unit minimum downtime constraint, unit minimum operating time constraint and branch power flow constraint.

[0039] Furthermore, the system power balance constraint is:

[0040]

[0041] in, is the active power generated by unit g at time t on day i; N g Assemble for the crew; is the load of load node n at time t on day i; N n is the set of load nodes; N t is the scheduling period set;

[0042] The generator set output constraint is:

[0043]

[0044] in, is the active power generated by unit g at time t on day i; is the minimum active power generated by unit g; is the maximum active power generated by unit g, i.e. the rated capacity of unit g;

[0045] The generator set climbing constraint is:

[0046]

[0047]

[0048] in, is the active power generated by unit g at time t on day i; is the active power generated by unit g at time t-1 on day i; is the start and stop status of unit g at time t on day i; is the start and stop status of unit g at time t-1 on day i; It is the upper limit of the unit's ramp-up power increment, i.e. the maximum power that can be increased within a specific time period; It is the upper limit of the unit's g-down ramp power increment, that is, the maximum power that can be reduced within a specific time period; is the power limit when unit g starts; is the power limit when unit g is shut down;

[0049] The spinning reserve capacity constraint is:

[0050]

[0051] in, is the start and stop status of unit g at time t on day i; is the maximum active power of unit g, that is, the rated capacity of unit g; Q s is the proportion of spinning reserve capacity, is the load of load node n at time t on day i; N n is the set of load nodes; N t is the scheduling period set;

[0052] The minimum operating time constraint of the unit is:

[0053]

[0054]

[0055] in, is the minimum operating time of unit g; is the start and stop status of unit g at time t on day i; is the start and stop status of unit g at time t-1 on day i; is the start and stop status of unit g at time k on day i;

[0056] The minimum downtime constraint of the unit is:

[0057]

[0058]

[0059] in, is the minimum downtime of unit g; is the start and stop status of unit g at time t on day i; is the start and stop status of unit g at time t-1 on day i;

[0060] The branch power flow constraint is:

[0061]

[0062] in, is the active power limit of line l, LF lg is the power transfer factor of line l relative to unit g, LF ln is the power transfer factor of line l relative to load node n; is the active power generated by unit g at time t on day i; is the load of load node n at time t on day i; N n is the set of load nodes; N l is the set of lines; N t A set of scheduling periods.

[0063] Based on the above method embodiments, the present invention provides corresponding device embodiments.

[0064] An embodiment of the present invention provides a device for selecting a hot start point of a unit combination scheme, comprising: a data acquisition module, an undirected graph generation module, a graph similarity evaluation module, and a hot start point selection module;

[0065] The data acquisition module is used to obtain load curve data of several historical typical days of the power grid, load curve data of the power grid operation day and the topology structure of the power grid;

[0066] The undirected graph generation module is used to generate undirected graphs of several historical typical days based on the topological structure of the power grid and the load curve data of several historical typical days of the power grid; and to generate an undirected graph of the power grid operation day based on the topological structure of the power grid and the load curve data of the power grid operation day;

[0067] The graph similarity assessment module is configured to sequentially input the undirected graphs of each historical typical day and the undirected graphs of the power grid operation day into a graph similarity assessment network model to generate similarity scores between the undirected graphs of the power grid operation day and the undirected graphs of each historical typical day; wherein each training sample in training the graph similarity assessment network model includes: undirected graphs of two different historical typical days and actual similarity labels between the two undirected graphs; the actual similarity of the undirected graphs of the two different historical typical days is calculated based on the optimal unit combination schemes of the two different historical typical days; the optimal unit combination scheme is the unit combination scheme with the lowest total operating cost;

[0068] The hot start point selection module is used to select the undirected graph of the historical typical day with the highest similarity score as the target undirected graph; use the optimal unit combination plan of the typical day corresponding to the target undirected graph as the target unit combination plan for the power grid operation day; and use the target unit combination plan as the hot start point for solving the total operating cost minimization model.

[0069] Based on the above method embodiment, the present invention provides a corresponding electronic device embodiment.

[0070] An embodiment of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for selecting a hot start point of a unit combination scheme as described in any one of the above-mentioned method embodiments can be implemented.

[0071] Based on the above method embodiment, the present invention provides a corresponding storage medium embodiment.

[0072] An embodiment of the present invention provides a storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for selecting a hot start point of a unit combination scheme described in any one of the above method embodiments can be implemented.

[0073] Compared with the prior art, the present invention has the following beneficial effects:

[0074] The embodiment of the present application provides a kind of unit combination scheme hot start point selection method, device, electronic equipment and storage medium.The scheme is by constructing the undirected graph of power grid operation day and several historical typical days, these undirected graph is input into graph similarity evaluation network model, the similarity score between power grid operation day and each historical typical day can be calculated, further, the unit combination scheme corresponding to the historical typical day with highest similarity score is selected as the hot start point of power grid operation day, effectively avoid the process of total operation cost minimization model from beginning to solve, by using previous effective solution, it can quickly guide solving algorithm to converge to high-quality solution faster, so as to greatly shorten the calculation time, when processing complex power system scheduling, it can significantly improve solving efficiency, ensure that timely scheduling decision can be made in power demand peak period, maintain the stability and reliability of power grid. BRIEF DESCRIPTION OF DRAWINGS

[0075] Figure 1 It is a kind of unit combination scheme hot start point selection method flow diagram provided by the embodiment of the present application.

[0076] Figure 2 It is a kind of unit combination scheme hot start point selection device structure diagram provided by the embodiment of the present application. DETAILED DESCRIPTION

[0077] The technical solutions in the embodiments of the present application will be described clearly and completely in the embodiments of the present application in conjunction with the drawings, obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments.Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0078] As Figure 1 The embodiment of the present application provides a kind of unit combination scheme hot start point selection method, at least includes the following steps:

[0079] Step S1, obtain the load curve data of several historical typical days of power grid, the load curve data of power grid operation day and the topology structure of power grid;

[0080] Specifically, a historical typical day refers to a representative date selected from historical data in the power system. The load curves on these dates can reflect the operation of the power grid under specific conditions and can represent the operating characteristics of the power grid under similar conditions, which is helpful to optimize the unit combination plan and improve the operation efficiency of the power grid; a power grid operation day refers to a day when the system needs to perform actual operation and load scheduling, that is, the date that is currently being analyzed, scheduled or predicted. The load curve data and power grid topology structure of this day are known and are used to optimize the unit combination plan of the power grid on this day; the load curve data is the load value of the load node at the scheduling moment; the topology of the power grid refers to the network layout formed by each node and transmission line in the power grid; among them, the nodes include: power generation nodes, substation nodes and load nodes.

[0081] Step S2: generating an undirected graph of several historical typical days based on the topological structure of the power grid and the load curve data of several historical typical days of the power grid; generating an undirected graph of the power grid operation day based on the topological structure of the power grid and the load curve data of the power grid operation day;

[0082] Specifically, in a preferred embodiment, the method for generating an undirected graph includes:

[0083] Obtaining load curve data of the power grid corresponding to a date and the topology of the power grid; wherein the load curve data includes the load value of the load node at the scheduling time;

[0084] According to the topological structure of the power grid, the vertices and edges of the undirected graph corresponding to the date are generated; wherein the vertices of the undirected graph are the load nodes in the topological structure; and the edges of the undirected graph are the transmission lines in the topological structure;

[0085] According to the connection relationship of load nodes in the topological structure of the power grid, the adjacency matrix of the undirected graph is calculated and generated;

[0086] According to the load curve data of the power grid on the corresponding date, the characteristic value of each vertex in the undirected graph is generated.

[0087] Exemplarily, calculating and generating an adjacency matrix of an undirected graph based on the connection relationship of load nodes in the topological structure of the power grid includes:

[0088] The connection between each load node is checked. If there is a direct transmission line connecting two nodes in the power grid, the intersection position corresponding to these two nodes in the adjacency matrix is ​​set to 1, indicating that they are connected. If there is no direct connection between the two nodes, the corresponding position in the matrix remains 0. The matrix is ​​symmetrical, so a line will be marked as 1 symmetrically at the positions of two nodes, indicating a bidirectional connection between them.

[0089] Exemplarily, generating the characteristic value of each vertex in the undirected graph based on the load curve data of the power grid corresponding to the date includes:

[0090] A day is divided into multiple time periods, and the active load value of the load node in each time period is extracted from the load curve data in sequence. The feature value of each vertex is composed of the active load value set of the load node in these time periods.

[0091] It can be understood that these eigenvalues ​​not only reflect the load levels of load nodes in different time periods, but also reveal the load trends over time, providing important basic data for subsequent graph similarity assessments. Furthermore, the integration of these eigenvalues ​​provides strong support for dynamic analysis and optimized scheduling of power grids.

[0092] Step S3: sequentially inputting the undirected graphs of each historical typical day and the undirected graphs of the power grid operation day into a graph similarity evaluation network model to generate similarity scores between the undirected graphs of the power grid operation day and the undirected graphs of each historical typical day; wherein, each training sample in training the graph similarity evaluation network model includes: undirected graphs of two different historical typical days and actual similarity labels between the two undirected graphs; the actual similarity of the undirected graphs of the two different historical typical days is calculated based on the optimal unit combination schemes of the two different historical typical days; the optimal unit combination scheme is the unit combination scheme with the lowest total operating cost;

[0093] In an optional embodiment, the training process of the graph similarity evaluation network model includes:

[0094] Randomly select a preset number of training samples and repeat the training operation until the graph similarity evaluation network model converges;

[0095] The training operation includes:

[0096] Input the current training sample into the graph similarity evaluation network model to obtain the similarity score of the current training sample;

[0097] According to the mean square error function, the error between the similarity score of the current training sample and the actual similarity is calculated to generate the mean square error;

[0098] Based on the mean square error, the graph similarity evaluation network model is judged to see whether it has converged;

[0099] If the model has not converged, the Adam optimizer is used to optimize the model parameters according to the current mean square error, and the next batch of training samples is selected to update the current training samples.

[0100] It can be understood that judging whether the graph similarity evaluation network model has converged is based on the mean square error, including: judging whether the performance of the model has stabilized within a smaller range based on the calculated mean square error. If the mean square error is lower than the preset threshold, the model is considered to have converged; if not, the model still needs further optimization.

[0101] In an optional embodiment, the structure of the graph similarity evaluation network model includes: an input layer, a graph convolution layer, and a fully connected layer;

[0102] The input layer is used to receive two undirected graphs to be evaluated;

[0103] The graph convolution layer is used to extract features corresponding to undirected graphs. It includes three layers of convolution processing. The first layer converts 24-dimensional input features into 128 dimensions, the second layer converts 128-dimensional features into 96 dimensions, and the third layer converts 96-dimensional features into 48-dimensional features. The output of all convolution layers uses the tanh activation function to increase the nonlinear expression ability of the model and effectively capture the structural information of the graph.

[0104] The fully connected layer is used to fuse and transform the extracted features. It contains three fully connected layers. The first layer maps the 48-dimensional input features to 144 dimensions, the second layer maps the 144 dimensions to 36 dimensions, and the third layer outputs a 1-dimensional similarity evaluation value. The third layer is processed by the sigmoid activation function to obtain the final similarity score.

[0105] In a preferred embodiment, the actual similarity is calculated using the following formula:

[0106]

[0107] Among them, η ij is the actual similarity between the undirected graphs of the i-th day and the j-th day in the historical typical day; is the start and stop status of unit g at time t on day i; is the start and stop status of unit g at time t on day j; is the maximum active power of unit g, that is, the rated capacity of unit g; N g is the unit set; N t A set of scheduling periods.

[0108] The optimal unit combination solution is determined by the following methods:

[0109] Preferably, a solver for solving a total operating cost minimization model, hardware parameters of the unit, cost parameters of the unit, hardware parameters of the transmission line, and load curve data of a corresponding date are obtained; wherein the hardware parameters of the unit include: the maximum active power of the unit, the upper limit of the unit's ramp-up power increment, the upper limit of the unit's ramp-down power increment, the power limit when the unit is started, the power limit when the unit is shut down, the ratio of the spinning reserve capacity, and the minimum operating time of the unit; the cost parameters of the unit include the fuel cost coefficient of the unit; the hardware parameters of the transmission line include: the active power limit of the line, the ratio of the spinning reserve capacity, the power transfer factor of the line relative to the unit, and the power transfer factor of the line relative to the load node; the load curve data include: the load value of the load node at the scheduling time;

[0110] Define the decision variables of the total operating cost minimization model in the solver; the decision variables include the start and stop status of the unit and the active power generated by the unit;

[0111] Convert the objective function and constraints into expressions and import them into the solver;

[0112] The hardware parameters of the unit, the cost parameters of the unit, the hardware parameters of the transmission line, and the load curve data of the corresponding date are set as the parameters of the solver, and the solver is called to solve the problem so that the solver calculates the start and stop status and active power generation for each scheduling period on the corresponding date as the optimal unit combination plan for the current date.

[0113] It should be noted that the solver is an existing commercial solver, which can be selected from Gurobi, CPLEX, MOSEK, GLPK, SCIP, COIN-OR, Pyomo and AMPL for solving.

[0114] Step S4: Select the undirected graph of the historical typical day with the highest similarity score as the target undirected graph; and use the optimal unit combination plan of the typical day corresponding to the target undirected graph as the target unit combination plan for the power grid operation day.

[0115] It can be understood that using the optimal unit combination plan for a typical day corresponding to the target undirected graph as the target unit combination plan for the grid operation day ensures the grid achieves optimal operating efficiency and economy under current operating conditions. By referencing the optimal plans for historical typical days, this method improves the scientific nature and accuracy of unit combinations, provides a reliable decision-making basis for grid operation days, and reduces operational risks caused by insufficient information or misjudgments. Furthermore, leveraging historical data experience, it is possible to optimize overall operating costs while maintaining safe and stable grid operation, thereby enhancing the flexibility and responsiveness of grid management.

[0116] Step S5: Using the target unit combination scheme as a hot start point for solving the total operating cost minimization model.

[0117] In a preferred embodiment, the total operating cost minimization model includes an objective function and constraints;

[0118] The objective function includes:

[0119]

[0120] Among them, F G represents the total operating cost of the unit combination; is the start and stop status of unit g at time t on day i; is the active power generated by unit g at time t on day i; is the start and stop status of unit g at time i-1 on day i; S UD The startup and shutdown costs of the unit; is the fuel cost of unit g at time t on day i;

[0121] Fuel cost is calculated by the following formula

[0122]

[0123] Among them, a g 、b g and c g is the fuel cost coefficient of unit g;

[0124] The constraints include: system power balance constraint, generator set output constraint, generator set ramp constraint, spinning reserve capacity constraint, unit minimum downtime constraint, unit minimum operating time constraint and branch power flow constraint.

[0125] In a preferred embodiment, the system power balance constraint is:

[0126]

[0127] in, is the active power generated by unit g at time t on day i; N g Assemble for the crew; is the load of load node n at time t on day i; N n is the set of load nodes; N t is the scheduling period set;

[0128] The generator set output constraint is:

[0129]

[0130] in, is the active power generated by unit g at time t on day i; is the minimum active power generated by unit g; is the maximum active power generated by unit g, i.e. the rated capacity of unit g;

[0131] The generator set climbing constraint is:

[0132]

[0133]

[0134] in, is the active power generated by unit g at time t on day i; is the active power generated by unit g at time t-1 on day i; is the start and stop status of unit g at time t on day i; is the start and stop status of unit g at time t-1 on day i; It is the upper limit of the unit's ramp-up power increment, i.e. the maximum power that can be increased within a specific time period; It is the upper limit of the unit's g-down ramp power increment, that is, the maximum power that can be reduced within a specific time period; is the power limit when unit g starts; is the power limit when unit g is shut down;

[0135] The spinning reserve capacity constraint is:

[0136]

[0137] in, is the start and stop status of unit g at time t on day i; is the maximum active power of unit g, that is, the rated capacity of unit g; Q s is the proportion of spinning reserve capacity, is the load of load node n at time t on day i; N n is the set of load nodes; N t is the scheduling period set;

[0138] The minimum operating time constraint of the unit is:

[0139]

[0140]

[0141] in, is the minimum operating time of unit g; is the start and stop status of unit g at time t on day i; is the start and stop status of unit g at time t-1 on day i; is the start and stop status of unit g at time k on day i;

[0142] The minimum downtime constraint of the unit is:

[0143]

[0144]

[0145] in, is the minimum downtime of unit g; is the start and stop status of unit g at time t on day i; is the start and stop status of unit g at time t-1 on day i;

[0146] The branch power flow constraint is:

[0147]

[0148] in, is the active power limit of line l, LF lg is the power transfer factor of line l relative to unit g, LF ln is the power transfer factor of line l relative to load node n; is the active power generated by unit g at time t on day i; is the load of load node n at time t on day i; N n is the set of load nodes; N l is the set of lines; N t A set of scheduling periods.

[0149] For example, N is the start / stop status of unit g at time t on day i, and can be 0 or 1, where 0 means shutdown and 1 means startup; t is the set of scheduling periods, N t ={1, 2, 3, ..., 24}.

[0150] Understandably, when using a direct solution approach to solve a total operating cost minimization model, it typically starts with a random or default initial state. This can lead to slower convergence, as the initial point is often far from the optimal solution. This requires more iterations to reach a satisfactory solution, increasing computational time and resource consumption. Furthermore, the quality of the solution obtained by direct solution may be suboptimal, as it explores a larger search space and may miss some potentially high-quality solutions.

[0151] The target unit commitment scheme is used as the hot starting point for solving the total operating cost minimization model. The hot starting point solution method utilizes an undirected graph of historical typical days and the corresponding optimal unit commitment scheme to select an initial state close to the optimal solution, providing a good starting point for the optimization process. This method ensures faster convergence of the optimization algorithm, significantly reducing the number of required iterations, thereby reducing computing time and resource consumption. Furthermore, because the hot starting point is selected based on historical data and actual operating conditions, it can better guide the search process, improve the quality of the solution, and avoid falling into local optimal solutions.

[0152] Another embodiment of the present invention provides a method for selecting a hot start point of a unit combination scheme, which includes, in addition to steps S1 to S5 of the above embodiment, the following steps after step S5:

[0153] Step S6: input the target unit combination scheme into the solver for solution, and determine whether the solution of the total operating cost minimization model at this time is feasible; if the solution is feasible, the target unit combination scheme at this time is the hot start point; if the solution is not feasible, the undirected graph of the historical typical day with the highest similarity score that has not been selected is used as the target undirected graph, and the optimal unit combination scheme of the typical day corresponding to the target undirected graph is updated to the target unit combination scheme of the power grid operation day, and the target unit combination scheme at this time is input into the solver for re-solution until the solution of the total operating cost minimization model is feasible.

[0154] Specifically, in a preferred embodiment, three grid operation days were selected to set the hot start point to solve the total operating cost minimization model, and the solution time was compared with the solution time directly solved by the solver. The comparison results are as follows:

[0155]

[0156]

[0157] Among them, in the grid operation day 1, the unit combination scheme corresponding to the historical typical day 03-01 is not feasible to solve; in the grid operation day 2, the unit combination scheme corresponding to the historical typical day 05-12 is not feasible to solve.

[0158] Based on the above method embodiments, the present invention provides corresponding device embodiments.

[0159] like Figure 2 As shown, an embodiment of the present invention provides a device for selecting a hot start point of a unit combination scheme, comprising: a data acquisition module 101 , an undirected graph generation module 102 , a graph similarity evaluation module 103 and a hot start point selection module 104 .

[0160] The data acquisition module 101 is used to obtain load curve data of several historical typical days of the power grid, load curve data of the power grid operation day and the topology structure of the power grid;

[0161] The undirected graph generation module 102 is used to generate undirected graphs of several historical typical days based on the topology of the power grid and the load curve data of several historical typical days of the power grid; and to generate an undirected graph of the power grid operation day based on the topology of the power grid and the load curve data of the power grid operation day;

[0162] The graph similarity evaluation module 103 is configured to sequentially input the undirected graphs of each historical typical day and the undirected graphs of the power grid operation day into a graph similarity evaluation network model to generate similarity scores between the undirected graphs of the power grid operation day and the undirected graphs of each historical typical day; wherein each training sample in training the graph similarity evaluation network model includes: undirected graphs of two different historical typical days and actual similarity labels between the two undirected graphs; the actual similarity of the undirected graphs of the two different historical typical days is calculated based on the optimal unit combination schemes of the two different historical typical days; the optimal unit combination scheme is the unit combination scheme with the lowest total operating cost;

[0163] The hot start point selection module 104 is used to select the undirected graph of the historical typical day with the highest similarity score as the target undirected graph; use the optimal unit combination plan of the typical day corresponding to the target undirected graph as the target unit combination plan for the power grid operation day; and use the target unit combination plan as the hot start point for solving the total operating cost minimization model.

[0164] It should be noted that the embodiments of the device described above correspond to the above-mentioned embodiments of the present invention, and can implement any of the methods described above in the present invention. In addition, the embodiments of the above-mentioned device are merely schematic, wherein the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the scheme of this embodiment. In addition, in the drawings of the embodiment of the device provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which can be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art can understand and implement it without paying any creative work.

[0165] Based on the above method embodiment of the present invention, a corresponding electronic device embodiment is provided.

[0166] An embodiment of the present application provides an electronic device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, when the computer program is executed by the processor, a method for selecting a heat start point of a unit combination scheme according to any one of the present application is implemented, or when the computer program is executed by the processor, functions of each module in each device embodiment are implemented.

[0167] For example, the computer program can be divided into one or more modules, the one or more modules are stored in the memory and executed by the processor to complete the present application. The one or more modules can be a series of computer program instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program in the terminal device.

[0168] The terminal device can be a desktop computer, a notebook computer, a palm computer, a cloud server and other computing devices. The terminal device can include, but is not limited to, a processor and a memory.

[0169] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. The processor is a control center of the terminal device, and is connected with all parts of the terminal device through various interfaces and lines.

[0170] The memory can be used to store the computer programs and / or modules, and the processor implements various functions of the terminal device by running or executing the computer programs and / or modules stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required for a function, etc.; the data storage area can store data created based on the use of the mobile phone, etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage device.

[0171] Based on the above method embodiment, the present invention provides a corresponding storage medium embodiment;

[0172] Another embodiment of the present invention provides a storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the storage medium is located is controlled to execute the method for selecting a hot start point of any of the above-mentioned unit combination schemes of the present invention.

[0173] The above-mentioned storage medium is a computer-readable storage medium, and the computer program includes computer program code, which may be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the content contained in the computer-readable medium may be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0174] In the description of this specification, the reference terms "one embodiment," "some embodiments," "example," "specific example," or "some examples" mean that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. Moreover, the specific features, structures, materials, or characteristics described may be combined in any appropriate manner in any one or more embodiments or examples. In addition, those skilled in the art may combine and integrate different embodiments or examples described in this specification, as well as features of different embodiments or examples, unless they are mutually inconsistent.

[0175] The above is a preferred embodiment of the present invention. 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 invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for selecting a hot start point for a unit combination scheme, characterized in that: include: Obtain load curve data of several typical historical days of the power grid, load curve data of power grid operation days, and topological structure of the power grid; Generate an undirected graph of several historical typical days based on the topological structure of the power grid and the load curve data of several historical typical days of the power grid; generate an undirected graph of the power grid operation day based on the topological structure of the power grid and the load curve data of the power grid operation day; The undirected graphs of each historical typical day and the undirected graphs of the power grid operation day are sequentially input into the graph similarity evaluation network model to generate similarity scores between the undirected graphs of the power grid operation day and the undirected graphs of each historical typical day; wherein, each training sample in training the graph similarity evaluation network model includes: undirected graphs of two different historical typical days and actual similarity labels between the two; the actual similarity of the undirected graphs of the two different historical typical days is calculated based on the optimal unit combination schemes of the two different historical typical days; the optimal unit combination scheme is the unit combination scheme with the lowest total operating cost; The undirected graph of the historical typical day with the highest similarity score is selected as the target undirected graph; the optimal unit combination plan of the typical day corresponding to the target undirected graph is used as the target unit combination plan for the power grid operation day; The target unit combination scheme is used as the hot starting point for solving the total operating cost minimization model.

2. The method for selecting a hot start point of a unit combination scheme according to claim 1, characterized in that: The methods for generating undirected graphs include: Obtaining load curve data of the power grid corresponding to a date and the topology of the power grid; wherein the load curve data includes the load value of the load node at the scheduling time; According to the topological structure of the power grid, the vertices and edges of the undirected graph corresponding to the date are generated; wherein the vertices of the undirected graph are the load nodes in the topological structure; and the edges of the undirected graph are the transmission lines in the topological structure; According to the connection relationship of load nodes in the topological structure of the power grid, the adjacency matrix of the undirected graph is calculated and generated; According to the load curve data of the power grid on the corresponding date, the characteristic value of each vertex in the undirected graph is generated.

3. The method for selecting a hot start point of a unit combination scheme according to claim 1, characterized in that: The training process of the graph similarity evaluation network model includes: Randomly select a preset number of training samples and repeat the training operation until the graph similarity evaluation network model converges; The training operation includes: Input the current training sample into the graph similarity evaluation network model to obtain the similarity score of the current training sample; According to the mean square error function, the error between the similarity score of the current training sample and the actual similarity is calculated to generate the mean square error; Based on the mean square error, the graph similarity evaluation network model is judged to see whether it has converged; If the model has not converged, the Adam optimizer is used to optimize the model parameters according to the current mean square error, and the next batch of training samples is selected to update the current training samples.

4. The method for selecting a hot start point of a unit combination scheme according to claim 1, characterized in that: The actual similarity is calculated using the following formula: Among them, η ij is the actual similarity between the undirected graphs of day i and day j in the historical typical day; is the start and stop status of unit g at time t on day i; is the start and stop status of unit g at time t on day j; is the maximum active power of unit g, that is, the rated capacity of unit g; N g is the unit set; N t A set of scheduling periods.

5. The method for selecting a hot start point of a unit combination scheme according to claim 1, characterized in that: The optimal unit combination solution is determined by the following methods: Obtain a solver for solving the total operating cost minimization model, hardware parameters of the unit, cost parameters of the unit, hardware parameters of the transmission line, and load curve data for the corresponding date; wherein the hardware parameters of the unit include: the maximum active power generation of the unit, the upper limit of the unit's ramp-up power increment, the upper limit of the unit's ramp-down power increment, the power limit at unit startup, the power limit at unit shutdown, the ratio of spinning reserve capacity, and the minimum operating time of the unit; the cost parameters of the unit include the fuel cost coefficient of the unit; the hardware parameters of the transmission line include: the active power limit of the line, the ratio of spinning reserve capacity, the power transfer factor of the line relative to the unit, and the power transfer factor of the line relative to the load node; the load curve data include: the load value of the load node at the scheduling time; Define the decision variables of the total operating cost minimization model in the solver; the decision variables include the start and stop status of the unit and the active power generated by the unit; Convert the objective function and constraints into expressions and import them into the solver; The hardware parameters of the unit, the cost parameters of the unit, the hardware parameters of the transmission line, and the load curve data of the corresponding date are set as the parameters of the solver, and the solver is called to solve the problem so that the solver calculates the start and stop status and active power generation for each scheduling period on the corresponding date as the optimal unit combination plan for the current date.

6. The method for selecting a hot start point of a unit combination scheme according to claim 1, characterized in that: The total operating cost minimization model includes an objective function and constraints; The objective function includes: Among them, F G represents the total operating cost of the unit combination; is the start and stop status of unit g at time t on day i; is the active power generated by unit g at time t on day i; is the start and stop status of unit g at time t-1 on day i; S UD The startup and shutdown costs of the unit; is the fuel cost of unit g at time t on day i; Fuel cost is calculated by the following formula Among them, a g 、b g and c g is the fuel cost coefficient of unit g; The constraints include: system power balance constraint, generator set output constraint, generator set ramp constraint, spinning reserve capacity constraint, unit minimum downtime constraint, unit minimum operating time constraint and branch power flow constraint.

7. The method for selecting a hot start point of a unit combination scheme according to claim 6, characterized in that: The system power balance constraint is: in, is the active power generated by unit g at time t on day i; N g Assemble for the crew; is the load of load node n at time t on day i; N n is the set of load nodes; N t is the scheduling period set; The generator set output constraint is: in, is the active power generated by unit g at time t on day i; is the minimum active power generated by unit g; is the maximum active power generated by unit g, i.e. the rated capacity of unit g; The generator set climbing constraint is: in, is the active power generated by unit g at time t on day i; is the active power generated by unit g at time t-1 on day i; is the start and stop status of unit g at time t on day i; is the start and stop status of unit g at time t-1 on day i; It is the upper limit of the unit's ramp-up power increment, i.e. the maximum power that can be increased within a specific time period; It is the upper limit of the unit's g-down ramp power increment, that is, the maximum power that can be reduced within a specific time period; is the power limit when unit g starts; is the power limit when unit g is shut down; The spinning reserve capacity constraint is: in, is the start and stop status of unit g at time t on day i; is the maximum active power of unit g, that is, the rated capacity of unit g; Q s is the proportion of spinning reserve capacity, is the load of load node n at time t on day i; N n is the set of load nodes; N t is the scheduling period set; The minimum operating time constraint of the unit is: in, is the minimum operating time of unit g; is the start and stop status of unit h at time t on day i; is the start and stop status of unit g at time t-1 on day i; is the start and stop status of unit g at time k on day i; The minimum downtime constraint of the unit is: in, is the minimum downtime of unit g; is the start and stop status of unit h at time t on day i; is the start and stop status of unit g at time t-1 on day i; The branch power flow constraint is: in, is the active power limit of line l, LF lg is the power transfer factor of line l relative to unit g, LF ln is the power transfer factor of line l relative to load node n; is the active power generated by unit g at time t on day i; is the load of load node n at time t on day i; N n is the set of load nodes; N l is the set of lines; N t A set of scheduling periods.

8. A device for selecting a hot start point for a unit combination scheme, characterized in that: include: Data acquisition module, undirected graph generation module, graph similarity evaluation module and hot start point selection module; The data acquisition module is used to obtain load curve data of several historical typical days of the power grid, load curve data of the power grid operation day and the topology structure of the power grid; The undirected graph generation module is used to generate undirected graphs of several historical typical days based on the topological structure of the power grid and the load curve data of several historical typical days of the power grid; and to generate an undirected graph of the power grid operation day based on the topological structure of the power grid and the load curve data of the power grid operation day; The graph similarity assessment module is configured to sequentially input the undirected graphs of each historical typical day and the undirected graphs of the power grid operation day into a graph similarity assessment network model to generate similarity scores between the undirected graphs of the power grid operation day and the undirected graphs of each historical typical day; wherein each training sample in training the graph similarity assessment network model includes: undirected graphs of two different historical typical days and actual similarity labels between the two undirected graphs; the actual similarity of the undirected graphs of the two different historical typical days is calculated based on the optimal unit combination schemes of the two different historical typical days; the optimal unit combination scheme is the unit combination scheme with the lowest total operating cost; The hot start point selection module is used to select the undirected graph of the historical typical day with the highest similarity score as the target undirected graph; use the optimal unit combination plan of the typical day corresponding to the target undirected graph as the target unit combination plan for the power grid operation day; and use the target unit combination plan as the hot start point for solving the total operating cost minimization model.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the method for selecting a hot start point of a unit combination scheme as described in any one of claims 1 to 7 can be implemented.

10. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, it can implement the method for selecting a hot start point of a unit combination scheme as described in any one of claims 1 to 7.

Images