Solver training method and solving method for power system security constraint unit commitment model

Abstract

The invention belongs to the related technical field of power system dispatching, and discloses a solver training method and solving method for a power system security constraint unit commitment model, and the training method comprises the steps: obtaining a plurality of unit commitment models, and generating a plurality of groups of near-optimal solutions for each unit commitment model through a near-optimal solution generation method; constructing a training sample set, wherein each training sample comprises parameter characteristics and a near-optimal solution of the unit commitment model; training the graph neural network to obtain a solver; the near-optimal solution generation method comprises the following steps: adding a near-optimal constraint to a unit commitment model to obtain a near-optimal solution generation model: carrying out randomized disturbance on a coefficient of a target function and adding a diversity constraint to obtain an updated model and solving the updated model, updating a newest solution into a currently obtained solution, and adding the currently obtained solution into a solution set; and repeatedly executing the process until a preset number of near-optimal solutions exist in the solution set. And inputting the parameter characteristics of the to-be-solved unit commitment model into the solver so as to quickly predict the value of the power resource scheduling variable.

Description

Technical Field

[0001] The present invention belongs to the technical field related to power system dispatching, and more specifically, relates to a solver training method and a solving method for a power system safety constraint unit combination model. Background Art

[0002] As the scale of power systems continues to increase, independent system operators have an urgent need to solve the large-scale security-constrained unit commitment (SCUC) problem more efficiently. The purpose of independent system operators in solving the unit commitment problem is to optimize the configuration combination of power generation resources and determine the operating status and output level of power generation resources under the premise of meeting the operating conditions such as system balance constraints, reserve constraints, and network constraints, so as to minimize the total operating cost of the entire system.

[0003] At present, large-scale SCUC problems are usually modeled as mixed integer linear programming problems, and the integer programming solver is called to perform calculations using the branch and bound method. However, due to the NP-hard nature of mixed integer programming problems, it is difficult to search for the optimal solution to large-scale SCUC problems, and the amount of calculation is huge. The independent system operators generally only reserve 1-3 hours to solve large-scale SCUC problems. Therefore, how to find a high-quality approximate optimal solution in a relatively short time is of great significance for system operators to reduce system operating costs. Summary of the invention

[0004] In view of the above defects or improvement needs of the prior art, the present invention provides a solver training method and a solution method for a power system safety constraint unit combination model, the purpose of which is to find a high-quality approximate optimal solution for the power system safety constraint unit combination model in a relatively short time.

[0005] To achieve the above object, the present invention provides a solver training method for a power system safety constraint unit combination model, which comprises:

[0006] Acquire multiple safety-constrained unit commitment models of the historical power system, and for each unit commitment model, generate multiple groups of near-optimal solutions for power resource dispatching variables in the corresponding unit commitment model by using a near-optimal solution generation method;

[0007] Constructing a training sample set, wherein the input of each training sample is the parameter feature of the unit commitment model, and the label is the near-optimal solution of the corresponding unit commitment model;

[0008] Using the training sample set to train the graph neural network, a solver for predicting the values ​​of power resource scheduling variables in the unit commitment model based on the parameter characteristics of the unit commitment model is obtained;

[0009] Wherein, the near-optimal solution generation method comprises:

[0010] Step S11: Obtain the optimal solution X of the unit combination model * And the objective function value z corresponding to the optimal solution * , set the number of near-optimal solutions N sol >1 and tolerance q, add near-optimal constraints to the unit commitment model to obtain a near-optimal solution generation model, wherein the near-optimal constraint is that the value of the objective function of the unit commitment model does not exceed (1+q)z * , initialize the solution set S = {X *}, initialize the latest solution X new =X * ;

[0011] Step S12: Randomly perturb the coefficients in the objective function of the near-optimal solution generation model, and add diversity constraints to the near-optimal solution generation model to obtain an updated model, wherein the perturbation amplitude does not exceed 10% of the original coefficient value, and the diversity constraints are used to constrain the solution to be generated to be different from the current latest solution X new ;

[0012] Step S13: Solve and update the model, and convert the latest solution X new Update to the current solution, and add the current solution to the solution set S;

[0013] Step S14: Determine whether the number of solutions in the solution set S reaches N sol If not, jump to step S12; if so, output the solution set S as multiple groups of near-optimal solutions for the corresponding unit commitment model.

[0014] Preferably, the variable is an integer-valued variable.

[0015] Preferably, the variable is a binary variable;

[0016] In step S12, the diversity constraint is:

[0017]

[0018] In the formula, i is the variable number, B is the variable set, and x i is the i-th variable to be decided, For the latest solution new The solution corresponding to the i-th variable in Represents x i is the i-th variable in the variable set B and its latest solution X new The value in is 0, Represents x i is the i-th variable in the variable set B and its latest solution Xnew The value in is 1.

[0019] Preferably, the variables include startup variables, shutdown variables and state variables of the unit.

[0020] Preferably, the tolerance q does not exceed 10%.

[0021] The present invention also provides a method for solving a power system safety constraint unit combination model, which comprises:

[0022] Determining a safety constraint unit commitment model to be solved for the power system, wherein the unit commitment model has power resource dispatch variables to be decided;

[0023] Extracting parameter features of the unit commitment model and inputting them into a solver, and using the solver to predict the value of the power resource scheduling variable;

[0024] Wherein, the solver is trained by the training method as described above.

[0025] Preferably, after using the solver to predict the value of the power resource scheduling variable, the method further includes correcting the prediction result:

[0026] Classify the variables whose predicted values ​​reach the preset confidence level as credible variables;

[0027] Adding a deviation term to the objective function in the unit commitment model to obtain a modified model, wherein the deviation term=W*accumulated deviation, the accumulated deviation is the accumulation of absolute deviations between all the credible variables to be decided and the predicted values ​​of the corresponding credible variables, and W is a preset weight coefficient;

[0028] The correction model is solved to obtain the value of the power resource scheduling variable.

[0029] Preferably, after the correction is performed, a solution quality assessment step is further included:

[0030] Calculate the sum of the deviations between the solution results of all variables corresponding to any device in the power system in the correction model and the predicted values ​​of the variables in the solver to obtain the decision deviation of the corresponding device;

[0031] The equipment in the power system is divided into stable equipment and unstable equipment according to the decision deviation;

[0032] The solution results of the variables corresponding to the stable equipment are directly substituted into the unit combination model, and the unit combination model is re-solved to re-obtain the values ​​of the variables corresponding to the unstable equipment.

[0033] The present invention also provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of any of the above methods are implemented.

[0034] The present invention also provides a computer program product, comprising a computer program or instructions, wherein when the computer program or instructions are executed by a processor, the steps of any of the above methods are implemented.

[0035] In general, compared with the prior art, the above technical solutions conceived by the present invention mainly have the following beneficial effects:

[0036] 1. The present invention uses a trained neural network as a solver. As long as the parameter characteristics of the unit combination model are input into the solver, the value of the power resource scheduling variable can be quickly predicted, so that the unit combination model can be solved in a short time. Since training a neural network requires a large amount of training data, and the amount of historical data is relatively limited, in the historical data, a safety constraint unit combination model has only one optimal solution. Too little training data will lead to poor training effect of the neural network. Therefore, the present invention also mentions a near-optimal solution generation method. By adding near-optimal constraints to the unit combination model, the solution can be made to be a near-optimal solution, that is, close to the optimal solution. By adding disturbances and diversity constraints, it can be ensured that the solution obtained each time is different from the existing solution, avoiding repeated acquisition of the same near-optimal solution. In this way, multiple groups of near-optimal solutions of the unit combination model can be quickly obtained, thereby expanding the training data, improving the training effect of the neural network, and thus improving the accuracy of the solver. As long as the model parameters of the power system safety constraint unit combination model to be solved are input into the solver, the values ​​of the power resource scheduling variables in the corresponding unit combination model can be obtained. This enables fast solution of fast models.

[0037] 2. Preferably, the method proposed in the present invention is used to predict the value of an integer variable, with higher prediction accuracy and a simpler process.

[0038] 3. Preferably, after using the solver to predict the value of the variable, the present invention also performs a correction step to improve the accuracy of the obtained solution.

[0039] 4. Preferably, after executing the correction step, the present invention also executes a solution quality evaluation step to further improve the accuracy of the obtained solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 is a flowchart of the steps of a solver training method in one embodiment of the present invention;

[0041] Figure 2 Schematic diagram of the effect of a method for generating a near-optimal solution in one embodiment of the present invention;

[0042] Figure 3 is a flowchart of the steps of a method for solving a power system safety constraint unit combination model in one embodiment of the present invention;

[0043] Figure 4 is a flow chart of steps for correcting prediction results in one embodiment of the present invention;

[0044] Figure 5 is a flow chart of steps for evaluating the quality of a solution in one embodiment of the present invention;

[0045] Figure 6 It is a time-effectiveness comparison diagram between the method of the present invention and the traditional method. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0047] Example 1

[0048] like Figure 1 The figure is a flowchart of the steps of the solver training method in one embodiment of the present invention, which mainly includes steps S1 to S3. The key steps are introduced below.

[0049] Step S1: obtaining multiple safety-constrained unit combination models of the historical power system, and for each unit combination model, generating multiple groups of near-optimal solutions for power resource dispatching variables in the corresponding unit combination model by a near-optimal solution generation method.

[0050] Since training a neural network requires a large amount of training data, and the amount of historical data is relatively limited, in the historical data, a safety constraint unit combination model has only one optimal solution. Too little training data will lead to poor training results of the neural network. Therefore, this step is mainly to achieve data enhancement of the training data, generate multiple groups of near-optimal solutions for each safety constraint unit combination model, and thus expand the training data.

[0051] Among them, multiple groups of near-optimal solutions for power resource dispatching variables in the corresponding unit combination model are generated by a near-optimal solution generation method, including steps S11 to S14, and the key steps are introduced below.

[0052] Step S11: Obtain the optimal solution X of the unit combination model * And the objective function value z corresponding to the optimal solution * , set the number of near-optimal solutions N sol>1 and tolerance q, add near-optimal constraints to the unit commitment model to obtain a near-optimal solution generation model. The near-optimal constraint is that the value of the objective function of the unit commitment model does not exceed (1+q)z * , initialize the solution set S = {X *}, initialize the latest solution X new =X * .

[0053] The unit combination model and the optimal solution of the unit combination model can be directly obtained from historical data. * , and can also obtain or calculate the function value z of the objective function when the unit combination model takes the optimal solution * .

[0054] A general mode of the unit combination model is given below, but it is not limited to this:

[0055]

[0056] In the formula, u g,t ,s g,t ,d g,t They represent the state variable, startup variable and shutdown variable of unit g at time t respectively. g,t represents the output level of unit g at time t, g(·) is the system-level constraint of the unit commitment model, h(·) is the single-machine constraint of the unit commitment model, and f(·) is the objective function.

[0057] Construct solution set S and latest solution X new , when initialized, the solution set S only contains the optimal solution X * , and the optimal solution X * As the latest solution at the initial moment new .

[0058] Set the number of near-optimal solutions N as needed sol , for example, N sol You can take 10.

[0059] The tolerance q is set as needed, which reflects the degree of deviation of the expanded data. The larger the tolerance q, the higher the generation speed but the lower the quality of the near-optimal solution. The smaller the tolerance q, the lower the generation speed but the higher the quality of the near-optimal solution. In this embodiment, q=0.5%.

[0060] Add near-optimal constraints to the unit commitment model to obtain a near-optimal solution generation model. The near-optimal constraint is that the value of the objective function of the unit commitment model does not exceed (1+q)z * , the quality of the expanded near-optimal solution is guaranteed by the near-optimal constraint. In a specific implementation, the near-optimal constraint can be specifically expressed as:

[0061] f(ug,t ,s g,t ,d g,t ,p g,t )≤(1+q)z *

[0062] Assuming that the original constraint set of the unit commitment model is set C, after adding the near-optimal constraints, the constraint set of the near-optimal solution generation model can be expressed as:

[0063] C∪{f(u g,t ,s g,t ,d g,t ,p g,t )≤(1+q)z *}

[0064] Step S12: Randomly perturb the coefficients in the objective function of the near-optimal solution generation model, and add diversity constraints to the near-optimal solution generation model to obtain an updated model, wherein the perturbation amplitude does not exceed 10% of the original coefficient value, and the diversity constraint is used to constrain the solution to be generated to be different from the current latest solution X new .

[0065] Among them, the objective function coefficients of the near-optimal solution generation model are randomly perturbed. For example, the objective function f(u g,t ,s g,t ,d g,t ,p g,t ) each variable has a corresponding coefficient. If the objective function is a cost function, the coefficient is actually a unit price. The coefficient is randomly perturbed, and the perturbation amplitude does not exceed 10% of the original coefficient value. In this embodiment, the perturbation interval can be 0.95 to 1.05 times the original coefficient value. The objective function is perturbed in each cycle, the purpose of which is to change the optimization direction and improve the diversity of the generated near-optimal solutions.

[0066] Add diversity constraints to the near-optimal solution generation model. Diversity constraints are used to constrain the solution to be generated to be different from the current latest solution X new , to avoid repeatedly generating the same solution.

[0067] In one embodiment, the power resource scheduling variable targeted by the method is an integer variable. Compared with continuous time variables, the method is more suitable for integer variables and is more accurate in predicting integer variables. Specifically, the integer variable can be, for example, a startup variable, a shutdown variable, and a state variable of a unit. In each prediction period, the predicted value of the integer variable is a constant value.

[0068] Furthermore, the integer variable is a binary variable, which can only be 0 or 1. For example, for the power-on variable, 1 indicates that the power-on action is executed, and 0 indicates that the power-on action is not executed. At this time, the multi-style constraint can be expressed as:

[0069]

[0070] In the formula, i is the variable number, B is the variable set, and x i is the i-th variable to be decided, For the latest solution new The solution corresponding to the i-th variable in Represents x i is the i-th variable in the variable set B and its latest solution X new The value in is 0, Represents x i is the i-th variable in the variable set B and its latest solution X new The value in is 1.

[0071] Step S13: Solve and update the model, and convert the latest solution X new Update to the current solution and add the current solution to the solution set S.

[0072] Specifically, the update model is solved. When a solution is found, it is stored in the solution set S as a near-optimal solution. At this time, the number of solutions stored in the solution set S is accumulated by 1, and the latest solution X is also added. new Update to the current solution.

[0073] Step S14: Determine whether the number of solutions in the solution set S reaches N sol If not, jump to step S12; if so, output the solution set S as multiple groups of near-optimal solutions for the corresponding unit commitment model.

[0074] Specifically, steps S12 to S13 are repeated until the solutions in the solution set S meet the quantity requirement. At this time, a unit commitment model is expanded from a corresponding set of solutions to corresponding sets of solutions, which can be specifically expressed as:

[0075]

[0076] In the formula, s q is the qth unit combination model, It is the p-th near-optimal solution of the q-th unit combination model.

[0077] like Figure 2 The figure is a schematic diagram of the effect of the near-optimal solution generation method provided by an embodiment of the present invention. For each unit combination model, 25 near-optimal solutions can be quickly generated in a short time through the above-mentioned near-optimal solution generation method, which greatly reduces the time spent on finding the optimal solution.

[0078] Step S2: construct a training sample set, where the input of each training sample is the parameter feature of the unit combination model, and the label is the near-optimal solution of the corresponding unit combination model.

[0079] Specifically, a unit combination model and one of its solutions can construct a training sample, and a unit combination model can construct multiple training samples to extract parameter features of the unit combination model. For example, when the prediction variable is the unit start-stop state variable, the input parameter features are preferably parameters closely related to the unit start-stop state, such as the maximum / minimum technical output of the thermal power unit, the load power curve and other related parameters. With the parameter features as input and the near-optimal solution of the unit combination model as a label, a training sample can be constructed, and in this way, a training sample set is constructed.

[0080] Step S3: Use the training sample set to train the graph neural network to obtain a solver that predicts the values ​​of power resource scheduling variables in the unit combination model based on the parameter characteristics of the unit combination model.

[0081] Specifically, the graph neural network can select an existing neural network, such as a convolutional neural network.

[0082] The parameter features of the unit commitment model in the training sample are input into the graph neural network, the deviation between the prediction result and the label is calculated through the loss function, and the parameters of the graph neural network are adjusted through back propagation until the loss function converges and the training is completed. A solver for predicting the values ​​of power resource scheduling variables in the unit commitment model based on the parameter features of the unit commitment model is obtained.

[0083] Taking binary variables as an example, the loss function can be designed as:

[0084]

[0085] in:

[0086]

[0087] U1={x i |x i ∈X,x i =0}

[0088] U2={x i |x i ∈X,x i =1}

[0089] In the formula, S is the solution set of a unit commitment model, X is a set of solutions in the solution set, and x i is the solution of the i-th variable in X, U1 represents the set of variables whose values ​​are 0 in the solution set, U2 represents the set of variables whose values ​​are 1 in the solution set, and P i The variable x predicted by the graph neural network i The value of , after passing through the sigmoid activation function, must be between [0,1].

[0090] Through the above training method, a solver for solving the power system security constraint unit combination model can be trained. As long as the model parameters of the power system security constraint unit combination model to be solved are input into the solver, the values ​​of the power resource dispatch variables in the corresponding unit combination model can be obtained, thereby realizing the rapid solution of the fast model.

[0091] Example 2

[0092] The present invention also proposes a method for solving the power system safety constraint unit combination model, such as Figure 3 As shown, the method includes:

[0093] Step S4: determining a safety constraint unit combination model to be solved for the power system, wherein the unit combination model has power resource dispatching variables;

[0094] Step S5: extracting parameter features of the safety constraint unit commitment model and inputting them into a solver, and using the solver to predict the values ​​of power resource dispatch variables;

[0095] The solver is trained by the training method in Example 1.

[0096] Based on step S4 and step S5, the unit combination model can be solved to obtain the value of the power resource scheduling variable, and the power resource scheduling can be achieved based on the value.

[0097] Considering that the prediction result of the solver may be biased, in one embodiment, after the solver continues to make predictions, the prediction result is further corrected, that is, after step S5, the following steps are further performed:

[0098] Step S6: Correct the prediction result.

[0099] like Figure 4 As shown, the correction includes the following steps.

[0100] Step S61: Classify the variables whose predicted values ​​have confidence levels that reach a preset level of credibility as credible variables.

[0101] The solver is actually a trained neural network. It can be understood that the output end of the neural network has multiple channels. The i-th channel corresponds to the predicted value of the i-th variable and its confidence. The credibility of the predicted value of the variable can be evaluated based on the confidence.

[0102] For example, for binary variables, the confidence standard can be unified, that is, the confidence of all channel outputs is 1. If the confidence of 1 is very low, it means that its predicted value is 0. More specifically, a first confidence threshold α1 and a second confidence threshold α2 can be set. The first confidence threshold α1 is greater than or equal to 0.8, and the second confidence threshold α2 is less than or equal to 0.2. For example, the first confidence threshold α1 can be 0.9, and the second confidence threshold α2 can be 0.1. Variables whose confidence of predicted values ​​is greater than or equal to the first confidence threshold α1 can be classified into the first set; variables whose confidence of predicted values ​​is less than or equal to the second confidence threshold α2 can be classified into the second set; and the remaining variables are classified into the third set, which can be specifically expressed as follows:

[0103]

[0104] In the formula, J is the set of numbers of all variables, h i For variable x i Taking a confidence level of 1, J1 represents the set of numbers of variables whose predicted values ​​are 1 and whose predicted results are reliable, J0 represents the set of numbers of variables whose predicted values ​​are 0 and whose predicted results are reliable, J represents the set of numbers of variables whose predicted results are unreliable, and J0∪J1 represents the set of numbers of all variables whose predicted results are reliable (i.e., credible variables).

[0105] Step S62: Add a deviation term to the objective function in the unit commitment model to obtain a modified model, where the deviation term = W* cumulative deviation, where the cumulative deviation is the accumulation of absolute deviations between all credible variables to be decided and the predicted values ​​of the corresponding credible variables, and W is a preset weight coefficient.

[0106] Continuing with the above example, at this point, the deviation term can be expressed as:

[0107]

[0108] By adjusting the objective function in the original unit commitment model, the following modified model is obtained:

[0109]

[0110] Step S63: Solve the modified model to obtain the value of the power resource scheduling variable.

[0111] It should be noted that the weight coefficient W can be adjusted flexibly. For example, W can take a larger number so that the deviation term has a larger proportion than the original objective function.

[0112] In some embodiments, considering that the obtained solution may still be of low quality, in order to further improve the quality of the solution, further, after step S6, the following steps are performed:

[0113] Step S7: Evaluate the quality of the solution.

[0114] like Figure 5 As shown, evaluating the quality of the solution includes the following steps.

[0115] Step S71: Calculate the sum of the deviations between the solution results of all variables corresponding to any device in the power system in the modified model and their predicted values ​​in the solver to obtain the decision deviation of the corresponding device.

[0116] For example, for a generator set, which includes startup variables, shutdown variables and state variables, the previous method can be used to obtain the solution results of each variable in the modified model and the predicted value obtained by the solver. The deviation between the solution result and the predicted value of each variable of the device is calculated and accumulated to obtain the decision deviation of the device.

[0117] For example, the decision deviation of device g can be specifically expressed as:

[0118]

[0119] In the formula, x j is the solution result of the j-th variable of device g in solving the modified model, is the predicted value of the j-th variable of device g in the solver.

[0120] Step S72: Divide the equipment in the power system into stable equipment and unstable equipment according to the decision deviation.

[0121] Step S73: directly substitute the solution results of the variables corresponding to the stable equipment into the unit combination model, and re-solve the unit combination model to re-obtain the values ​​of the variables corresponding to the unstable equipment.

[0122] Specifically, the larger the m(g), the more unstable the variable value corresponding to the device is, and re-optimization is needed. At this time, the device with a decision deviation greater than the threshold can be re-solved. The solution process is the same as above and will not be repeated here.

[0123] like Figure 6 The figure shows a time efficiency comparison diagram between the method of the present invention and the traditional method, wherein the numbers outside the circle represent the serial numbers of the unit combination models, and the divergent length along the radial direction starting from the center of the circle is the solution time. It can be seen from the figure that the solution time using the solution method proposed by the present invention is greatly reduced. Figure 6 A relatively simple unit combination model is selected. The unit combination model used in actual engineering is very large. The time efficiency advantage of using the present invention for solving is more obvious.

[0124] Example 3

[0125] The present invention also relates to a computer-readable storage medium having a computer program stored thereon, and the computer program implements the steps of the above method when executed by a processor.

[0126] Specifically, the memory may include a high-speed random access memory, and may also include a non-volatile memory, such as a hard disk, an internal memory, a plug-in hard disk, a smart media card (SMC), a 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 devices.

[0127] Example 4

[0128] The embodiment of the present invention provides a computer program product or a computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device performs the steps of the method of the above embodiment of the present invention.

[0129] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification. It should be noted that "in one embodiment", "for example", "for example", etc. of the present invention are intended to illustrate the present invention, rather than to limit the present invention.

[0130] The above embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the patent application. It should be pointed out that, for those of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention.

Claims

1. A solver training method for a power system safety constraint unit combination model, characterized in that: include: Acquire multiple safety-constrained unit commitment models of the historical power system, and for each unit commitment model, generate multiple groups of near-optimal solutions for power resource dispatching variables in the corresponding unit commitment model by using a near-optimal solution generation method; Constructing a training sample set, wherein the input of each training sample is the parameter feature of the unit commitment model, and the label is the near-optimal solution of the corresponding unit commitment model; Using the training sample set to train the graph neural network, a solver for predicting the values ​​of power resource scheduling variables in the unit commitment model based on the parameter characteristics of the unit commitment model is obtained; Wherein, the near-optimal solution generation method comprises: Step S11: Obtain the optimal solution X of the unit combination model * And the objective function value z corresponding to the optimal solution * , set the number of near-optimal solutions N sol >1 and tolerance q, add near-optimal constraints to the unit commitment model to obtain a near-optimal solution generation model, wherein the near-optimal constraint is that the value of the objective function of the unit commitment model does not exceed (1+q)z * , initialize the solution set S = {X * }, initialize the latest solution X new =X * ; Step S12: Randomly perturb the coefficients in the objective function of the near-optimal solution generation model, and add diversity constraints to the near-optimal solution generation model to obtain an updated model, wherein the perturbation amplitude does not exceed 10% of the original coefficient value, and the diversity constraints are used to constrain the solution to be generated to be different from the current latest solution X new ; Step S13: Solve and update the model, and convert the latest solution X new Update to the current solution, and add the current solution to the solution set S; Step S14: Determine whether the number of solutions in the solution set S reaches N sol If not, jump to step S12; if so, output the solution set S as multiple groups of near-optimal solutions for the corresponding unit commitment model.

2. The solver training method for the power system safety constraint unit combination model according to claim 1, characterized in that: The variable is an integer-valued variable.

3. The solver training method for the power system safety constraint unit combination model according to claim 1, characterized in that: The variable is a binary variable; In step S12, the diversity constraint is: In the formula, i is the variable number, B is the variable set, and x i is the i-th variable to be decided, For the latest solution new The solution corresponding to the i-th variable in Represents x i is the i-th variable in the variable set B and its latest solution X new The value of is 0, Represents x i is the i-th variable in the variable set B and its latest solution X new The value in is 1.

4. The solver training method for the power system safety constraint unit combination model according to claim 2 or 3, characterized in that: The variables include startup variables, shutdown variables and state variables of the unit.

5. The solver training method for the power system safety constraint unit combination model according to claim 1, characterized in that: The tolerance q does not exceed 10%.

6. A method for solving a power system safety constraint unit combination model, characterized in that: include: Determining a safety constraint unit commitment model to be solved for the power system, wherein the unit commitment model has power resource dispatch variables to be decided; Extracting parameter features of the unit commitment model and inputting them into a solver, and using the solver to predict the value of the power resource scheduling variable; Wherein, the solver is trained by the training method according to any one of claims 1 to 5.

7. The method for solving the power system safety constraint unit combination model according to claim 6, characterized in that: After using the solver to predict the value of the power resource scheduling variable, the prediction result is also corrected: Classify the variables whose predicted values ​​reach the preset confidence level as credible variables; Adding a deviation term to the objective function in the unit commitment model to obtain a modified model, wherein the deviation term=preset weight coefficient×accumulated deviation, and the accumulated deviation is the accumulation of absolute deviations between all the credible variables to be decided and the predicted values ​​of the corresponding credible variables; The correction model is solved to obtain the value of the power resource scheduling variable.

8. The method for solving the power system safety constraint unit combination model according to claim 7, characterized in that: After making the corrections, a solution quality assessment step is also included: Calculate the sum of the deviations between the solution results of all variables corresponding to any device in the power system in the correction model and the predicted values ​​of the variables in the solver to obtain the decision deviation of the corresponding device; The equipment in the power system is divided into stable equipment and unstable equipment according to the decision deviation; The solution results of the variables corresponding to the stable equipment are directly substituted into the unit combination model, and the unit combination model is re-solved to re-obtain the values ​​of the variables corresponding to the unstable equipment.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

10. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 8 are implemented.

Images