A Robust Unit Commitment Accelerated Optimization Method for Power Systems Based on Representative Scenarios

Abstract

The present invention belongs to the technical field of power systems, and specifically discloses a robust unit commitment acceleration optimization method based on representative scenarios, which includes the following steps: constructing a robust uncertainty set according to historical predicted power generation data and historical actual power generation data; constructing a unit commitment optimization model; calculating the system net load and system net load rate of the power system for each time period; respectively constructing a first auxiliary mixed-integer linear programming model and a second auxiliary mixed-integer linear programming model to determine the minimum net load representative scenario and the maximum net load representative scenario; determining the set of must-run units; determining the set of must-stop units; using the set of must-run units and the set of must-stop units as the optimization plan, and constructing a robust unit commitment reduction model to achieve the acceleration optimization of the robust unit commitment problem of the power system. The present invention can improve the calculation efficiency of the robust unit commitment problem and solve the problem of low calculation efficiency of the robust unit commitment problem in the prior art.

Description

Technical Field

[0001] The invention belongs to the technical field of power systems, and particularly relates to a method for accelerating the optimization of a robust unit commitment in a power system based on representative scenarios. Background Art

[0002] The security-constrained unit commitment (SCUC) problem is a typical non-convex, large-scale mixed-integer optimization problem, which contains a large number of binary variables and continuous variables and is restricted by a series of equality and inequality constraints. The common solution method for this problem is the branch-and-bound method, which searches for the optimal solution by gradually constructing a solution space tree and pruning. However, since the worst-case computational complexity of the SCUC problem is exponential with respect to the number of binary variables, the branch-and-bound method still faces great computational pressure in the face of large-scale power systems. Therefore, recent research on accelerating the optimization method of security-constrained unit commitment mainly focuses on how to make the branch-and-bound method search for feasible solutions and even optimal solutions more efficiently. The first type of method is to identify infeasible solutions in the solution space of the branch-and-bound method through methods such as neighborhood search, internal information induction of the solver, and reinforcement learning, avoiding unnecessary searches, thereby improving the search efficiency. The second type of method is to reduce the model scale through methods such as unit aggregation and time period aggregation, reduce the decision space to be searched, and reduce the computational burden of the branch-and-bound method. The third type of method directly generates feasible solutions through data-driven or heuristic methods, or uses them as initial heat starts to improve the search efficiency of the branch-and-bound method. The fourth type of method is to repair the feasible solutions or use them to identify the must-run units and must-stop units after parallelly solving multiple small-scale optimization problems, thereby improving the search efficiency of the branch-and-bound method.

[0003] However, with the increasing access of new energy to the power system, the safe operation of the power system requires effective uncertainty management means to ensure. Robust optimization provides a promising solution for new energy uncertainty management, but the two-stage robust unit commitment problem is more complex than the previously mentioned security-constrained unit commitment problem, with long computational time-consuming, and often cannot meet the daily operation and production requirements, thus limiting its application in large-scale systems. The previously mentioned security-constrained unit commitment acceleration methods are not targeted acceleration methods for the two-stage robust unit commitment problem. Therefore, there is still much room for improvement in the two-stage robust unit commitment acceleration optimization method. Summary of the Invention

[0004] In view of the above deficiencies in the prior art, a method for accelerating the optimization of a robust unit commitment in a power system based on representative scenarios provided by the present invention solves the problem of low computational efficiency of the robust unit commitment problem in the prior art.

[0005] To achieve the above invention purpose, the technical solution adopted by the present invention is: a method for accelerating the optimization of a robust unit commitment in a power system based on representative scenarios, including the following steps:

[0006] S1. Obtain the historical predicted power generation data and historical actual power generation data of the new energy power station, and construct a robust uncertainty set based on the historical predicted power generation data and historical actual power generation data;

[0007] S2. Obtain the day-ahead optimal scheduling data of the power system, and construct a unit commitment optimization model;

[0008] S3. Set the new energy prediction deviation to 0, and calculate the system net load and system net load rate of each time period according to the day-ahead predicted output data of the new energy power station and the day-ahead load prediction data of each load node in the power system;

[0009] S4. Considering the new energy prediction deviation within the limit range of the robust uncertainty set, use the objective function to construct the first auxiliary mixed-integer linear programming model and the second auxiliary mixed-integer linear programming model respectively, and substitute the calculation results of the system net load and system net load rate into the first auxiliary mixed-integer linear programming model and the second auxiliary mixed-integer linear programming model respectively to determine the minimum net load representative scenario and the maximum net load representative scenario;

[0010] Among them, the first auxiliary mixed-integer linear programming model is an auxiliary mixed-integer linear programming model that minimizes the product sum of the reciprocal of the system net load rate and the system net load, and the second auxiliary mixed-integer linear programming model is an auxiliary mixed-integer linear programming model that maximizes the product sum of the system net load rate and the system net load;

[0011] S5. Substitute the minimum net load representative scenario into the unit commitment optimization model for solution, and set the discriminant threshold of the must-run units according to the minimum and maximum technical output parameters of the units participating in the optimization to determine the set of must-run units;

[0012] S6. Substitute the maximum net load representative scenario and the set of must-run units into the unit commitment optimization model for solution, and determine the set of must-stop units according to the must-stop unit discriminant rule, combining the solution result and the set of must-run units;

[0013] S7. Take the set of must-run units and the set of must-stop units as the optimization plan, and combine the robust uncertainty set and the unit commitment optimization model to construct a robust unit commitment reduction model to achieve accelerated optimization of the robust unit commitment problem of the power system.

[0014] The beneficial effect of the above solution is that without significantly reducing the robustness of the optimization result, by quickly solving the unit commitment model corresponding to the representative scenario, accurately determining the set of must-run units and the set of must-stop units, and then effectively reducing the number of binary variables to be decided, thereby accelerating the solution process of the robust unit commitment problem.

[0015] Further, in S1, a robust uncertainty set is constructed based on historical predicted power generation data and historical actual power generation data, specifically including:

[0016] S11. Calculate the output prediction deviation based on historical predicted power generation data and historical actual power generation data;

[0017] The formula for calculating the output prediction deviation is:

[0018]

[0019] where represents the output prediction deviation of the new energy power station at time period on the day, represents the actual power generation data of the new energy power station at time period on the day, represents the predicted power generation data of the new energy power station at time period on the day,

[0020] represents the set of new energy power stations,

[0021] represents the set of historical data days, represents the set of scheduling time periods;

[0022]

[0023]

[0024]

[0025] where represents the output prediction deviation of the new energy power station within the robust uncertainty set at time period , represents the average value of the output prediction deviation of the new energy power station at time period within days, represents the standard deviation of the output prediction deviation of the new energy power station Represents the conservatism control coefficient, and the default value is 1.5.

[0026] The beneficial effects of the above further solution are as follows: By defining the output prediction deviation range of each time period of the new energy power station within the robust uncertainty set, it is ensured that when the subsequent robust unit commitment model copes with the uncertainty of new energy power generation, it can not only ensure the safe operation of the system, but also avoid excessive conservatism, thus achieving the balance between robustness and economy.

[0027] Furthermore, in S2, the day-ahead optimal scheduling data includes power grid topology data, cost data of units participating in optimization, operation characteristic data of units participating in optimization, day-ahead load prediction data, and day-ahead predicted output data of new energy power stations;

[0028] Construct a unit commitment optimization model, specifically including:

[0029] Construct a unit commitment optimization model and the corresponding constraint conditions of the unit commitment optimization model;

[0030] The constraint conditions corresponding to the unit commitment optimization model include upper and lower limits of unit output constraints, unit start-stop status constraints, minimum continuous start-stop constraints of units, unit ramp rate constraints, system supply-demand balance constraints, line safety constraints, and new energy power station curtailment constraints;

[0031] The unit commitment optimization model is:

[0032]

[0033] Among them, Represents the total operating cost, Represents the unit participating in optimization In time period The operating status, Represents the no-load cost of the unit participating in optimization, Represents the unit In time period Whether a startup event occurs, Represents the startup cost of the startup event, Represents the unit In time period Whether a shutdown event occurs, Represents the shutdown cost of the shutdown event, Represents time period Unit The output level, Represents the unit The variable power generation cost function, Represents the new energy curtailment penalty cost, Represents the new energy power station In time period The curtailment amount, Denote the set of units participating in the optimization;

[0034] The upper and lower limits of unit output are constrained as:

[0035]

[0036] Wherein, and respectively represent the minimum technical output and the maximum technical output of the unit;

[0037] The start-stop state constraint of the unit is:

[0038]

[0039]

[0040] Wherein, represents the operating state of the unit participating in the optimization at time period

[0041] The minimum continuous start-stop constraint of the unit is:

[0042]

[0043]

[0044] Wherein, represents whether the unit has a startup event at time period represents whether the unit has a shutdown event at time period , represents the minimum continuous startup time of the unit and represents the minimum continuous shutdown time of the unit;

[0045] The ramp rate constraint of the unit is:

[0046]

[0047] Wherein, and respectively represent the maximum down-ramp rate and the maximum up-ramp rate of the unit, represents the output level of the unit at time period ;

[0048] The system supply-demand balance constraint is:

[0049] ​

[0050] Among them, represents the day-ahead predicted output of the new energy power station in the time period ; represents the predicted deviation between the day-ahead predicted output of the new energy power station in the time period and the actual output of the next day; represents the node in the time period of the day-ahead predicted load value; represents the set of load nodes;

[0051] The line security constraint is:

[0052]

[0053]

[0054] Among them, , and respectively represent the power transfer distribution factors of the unit , the new energy power station and the node to the line ; represents the steady-state power flow limit value of the line ; represents the set of transmission lines;

[0055] The new energy power station curtailment constraint is:

[0056]

[0057] Among them, represents the curtailment amount of the new energy power station in the time period .

[0058] The beneficial effect of the above further solution is that by introducing a unit commitment model containing new energy prediction deviation and curtailment amount, as the basis for constructing a subsequent robust unit commitment model, it helps to optimize the unit start-stop plan and power generation plan of the power grid when considering the uncertainty of new energy power generation.

[0059] Furthermore, in S3, according to the day-ahead predicted output data of the new energy power station and the day-ahead load prediction data of each load node in the power system, calculate the system net load and system net load rate of the power system in each time period, and the formula used is:

[0060]

[0061]

[0062] Among them, represents the system net load, represents the system net load rate.

[0063] The beneficial effect of the above further solution is that the calculation result of the system net load rate can be used as the weight in the objective functions of the subsequent first auxiliary mixed-integer linear programming model and the second auxiliary mixed-integer linear programming model, and then can guide the two small-scale optimization models to focus on the periods with smaller and larger system net loads respectively, so as to effectively screen out representative scenarios.

[0064] Further, in S4, the first auxiliary mixed-integer linear programming model is:

[0065]

[0066]

[0067]

[0068] Among them, represents the system net load variable, represents the system net load rate. After solving the first auxiliary mixed-integer linear programming model, a set of optimal solutions can be obtained, which is the representative scenario of the minimum net load;

[0069] The second auxiliary mixed-integer linear programming model is:

[0070]

[0071]

[0072]

[0073] Among them, after solving the second auxiliary mixed-integer linear programming model, a set of optimal solutions can be obtained, which is the representative scenario of the maximum net load.

[0074] The beneficial effect of the above further solution is that through two small-scale optimization problems, representative scenarios with smaller and larger system net loads are quickly screened out within the robust uncertainty set, providing an effective basis for the analysis of the must-start and must-stop units in the subsequent robust unit commitment model.

[0075] Further, S5 specifically includes:

[0076] S51. Substitute the minimum net load representative scenario into the unit commitment optimization model to obtain a new unit commitment optimization model;

[0077] The new unit commitment optimization model is:

[0078]

[0079] Wherein, represents the minimum net load representative scenario;

[0080] S52. Solve the new unit commitment optimization model to obtain the optimal solutions of the start-up modes and the first output levels of the first set of units participating in the optimization;

[0081] S53. Use the must-start unit discrimination threshold to determine the must-start unit set from the optimal solutions of the start-up modes and the first output levels of the first set of units.

[0082] The beneficial effects of the above further solution are as follows: By compressing the search space of the complex robust unit commitment model into a single representative scenario, the optimal solutions of the start-up modes and the first output levels of the first set of units can be quickly obtained, and combined with the discrimination threshold set by the minimum / maximum technical output, the must-start units that still need to be started up all the time when the system net load is low can be accurately identified.

[0083] Further, S6 specifically includes:

[0084] S61. Substitute the maximum net load representative scenario into the unit commitment optimization model to obtain a second unit commitment optimization model;

[0085] The second unit commitment optimization model is:

[0086]

[0087] Wherein, represents the maximum net load representative scenario;

[0088] S62. Substitute the must-start unit set into the second unit commitment optimization model to obtain a third unit commitment optimization model;

[0089] S63. Solve the third unit commitment optimization model to obtain the optimal solutions of the start-up modes and the second output levels of the second set of units participating in the optimization;

[0090] S64. According to the must-stop unit discrimination rule, combine the optimal solutions of the start-up modes and the second output levels of the second set of units and the must-start unit set to determine the must-stop unit set.

[0091] The beneficial effects of the above further solution are as follows: By compressing the search space of the complex robust unit commitment model into a single representative scenario and combining the determined set of must-run units, the solution time can be effectively reduced, and the optimal solution of the second unit's start-up mode and output level can be obtained quickly. At the same time, through the must-stop unit discrimination rule, the must-stop units that will not start up even when the system net load is high can be accurately identified.

[0092] Further, S7 specifically includes:

[0093] S71. Combine the robust uncertainty set and the unit commitment optimization model to construct a robust unit commitment model;

[0094] The robust unit commitment model is:

[0095]

[0096]

[0097] Among them, , , , , , the vector is the first-stage decision variable, the vector is the second-stage decision variable, the vector is the uncertain parameter restricted by the robust uncertainty set , that is, the new energy prediction deviation, is the value function about the vector and the vector , is the parameter vector of the no-load cost, start-up cost and shutdown cost of the units participating in the optimization, is the parameter vector of the variable cost of the units participating in the optimization and the curtailment penalty cost parameter vector of the new energy power station. The matrix and the vector are the coefficient matrix and the constant vector of the first-stage independent inequality constraints respectively, corresponding to the unit start-stop state constraint and the unit minimum continuous start-stop constraint in the unit commitment optimization model. The matrix and the vector are the coefficient matrix and the constant vector of the second-stage independent inequality constraints respectively, corresponding to the line security constraint and the new energy power station curtailment constraint in the unit commitment optimization model. The matrix , the matrix and the vector are the two coefficient matrices and the constant vector of the first-stage and second-stage coupled inequality constraints respectively, corresponding to the unit output upper and lower limit constraint and the unit ramp rate constraint in the unit commitment optimization model; the matrix , the matrix , the matrix Sum vector are the three coefficient matrices and the constant vector of the equality constraints coupling the second stage and the robust uncertainty set respectively, corresponding to the system supply-demand balance constraint in the unit commitment optimization model;

[0098] S72. Take the set of must-run units and the set of must-stop units as additional constraints to obtain a reduced robust unit commitment model;

[0099] The reduced robust unit commitment model is:

[0100]

[0101]

[0102] Among them, represents the vector composed of the first-stage decision variables in the robust unit commitment model in the th variable.

[0103] The beneficial effect of the above further solution is that by taking the set of must-run units and the set of must-stop units as additional constraints to directly fix the determined unit operating states, the binary variables to be determined in the robust unit commitment model can be effectively reduced, and the solution efficiency can be improved. Description of the Drawings

[0104] Figure 1 Shown is a schematic flow chart of a method for accelerating the optimization of a robust unit commitment in a power system based on representative scenarios. Detailed Embodiments

[0105] Now, exemplary embodiments of the present invention will be described in detail with reference to the drawings. It should be understood that the embodiments shown and described in the drawings are merely exemplary, intended to illustrate the principles and spirit of the present invention, and not to limit the scope of the present invention.

[0106] As Figure 1 shown, a method for accelerating the optimization of a robust unit commitment in a power system based on representative scenarios includes the following steps:

[0107] S1. Obtain the historical predicted power generation data and historical actual power generation data of the new energy power station, and construct a robust uncertainty set based on the historical predicted power generation data and historical actual power generation data.

[0108] In this embodiment, constructing a robust uncertainty set based on the historical predicted power generation data and historical actual power generation data specifically includes:

[0109] S11. Calculate the power prediction deviation according to the historical predicted power generation data and historical actual power generation data;

[0110] The formula for calculating the output prediction deviation is as follows:

[0111]

[0112] Among them, represents the output prediction deviation of the new energy power station on the day and time period . represents the actual power generation data of the new energy power station on the day and time period . represents the predicted power generation data of the new energy power station on the day and time period . represents the set of new energy power stations, represents the set of historical data days, represents the set of dispatching time periods;

[0113] S12. Construct a robust uncertainty set according to the mean and standard deviation of the output prediction deviation;

[0114] The formula for constructing the robust uncertainty set is as follows:

[0115]

[0116]

[0117]

[0118] Among them, represents the output prediction deviation of the new energy power station within the robust uncertainty set during the time period , and its range is , represents the mean of the output prediction deviation of the new energy power station during the time period within days, represents the standard deviation of the output prediction deviation of the new energy power station during the time period within days, represents the conservatism control coefficient, and the default value is 1.5.

[0119] Exemplarily, the new energy power station may include a photovoltaic power station, a wind power station, etc.

[0120] Optionally, the data interval of the historical predicted power generation data and the historical actual power generation data is 1 hour or 15 minutes. In the case of missing historical data, ultra-short-term prediction data can be used for substitution.

[0121] S2. Obtain the day-ahead optimal scheduling data of the power system and construct a unit commitment optimization model.

[0122] In this embodiment, the day-ahead optimal scheduling data includes power grid topology data, cost data of units participating in optimization, operation characteristic data of units participating in optimization, day-ahead load prediction data, and day-ahead predicted output data of new energy power stations;

[0123] Constructing the unit commitment optimization model specifically includes:

[0124] Construct the unit commitment optimization model and the corresponding constraint conditions of the unit commitment optimization model;

[0125] The constraint conditions corresponding to the unit commitment optimization model include upper and lower limits of unit output constraints, unit start-stop status constraints, minimum continuous start-stop constraints of units, unit ramp rate constraints, system supply-demand balance constraints, line safety constraints, and new energy power station curtailment constraints;

[0126] The unit commitment optimization model is:

[0127]

[0128] Wherein, represents the total operating cost, represents the unit participating in optimization at time period operating status, represents the no-load cost of the unit participating in optimization, represents the unit at time period whether a start-up event occurs, represents the start-up cost of the start-up event, represents the unit at time period whether a shutdown event occurs, represents the shutdown cost of the shutdown event, represents time period unit output level, represents the unit variable power generation cost function, represents the new energy curtailment penalty cost, represents the new energy power station at time period curtailment amount, represents the set of units participating in optimization;

[0129] The upper and lower limits of the unit output are constrained as follows:

[0130]

[0131] Among them, and respectively represent the minimum technical output and the maximum technical output of the unit;

[0132] The start-stop state constraint of the unit is:

[0133]

[0134]

[0135] Among them, represents the operating state of the unit at time period ;

[0136] The minimum continuous start-stop constraint of the unit is:

[0137]

[0138]

[0139] Among them, represents whether the unit has a startup event at time period , represents whether the unit has a shutdown event at time period , , represents the minimum continuous startup time of the unit , represents the minimum continuous shutdown time of the unit ;

[0140] The ramp rate constraint of the unit is:

[0141]

[0142] Among them, and respectively represent the maximum down-ramp rate and the maximum up-ramp rate of the unit, represents the output level of the unit at time period ;

[0143] The system supply-demand balance constraint is:

[0144]

[0145] Among them, represents a new energy power station at time period the day-ahead predicted output power, represents a new energy power station at time period the predicted deviation between the day-ahead predicted output power and the actual output power of the next day, represents node at time period the day-ahead predicted load value, represents the set of load nodes;

[0146] The line security constraint is:

[0147]

[0148]

[0149] Among them, 、 and respectively represent the power transfer distribution factors of unit 、new energy power station and node to line the power transfer distribution factor, represents the steady-state power flow limit value of line ; represents the set of transmission lines;

[0150] The curtailment constraint of the new energy power station is:

[0151]

[0152] Among them, represents the curtailment amount of the new energy power station at time period .

[0153] Exemplarily, the power grid topology data may include the connection relationships of power network nodes and transmission lines, and the output power transfer distribution factor matrix of each type of unit and node load to each transmission line;

[0154] The cost data of the units participating in the optimization may include the start-up cost, shutdown cost, no-load cost, and variable power generation cost function of the units at each time period;

[0155] The operation characteristic data of the units participating in the optimization may include the minimum / maximum technical output power of the units, the ramping ability of the units at each time period, and the minimum continuous start-up and shutdown time of the units;

[0156] The day-ahead predicted load data may include the day-ahead predicted load data of each node, with a time resolution of 1 hour or 15 minutes;

[0157] The day-ahead predicted output data of the new energy power station may include the day-ahead predicted output data of the new energy power station, with a time resolution of 1 hour or 15 minutes.

[0158] S3. Set the new energy prediction deviation to 0, and calculate the system net load and system net load rate of the power system for each time period according to the day-ahead predicted output data of the new energy power station and the day-ahead load prediction data of each load node in the power system.

[0159] In this embodiment, according to the day-ahead predicted output data of the new energy power station and the day-ahead load prediction data of each load node in the power system, the system net load and system net load rate of the power system for each time period are calculated using the following formulas:

[0160]

[0161]

[0162] where represents the system net load, represents the system net load rate.

[0163] S4. Consider the new energy prediction deviation within the limit range of the robust uncertainty set, use the objective function to construct the first auxiliary mixed-integer linear programming model and the second auxiliary mixed-integer linear programming model respectively, and substitute the calculation results of the system net load and system net load rate into the first auxiliary mixed-integer linear programming model and the second auxiliary mixed-integer linear programming model respectively to determine the minimum net load representative scenario and the maximum net load representative scenario;

[0164] Among them, the first auxiliary mixed-integer linear programming model is an auxiliary mixed-integer linear programming model that minimizes the product sum of the reciprocal of the system net load rate and the system net load, and the second auxiliary mixed-integer linear programming model is an auxiliary mixed-integer linear programming model that maximizes the product sum of the system net load rate and the system net load.

[0165] In this embodiment, the first auxiliary mixed-integer linear programming model is:

[0166]

[0167]

[0168]

[0169] where represents the system net load variable, represents the system net load rate. After solving the first auxiliary mixed-integer linear programming model, a set of optimal solutions can be obtained , which is the minimum net load representative scenario;

[0170] The second auxiliary mixed-integer linear programming model is:

[0171]

[0172]

[0173]

[0174] Among them, after solving the second auxiliary mixed-integer linear programming model, a set of optimal solutions can be obtained, which is the maximum net load representative scenario.

[0175] Exemplarily, by solving the first auxiliary mixed-integer linear programming model, a set of minimum net load representative scenarios reflecting the potential of the next day can be obtained, that is, the output prediction deviation worst-case estimates of each new energy power station at each time period are obtained. .

[0176] Similarly, by solving the second auxiliary mixed-integer linear programming model, a set of maximum net load representative scenarios reflecting the potential of the next day can be obtained, that is, the output prediction deviation worst-case estimates of each new energy power station at each time period are obtained. .

[0177] S5. Substitute the minimum net load representative scenario into the unit commitment optimization model for solution, and set the must-run unit discrimination threshold according to the minimum and maximum technical output parameters of the units participating in the optimization to determine the set of must-run units.

[0178] In this embodiment, S5 specifically includes:

[0179] S51. Substitute the minimum net load representative scenario into the unit commitment optimization model to obtain a new unit commitment optimization model;

[0180] The new unit commitment optimization model is:

[0181]

[0182] Among them, represents the minimum net load representative scenario;

[0183] S52. Solve the new unit commitment optimization model to obtain the first unit startup mode and the first output level optimal solution of the unit set participating in the optimization;

[0184] S53. Determine the set of must - start units from the first unit start - up mode and the optimal solution of the first output level using the must - start unit discrimination threshold.

[0185] Exemplarily, substitute the minimum net load representative scenario into the unit commitment optimization model, that is, each new energy power station At each time period The worst - case estimated value of the output prediction deviation Substitute them one by one into the unit commitment optimization model to obtain the first unit commitment optimization model. Solve the new unit commitment optimization model to obtain the first unit start - up mode and the optimal solution of the first output level of the units participating in the optimization, that is . Set the must - start unit discrimination threshold based on the minimum and maximum technical outputs of the units participating in the optimization, that is As a constraint condition, determine the set of must - start units from the first unit start - up mode and the optimal solution of the first output level, that is , where the units participating in the optimization in the set of must - start units As long as they satisfy the set must - start unit discrimination threshold in all scheduling time periods They are the elements of the set of must - start units .

[0186] S6. Substitute the maximum net load representative scenario and the set of must - start units into the unit commitment optimization model for solution, and determine the set of must - stop units according to the must - stop unit discrimination rule, combining the solution result and the set of must - start units.

[0187] In this embodiment, S6 specifically includes:

[0188] S61. Substitute the maximum net load representative scenario into the unit commitment optimization model to obtain the second unit commitment optimization model;

[0189] The second unit commitment optimization model is:

[0190]

[0191] Among them, Represents the maximum net load representative scenario;

[0192] S62. Substitute the set of must - start units into the second unit commitment optimization model to obtain the third unit commitment optimization model;

[0193] S63. Solve the third unit commitment optimization model to obtain the second unit start - up mode and the optimal solution of the second output level of the units participating in the optimization;

[0194] S64. According to the must - stop unit discrimination rule, combine the second unit start - up mode, the optimal solution of the second output level and the set of must - start units to determine the set of must - stop units.

[0195] Exemplarily, substituting the maximum net load representative scenario into the unit commitment optimization model means substituting each new energy power station at each time period the worst estimated value of the output prediction deviation as known quantities, and substituting them one by one into the unit commitment optimization model in the form of adding additional constraints to obtain the second unit commitment optimization model. Then, substituting the set of must-run units into the second unit commitment optimization model in the form of adding additional constraints to obtain the third unit commitment optimization model, that is, the third unit commitment optimization model is . Solving the third unit commitment optimization model to obtain the optimal solution of the second unit startup mode and the second output level of the units participating in the optimization, that is . According to the must-stop unit discrimination rule, combining the second unit startup mode, the optimal solution of the second output level and the set of must-run units, determine the set of must-stop units, that is: . In the set of must-stop units, as long as the operating status of the units participating in the optimization is 0 in all scheduling time periods of the two representative scenarios, that is, not starting up, it is the element of the set of must-stop units . Element.

[0196] S7. Take the set of must-run units and the set of must-stop units as the optimization plan, and combine the robust uncertainty set and the unit commitment optimization model to construct a robust unit commitment reduction model to accelerate the optimization of the robust unit commitment problem of the power system.

[0197] In this embodiment, S7 specifically includes:

[0198] S71. Combine the robust uncertainty set and the unit commitment optimization model to construct a robust unit commitment model;

[0199] The robust unit commitment model is:

[0200]

[0201]

[0202] Among them, , , , , , vector is the first-stage decision variable, vector is the second-stage decision variable, vector is the uncertain parameter restricted by the robust uncertainty set , that is, the new energy prediction deviation, is about vector Sum vector The value function is the parameter vector for optimizing the no-load cost, start-up cost, and shutdown cost of the unit is the parameter vector for the variable cost of the unit participating in the optimization and the curtailment penalty cost parameter of the new energy power station. Matrix and vector are respectively the coefficient matrix and the constant vector of the independent inequality constraints in the first stage, corresponding to the unit start-stop state constraint and the minimum continuous start-stop constraint of the unit in the unit commitment optimization model. Matrix and vector are respectively the coefficient matrix and the constant vector of the independent inequality constraints in the second stage, corresponding to the line safety constraint and the new energy power station curtailment constraint in the unit commitment optimization model. Matrix Matrix and vector are respectively the two coefficient matrices and the constant vector of the coupled inequality constraints in the first stage and the second stage, corresponding to the upper and lower limits of the unit output and the unit ramp rate constraint in the unit commitment optimization model; Matrix Matrix Matrix and vector are respectively the three coefficient matrices and the constant vector of the equality constraints coupled by the second stage and the robust uncertainty set, corresponding to the system supply-demand balance constraint in the unit commitment optimization model;

[0203] S72. Taking the set of units that must be started and the set of units that must be stopped as additional constraints, a robust unit commitment reduced model is obtained;

[0204] The robust unit commitment reduced model is:

[0205]

[0206]

[0207] where represents the vector composed of the decision variables in the first stage of the robust unit commitment model in the th variable.

[0208] Exemplarily, the two equations of the robust unit commitment reduced model find the specific positions of the variables representing the operating states of the corresponding units in the robust unit commitment model according to the numbers of the units participating in the optimization, add constraints, and then use the column and constraint generation algorithm to solve the robust unit commitment reduced model.

[0209] Those of ordinary skill in the art will realize that the embodiments described herein are provided to assist the reader in understanding the principles of the present invention, and it should be understood that the scope of protection of the present invention is not limited to such specific statements and embodiments. Those of ordinary skill in the art can make various other specific deformations and combinations that do not depart from the essence of the present invention based on these technical revelations disclosed in the present invention, and these deformations and combinations are still within the scope of protection of the present invention.

Claims

1. A method for accelerating optimization of robust unit commitment in power system based on representative scenarios, characterized in that: The method comprises: S1. Obtain historical predicted power generation data and historical actual power generation data of new energy stations, and construct a robust uncertainty set based on the historical predicted power generation data and the historical actual power generation data; S2. Obtain the day-ahead optimal dispatching data of the power system and build a unit combination optimization model; S3, setting the new energy forecast deviation to 0, and calculating the system net load and system net load rate of the power system in each period according to the day-ahead forecast output data of the new energy station and the day-ahead load forecast data of each load node in the power system; S4. Considering the new energy prediction deviation within the limited range of the robust uncertainty set, using the objective function to respectively construct a first auxiliary mixed integer linear programming model and a second auxiliary mixed integer linear programming model, and substituting the calculation results of the system net load and the system net load rate into the first auxiliary mixed integer linear programming model and the second auxiliary mixed integer linear programming model, respectively, to determine the minimum net load representative scenario and the maximum net load representative scenario; The first auxiliary mixed integer linear programming model is an auxiliary mixed integer linear programming model for minimizing the sum of the product of the inverse of the system net load rate and the system net load, and the second auxiliary mixed integer linear programming model is an auxiliary mixed integer linear programming model for maximizing the sum of the product of the system net load rate and the system net load; S5, substituting the minimum net load representative scenario into the unit combination optimization model for solution, and setting the threshold for determining the must-start units according to the minimum and maximum technical output parameters of the units participating in the optimization, and determining the must-start unit set; S6, substituting the maximum net load representative scenario and the set of units that must be started into the unit commitment optimization model for solution, and determining the set of units that must be stopped according to the rules for distinguishing the units that must be stopped, combining the solution result and the set of units that must be started; S7. Taking the set of units that must be started and the set of units that must be stopped as optimization schemes, and combining the robust uncertainty set and the unit combination optimization model, a simplified robust unit combination model is constructed to achieve accelerated optimization of the robust unit combination problem of the power system.

2. The method according to claim 1, characterized in that In S1, constructing a robust uncertainty set based on the historical predicted power generation data and the historical actual power generation data specifically includes: S11, calculating the output prediction deviation according to the historical predicted power generation data and the historical actual power generation data; The formula used to calculate the output prediction deviation is: in, Represents new energy station No. Day time The output forecast deviation, Represents new energy station No. Day time The actual power generation data, Represents new energy station No. Day time Forecasted power generation data, Represents a collection of new energy stations. Represents a collection of historical data days, represents a set of scheduling periods; S12, constructing the robust uncertainty set according to the mean and standard deviation of the output prediction deviation; The construction of the robust uncertainty set The formula used is: in, Represents the robust uncertainty set of new energy stations Time The output prediction deviation ranges from , express Tiantian New Energy Station Time The mean of the output forecast deviation, express Tiantian New Energy Station Time The standard deviation of the output forecast deviation, Indicates the conservative control coefficient, the default value is 1.

5.

3. The method according to claim 2, characterized in that In S2, the day-ahead optimization dispatching data includes power grid topology data, cost data of the units participating in the optimization, operation characteristic data of the units participating in the optimization, day-ahead load forecast data, and day-ahead output forecast data of new energy stations; The constructing of the unit commitment optimization model specifically includes: Constructing the unit commitment optimization model and the constraint conditions corresponding to the unit commitment optimization model; The constraints corresponding to the unit combination optimization model include upper and lower limits of unit output, unit start and stop status constraints, minimum continuous start and stop constraints, unit climbing constraints, system supply and demand balance constraints, line safety constraints and new energy station power abandonment constraints; The unit commitment optimization model is: in, represents the total operating cost, Indicates the units involved in optimization In the period The operating status of represents the no-load cost of the units participating in the optimization, Indicates the unit In the period Whether a boot event occurs, Indicates the startup cost of the startup event, Indicates the unit In the period Whether a downtime event occurred, represents the downtime cost of the downtime event, Indicates time period unit The output level, Indicates the unit The variable generation cost function is It represents the penalty fee for abandoning electricity from new energy sources. Represents new energy station In the period of wasted power, Represents the set of units participating in optimization; The upper and lower limits of the unit output are: in, and Respectively represent the minimum technical output and maximum technical output of the unit; The start and stop state constraints of the unit are: in, Indicates the units involved in optimization In the period The operating status of The minimum continuous start and stop constraints of the unit are: in, Indicates the unit In the period Whether a boot event occurs, Indicates the unit In the period Whether a downtime event occurred , Indicates the unit Minimum continuous power-on time, Indicates the unit Minimum continuous shutdown time; The unit climbing constraint is: in, and They represent the maximum down-ramp rate and the maximum up-ramp rate of the unit respectively. Indicates time period unit Output level; The system supply and demand balance constraint is: in, Represents new energy station In the period The day-ahead forecast output is Represents new energy station In the period The forecast deviation between the day-ahead forecast output and the next day's actual output is Representation Node In the period The load forecast value of the day ahead is represents a set of load nodes; The line safety constraints are: in, , and Respectively represent the unit 、New Energy Station and nodes To line The power transfer distribution factor, Indicates line The steady-state power flow limit value is represents a collection of transmission lines; The constraints on power abandonment at new energy stations are: in, Represents new energy station In the period of wasted electricity.

4. The method according to claim 3, characterized in that In S3, the system net load and system net load rate of the power system in each period are calculated based on the day-ahead predicted output data of the new energy station and the day-ahead load forecast data of each load node in the power system, and the formula used is: in, Indicates the net load of the system. Indicates the net load rate of the system.

5. The method according to claim 4, characterized in that In S4, the first auxiliary mixed integer linear programming model is: in, represents the system net load variable, Represents the net load rate of the system. After solving the first auxiliary mixed integer linear programming model, a set of The optimal solution , which is the minimum net load representative scenario; The second auxiliary mixed integer linear programming model is: Among them, after solving the second auxiliary mixed integer linear programming model, a set of The optimal solution , which is the representative scenario of maximum net load.

6. The method according to claim 5, characterized in that The S5 specifically includes: S51, substituting the minimum net load representative scenario into the unit commitment optimization model to obtain a new unit commitment optimization model; The new unit combination optimization model is: in, Indicates the minimum net load representative scenario; S52, solving the new unit combination optimization model to obtain the optimal solution of the first startup mode and the first output level of each unit participating in the optimization unit set; S53: using the must-start unit discrimination threshold, determine the must-start unit set from the first unit startup methods and the first output level optimal solution.

7. The method according to claim 6, characterized in that The S6 specifically includes: S61, substituting the maximum net load representative scenario into the unit commitment optimization model to obtain a second unit commitment optimization model; The second unit combination optimization model is: in, Indicates the maximum net load representative scenario; S62, substituting the set of must-start units into the second unit combination optimization model to obtain a third unit combination optimization model; S63, solving the third unit combination optimization model to obtain the optimal solution for the second startup mode and the second output level of each unit participating in the optimization unit set; S64. Determine the set of units that must be shut down according to a rule for distinguishing the units that must be shut down, in combination with the second startup modes of each unit, the second optimal solution for the output level, and the set of units that must be shut down.

8. The method according to claim 7, characterized in that The S7 specifically includes: S71, constructing a robust unit commitment model by combining the robust uncertainty set and the unit commitment optimization model; The robust unit commitment model is: in, , , , , ,vector is the first-stage decision variable, vector is the second-stage decision variable, the vector is restricted to the robust uncertainty set The uncertainty parameter, i.e., the new energy prediction deviation, It's about vector and vector The value function of is the parameter vector of the no-load cost, startup cost and shutdown cost of the unit involved in the optimization, is the variable cost parameter of the optimized units and the penalty cost parameter vector of the power abandonment of the new energy station, the matrix and vector They are the coefficient matrix and constant vector of the independent inequality constraints in the first stage, corresponding to the unit start-stop state constraints and the unit minimum continuous start-stop constraints in the unit combination optimization model. and vector They are the coefficient matrix and constant vector of the independent inequality constraints in the second stage, corresponding to the line safety constraints and the power abandonment constraints of the new energy station in the unit commitment optimization model. ,matrix and vector are the two coefficient matrices and constant vector of the inequality constraints coupled in the first and second stages, corresponding to the upper and lower limits of unit output and the unit ramp constraints in the unit combination optimization model; the matrix ,matrix ,matrix and vector They are the three coefficient matrices and constant vectors of the equality constraints of the second stage and the robust uncertainty set coupling, corresponding to the system supply and demand balance constraints in the unit commitment optimization model; S72, taking the set of units that must be started and the set of units that must be stopped as additional constraints to obtain the simplified model of the robust unit commitment; The simplified model of robust unit commitment is: in, Represents the vector of decision variables in the first stage of the robust unit commitment model The variables.

Images