Training data enhancement and rapid generation method for unit commitment optimization problem

Abstract

The invention discloses a training data enhancement and rapid generation method for a unit commitment optimization problem, and relates to the technical field of power systems, and the method comprises the steps: obtaining historical high-time-resolution power system unit commitment input data including a scheduling time period set, recording the data as a first day-ahead data set, and calculating the system net load; resampling the scheduling period set into a scheduling period subset according to a time sequence resampling rule; extracting corresponding new energy output data and load prediction data from the first day-ahead data set according to time periods to obtain a second day-ahead data set; performing proportional adjustment and expansion on the new energy output data and the load prediction data in the second day-ahead data set according to a data enhancement rule to form a third day-ahead data set; and constructing a unit commitment model of which the number is consistent with that of the data set samples before the third day, and solving the unit commitment model. According to the method, the problems that a data driving method is applied to high-time-resolution unit combinatorial optimization and training data collection is difficult when an electric power spot market is cleared day by day are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems, and in particular to a method for enhancing and rapidly generating training data for a unit commitment optimization problem. Background Art

[0002] The security-constrained unit commitment (SCUC) problem is a typical non-convex, mixed-integer optimization problem commonly used in day-ahead clearing of electricity spot markets. The worst-case computational complexity of the SCUC problem is exponentially proportional to the number of integer variables. SCUC problems with high time resolutions (such as 15 minutes, 10 minutes, and 5 minutes) exponentially increase the number of integer variables and constraints. This increased computational complexity comes at the expense of achieving safer and more economical power system dispatch solutions compared to traditional 1-hour time resolutions. Therefore, efficiently solving SCUC problems with high time resolutions has become a key technical challenge that needs to be addressed.

[0003] Data-driven methods can quickly identify redundant line constraints and unit integer variables in the high-time-resolution SCUC problem to be solved, and are an effective measure to improve the computational efficiency of high-time-resolution SCUC problems. However, the application of data-driven methods faces a circular dependency problem: the inherent difficulty of solving high-time-resolution SCUC problems necessitates data-driven methods to improve efficiency, but sufficient high-quality training data requires the prior solution of a large number of high-time-resolution SCUC instances. Therefore, the efficient generation of diverse training data has become a key bottleneck restricting the application of data-driven methods to high-time-resolution SCUC problems. Summary of the Invention

[0004] In response to the above-mentioned deficiencies in the prior art, the present invention provides a method for enhancing and rapidly generating training data for the unit commitment optimization problem, which solves the difficulty in collecting training data faced when applying data-driven methods to high-time-resolution unit commitment optimization problems and day-ahead clearing of the electricity spot market.

[0005] In order to achieve the above-mentioned object of the invention, the technical solution adopted by the present invention is: a method for enhancing and rapidly generating training data for unit commitment optimization problems, comprising the following steps:

[0006] S1: Obtain historical high-time-resolution power system unit commitment input data including the dispatch period set, record it as the day-ahead dataset, and calculate the system net load in the day-ahead dataset;

[0007] S2: Based on the system net load in the data set before the first day and the preset time series resampling rules, the high-resolution scheduling period set is resampled into a subset of scheduling periods with a set time resolution;

[0008] S3: Based on the resampled scheduling period subset, the corresponding high-time-resolution renewable energy output data and load forecast data are extracted from the first-day-ahead dataset by time period to form the set time-resolution data, thus obtaining the second-day-ahead dataset.

[0009] S4: According to the preset data enhancement rules, the set time resolution new energy output data and load forecast data in the second day data set are proportionally adjusted and expanded to form the third day data set;

[0010] S5: Construct and solve a unit commitment model with a set time resolution that is consistent with the number of samples in the dataset before the third day, record the optimization solution results and their analysis results as new data, and complete the training data enhancement and rapid generation of the unit commitment optimization problem.

[0011] Furthermore, the timing resampling rules preset in S2 include:

[0012] (1) According to the 1-hour time resolution, the scheduling period set is decomposed into S scheduling period subsets that are associated with the scheduling time resolution and do not overlap with each other, where The first sample Schedule period subset The following scheduling periods are included:

[0013]

[0014] in, is the resampling interval, , is the scheduling time resolution, is the scheduling period set, is the subset symbol, is any symbol, Set for scheduling period The subscripts represent the scheduling period set No. , , , Until the scheduling periods;

[0015] (2) Count the maximum and minimum values of the system net load every hour and form four scheduling period subsets in a certain order. The specific steps include the following:

[0016] a1: Schedule period set Group by hour, The scheduling period included in the hour for:

[0017]

[0018] in, Set for scheduling period No. Scheduling period, subscript The value range is based on Hours decide, ;

[0019] a2: Statistics Sample No. The periods of maximum and minimum system net load values in all hourly dispatch periods;

[0020] a3: Based on the time periods of the maximum and minimum system net loads, four scheduling period subsets are formed in the order of maximum-minimum, minimum-maximum, maximum-maximum-minimum, and minimum-minimum-maximum:

[0021]

[0022]

[0023]

[0024]

[0025] in, For the The maximum-minimum scheduling period subset in samples, For the The subset of the minimum-maximum scheduling periods in the samples, For the The subset of the maximum-maximum-minimum scheduling periods in the samples, For the The subset of the minimum-minimum-maximum scheduling periods in the samples, and Respectively Sample No. The period of maximum and minimum system net load in all scheduling periods of the hour, subscript and Respectively represent Hour and Hour;

[0026] Obtained by the timing resampling rule A subset of scheduling periods.

[0027] Furthermore, the step S3 includes the following sub-steps:

[0028] S31: For each scheduling period subset, extract the renewable energy output data and load forecast data for the corresponding period in the data set before the first day to form 1-hour resolution data;

[0029] The new energy output data includes the new energy station's day-ahead output data. According to the high time resolution new energy station day-ahead output data of the sample Schedule period subset Extract data slices during the included scheduling period:

[0030]

[0031] in, For the Sample new energy stations In the Schedule period subset The day-ahead forecast output is For the Sample new energy stations Gather during the scheduled time Time The day-ahead forecast output is It is a collection of new energy stations. For the a subset of scheduling periods;

[0032] The load forecast data includes load day forecast data, which is obtained by The schedule period subset contains the schedule period extraction data slice:

[0033]

[0034] in, Indicates the Sample nodes In the Schedule period subset The load forecast value of the day before is Indicates the Sample nodes Gather during the scheduled time Next period The load forecast value of the day before is is the set of load nodes;

[0035] S32: Each scheduling period subset on each day constitutes an independent sample, and together constitutes the data set before the next day.

[0036] Furthermore, the S4 includes the following sub-steps:

[0037] S41: Constructing a set of proportional adjustment factors , the formula is:

[0038]

[0039] in, is the preset adjustment factor limit, To adjust the step size, the expansion factor is ;

[0040] S42: The sample 1-hour time resolution new energy output data and load forecast data in the second day data set, which are expanded from the scheduling period subset of the first day data set, are scaled and expanded according to the proportional adjustment factor:

[0041]

[0042]

[0043] in, and Respectively represent Sample No. New energy stations in the sample expanded from the scheduling period subset In the period Day-ahead forecast output and nodes In the period The load forecast value of the day before is and On this basis, according to New samples scaled by the scaling factor, Indicates the Scaling factors;

[0044] S43: Each scheduling period subset of each day After scaling and expansion by the proportional adjustment factors, an independent sample is formed, which together constitutes the data set before the third day.

[0045] Furthermore, the objective function of the unit commitment model with a set time resolution that is consistent with the number of samples in the data set before the third day constructed in S5 is:

[0046]

[0047] Among them, in Sample No. The scheduling period subset is The samples are scaled by the scaling factor. is the total operating cost, To participate in the optimization of the collection of units, For the A subset of scheduling periods, 、 and Respectively for units Time The binary variables of the running status, whether the startup event occurs, and whether the shutdown event occurs, 、 and Respectively for units Time No-load costs, startup costs and shutdown costs, For the crew Time Lower and output level The associated variable cost function for electricity generation.

[0048] The present invention also adopts a technical solution: a training data enhancement and rapid generation device for unit commitment optimization problems, comprising:

[0049] Memory, used to store computer programs;

[0050] A processor is used to execute the computer program to implement the above-mentioned training data enhancement and rapid generation method for the unit commitment optimization problem.

[0051] The present invention provides a method for enhancing and rapidly generating training data for unit commitment optimization problems. Based on time series resampling and proportional expansion adjustment, this method enhances and rapidly generates historical high-temporal-resolution unit commitment data, enabling rapid acquisition of training data. This method significantly reduces the difficulty of collecting training data and provides effective data support for the application of data-driven methods to large-scale, high-temporal-resolution unit commitment optimization problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 Flowchart of a method for training data augmentation and rapid generation for a unit commitment optimization problem. DETAILED DESCRIPTION

[0053] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0054] like Figure 1 As shown, a method for enhancing and rapidly generating training data for a unit commitment optimization problem includes the following steps:

[0055] S1: Obtain historical high-time-resolution power system unit commitment input data including the dispatch period set, record it as the day-ahead dataset, and calculate the system net load in the day-ahead dataset;

[0056] S2: Based on the system net load in the data set before the first day and the preset time series resampling rules, the high-resolution scheduling period set is resampled into a subset of scheduling periods with a set time resolution;

[0057] S3: Based on the resampled scheduling period subset, the corresponding high-time-resolution renewable energy output data and load forecast data are extracted from the first-day-ahead dataset by time period to form the set time-resolution data, thus obtaining the second-day-ahead dataset.

[0058] S4: According to the preset data enhancement rules, the set time resolution new energy output data and load forecast data in the second day data set are proportionally adjusted and expanded to form the third day data set;

[0059] S5: Construct and solve a unit commitment model with a set time resolution that is consistent with the number of samples in the dataset before the third day, record the optimization solution results and their analysis results as new data, and complete the training data enhancement and rapid generation of the unit commitment optimization problem.

[0060] In S1, historical high-time-resolution power system unit combination input data including the scheduling period set is obtained, wherein the input data of each day constitutes an independent sample and together constitutes the data set before the first day. The high-time-resolution power system unit combination input data of each sample of the data set before the first day includes:

[0061] Scheduling period collection: all scheduling periods of the next 24 hours, The specific number of time periods depends on the scheduling time resolution. High time resolution generally refers to 15 minutes, 10 minutes or 5 minutes. In this case, the number of scheduling time periods is The values are 96, 144, and 288 respectively. The following uses a 15-minute scheduling resolution as an example.

[0062] Grid topology data: the connection relationship between power network nodes and transmission lines, and the output power transfer distribution factor matrix of each type of unit and node load on each transmission line;

[0063] Cost data of the units involved in optimization: startup costs, shutdown costs, no-load costs, and variable power generation cost functions for each period of the units;

[0064] Participate in optimizing the unit's operating characteristic data: minimum / maximum technical output of the unit, unit climbing capability, and minimum continuous start-up and shutdown time of the unit;

[0065] Load day-ahead forecast data: Load day-ahead forecast data for each node, with a time resolution consistent with the scheduling time resolution;

[0066] The day-ahead predicted output data of new energy stations; the day-ahead predicted output data of new energy stations has a time resolution consistent with the scheduling time resolution.

[0067] The system net load for each sample in the dataset before the first day is calculated as follows:

[0068]

[0069] in, represents a set of scheduling periods, Indicates the Sample period The net load value of the system, Indicates the Sample nodes In the period The load forecast value of the day before is Indicates the Sample new energy stations In the period The day-ahead forecast output is Represents a collection of new energy sites, including photovoltaic sites and wind farms. Represents a set of load nodes;

[0070] The timing resampling rules preset in S2 include:

[0071] (1) According to the 1-hour time resolution, the scheduling period set is decomposed into S scheduling period subsets that are associated with the scheduling time resolution and do not overlap with each other, where The first sample Schedule period subset The following scheduling periods are included:

[0072]

[0073] in, is the resampling interval, , The scheduling time resolution can be 15 minutes, 10 minutes or 5 minutes. In this case, the number of scheduling period subsets is 4, 6 or 12. is the scheduling period set, is the subset symbol, is any symbol, Set for scheduling period The subscripts represent the scheduling period set No. , , , Until the scheduling periods;

[0074] (2) Count the maximum and minimum values of the system net load every hour and form four scheduling period subsets in a certain order. The specific steps include the following:

[0075] a1: Set of scheduling periods Group by hour, The scheduling period included in the hour for:

[0076]

[0077] in, Set for scheduling period No. Scheduling period, subscript The value range is based on Hours decide, ;

[0078] a2: Statistics Sample No. The periods of maximum and minimum system net load values in all hourly dispatch periods;

[0079] a3: Based on the time periods of the maximum and minimum system net loads, four scheduling period subsets are formed in the order of maximum-minimum, minimum-maximum, maximum-maximum-minimum, and minimum-minimum-maximum:

[0080]

[0081]

[0082]

[0083]

[0084] in, For the The maximum-minimum scheduling period subset in samples, For the The subset of the minimum-maximum scheduling periods in the samples, For the The subset of the maximum-maximum-minimum scheduling periods in the samples, For the The subset of the minimum-minimum-maximum scheduling periods in the samples, and Respectively Sample No. The period of maximum and minimum system net load in all scheduling periods of the hour, subscript and Respectively represent Hour and Hour;

[0085] Obtained by the timing resampling rule A subset of scheduling periods.

[0086] In S3, according to Resampled samples Schedule period subset , extract the corresponding high-time-resolution new energy output and load forecast data by time period from the data set before the first day, and form 1-hour time-resolution data:

[0087] The S3 includes the following steps:

[0088] S31: For each scheduling period subset, extract the renewable energy output data and load forecast data for the corresponding period in the data set before the first day to form 1-hour resolution data;

[0089] The new energy output data includes the new energy station's day-ahead output data. According to the high time resolution new energy station day-ahead output data of the sample Schedule period subset Extract data slices during the included scheduling period:

[0090]

[0091] in, For the Sample new energy stations In the Schedule period subset The day-ahead forecast output is Including the representative time period of each hour of the next day, so it is converted to a 1-hour time resolution, For the Sample new energy stations Gather during the scheduled time Time The day-ahead forecast output is high temporal resolution data. Schedule period subset period and scheduling period set Time All represent the same scheduling period. It is a collection of new energy sites, including photovoltaic sites and wind farms. For the a subset of scheduling periods;

[0092] The load forecast data includes load day forecast data, which is obtained by The schedule period subset contains the schedule period extraction data slice:

[0093]

[0094] in, Indicates the Sample nodes In the Schedule period subset The load forecast value of the day before is Indicates the Sample nodes Gather during the scheduled time Next period The load forecast value of the day before is is the set of load nodes;

[0095] S32: Each scheduling period subset on each day constitutes an independent sample, which together form the next day's day data set. Samples on the same day inherit the same grid topology data, cost data of the participating units, and operating characteristics data of the participating units. Only the corresponding scheduling period subset, 1-hour time resolution day-ahead load forecast data, and new energy station day-ahead output forecast data are retained.

[0096] The S4 includes the following steps:

[0097] S41: Constructing a set of proportional adjustment factors , the formula is:

[0098]

[0099] in, is the preset adjustment factor limit, To adjust the step size, the expansion factor is ;

[0100] S42: The sample 1-hour time resolution new energy output data and load forecast data in the second day data set, which are expanded from the scheduling period subset of the first day data set, are scaled and expanded according to the proportional adjustment factor:

[0101]

[0102]

[0103] in, and Respectively represent Sample No. New energy stations in the sample expanded from the scheduling period subset In the period Day-ahead forecast output and nodes In the period The load forecast value of the day before is and On this basis, according to New samples scaled by the scaling factor, Indicates the Scaling factors;

[0104] S43: Each scheduling period subset of each day After scaling and expansion by the corresponding adjustment factors, an independent sample is formed, which together constitutes the third-day day-ahead dataset. Samples of the same scheduling period subset on the same day inherit the same grid topology data, cost data of the optimized units, operating characteristics data of the optimized units, and scheduling period subset. Only the load day-ahead forecast data and the new energy station day-ahead output forecast data with a time resolution of 1 hour after scaling and expansion by the corresponding adjustment factors are retained.

[0105] In S5, the dataset is Sample No. The scheduling period subset is Before constructing the unit combination optimization model corresponding to the sample scaled by the proportional adjustment factor, the power system unit combination input data needs to be calculated according to the Sample high time resolution power system unit commitment input data and dispatch time resolution Converted to match 1 hour time resolution:

[0106] Cost data of units involved in optimization:

[0107]

[0108]

[0109]

[0110]

[0111] Among them, Sample No. Schedule period subset period and scheduling period set Time All represent the same scheduling period. 、 、 and Respectively Sample No. The scheduling period subset is The sample unit after scaling by the proportional adjustment factor Time No-load costs, startup costs, shutdown costs and unit In the period The variable cost function of power generation related to the output level is: 、 、 and Respectively Sample scheduling period set Time Lower unit No-load costs, startup costs, shutdown costs and unit In the period The power generation variable cost function related to the output level, the conversion ratio and the scheduling time resolution Related, for ;

[0112] Operational characteristic data of units involved in optimization:

[0113]

[0114]

[0115]

[0116]

[0117]

[0118]

[0119] in, 、 、 、 、 and Respectively Sample No. The scheduling period subset is The sample unit after scaling by the proportional adjustment factor The maximum technical output, minimum technical output, maximum ramp-up rate, maximum ramp-down rate, minimum continuous start-up time parameters and minimum continuous stop time parameters, 、 、 、 、 and Respectively Sample scheduling period set Lower unit Maximum technical output, minimum technical output, maximum ramp-up rate, maximum ramp-down rate, minimum continuous start-up time parameters and minimum continuous stop time parameters;

[0120] The objective function of the unit commitment model with a set time resolution that is consistent with the number of samples in the dataset before the third day constructed in S5 is:

[0121]

[0122] Among them, in Sample No. The scheduling period subset is The samples are scaled by the scaling factor. is the total operating cost, To participate in the optimization of the collection of units, For the A subset of scheduling periods, 、 and Respectively for units Time The binary variables of the running status, whether the startup event occurs, and whether the shutdown event occurs, 、 and Respectively for units Time No-load costs, startup costs and shutdown costs, For the crew Time Lower and output level The associated variable cost function for electricity generation.

[0123] The constraints of the above-mentioned unit commitment optimization model include upper and lower limits of unit output, unit start and stop status constraints, minimum continuous start and stop constraints, unit ramp constraints, power system supply and demand balance constraints, power system positive and negative reserve constraints, and line power flow security constraints.

[0124] The upper and lower limits of the unit output are:

[0125]

[0126] in, and Respectively Sample No. The scheduling period subset is The sample unit after scaling by the proportional adjustment factor The minimum and maximum technical output of

[0127] The start and stop state constraints of the unit are:

[0128]

[0129] in, Indicates the Sample No. The scheduling period subset is The sample unit after scaling by the proportional adjustment factor Time The following operating status;

[0130] The minimum continuous start and stop constraints of the unit are:

[0131]

[0132]

[0133] in, and Respectively represent Sample No. The scheduling period subset is The sample unit after scaling by the proportional adjustment factor Time Check whether startup and shutdown events occur. and Respectively represent the units The minimum continuous on time parameters and the minimum continuous off time parameters;

[0134] The unit climbing constraint is:

[0135]

[0136] in, and Respectively represent Sample No. The scheduling period subset is The sample unit after scaling by the proportional adjustment factor The maximum downhill climbing rate and the maximum uphill climbing rate, Indicates the unit In the period The output level of the

[0137] The power system supply and demand balance constraint is:

[0138]

[0139] in, and Respectively represent Sample No. The scheduling period subset is New energy stations under the sample after scaling by the proportional adjustment factor Time The day-ahead forecast output and nodes Time The load forecast value under It is a collection of new energy stations, including photovoltaic stations and wind farms. is the set of load nodes;

[0140] The positive and negative reserve constraints of the power system are:

[0141]

[0142]

[0143] in, Indicates the system positive reserve rate, Indicates the system negative reserve ratio.

[0144] The line power flow safety constraint is:

[0145]

[0146] in, For the line The steady-state power flow limit value, 、 and Respectively for units , new energy stations and nodes Line The power transfer distribution factor, For transmission line collection;

[0147] Finally, the unit combination optimization model is solved using a commercial solver to obtain the optimal unit start and stop schedule matrix. The optimization solution results included.

[0148] According to the results of the optimization solution of the unit combination optimization model, we can further analyze and determine the Sample No. The scheduling period subset is The set of units that must be started or stopped under the sample scaled by the proportional adjustment factor and the set of flexible units that have start-stop events:

[0149]

[0150]

[0151]

[0152] in, To participate in the optimization of the collection of units, Represents the set of units that must be started, when the optimal start-stop plan matrix of the unit Medium unit When all rows are 1, the unit This is the set element of the must-start unit. Represents the set of units that must be shut down, when the optimal start-stop plan matrix of the unit Medium unit When all rows are 0, the unit That is the set element of the must-stop group, Represents a set of flexible units. When the unit When it does not belong to the set of must-start units nor the set of must-stop units, it is a flexible unit set element. Represents the difference operation.

[0153] The above-mentioned set of must-start and must-stop and flexible units serves as an important analysis result, together with data such as the power flow of each line in each time period and the output of each unit in each time period, to constitute a complete analysis result.

[0154] Example 2, a training data enhancement and rapid generation device for a unit commitment optimization problem, comprising:

[0155] Memory, used to store computer programs;

[0156] A processor is used to execute the computer program to implement the above-mentioned training data enhancement and rapid generation method for the unit commitment optimization problem.

[0157] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand 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 descriptions and embodiments. Those skilled in the art can make various other specific variations and combinations based on the technical teachings disclosed in the present invention without departing from the essence of the present invention, and such variations and combinations are still within the scope of protection of the invention.

Claims

1. A method for enhancing and rapidly generating training data for unit commitment optimization problems, characterized in that: The following steps are involved: S1: Obtain historical high-time-resolution power system unit commitment input data including the dispatch period set, record it as the day-ahead dataset, and calculate the system net load in the day-ahead dataset; S2: Based on the system net load in the data set before the first day and the preset time series resampling rules, the high-resolution scheduling period set is resampled into a subset of scheduling periods with a set time resolution; S3: Based on the resampled scheduling period subset, the corresponding high-time-resolution renewable energy output data and load forecast data are extracted from the first-day-ahead dataset by time period to form the set time-resolution data, thus obtaining the second-day-ahead dataset. S4: According to the preset data enhancement rules, the set time resolution new energy output data and load forecast data in the second day data set are proportionally adjusted and expanded to form the third day data set; S5: Construct and solve a unit commitment model with a set time resolution that is consistent with the number of samples in the dataset before the third day, record the optimization solution results and their analysis results as new data, and complete the training data enhancement and rapid generation of the unit commitment optimization problem.

2. The method for enhancing and rapidly generating training data for the unit commitment optimization problem according to claim 1, characterized in that: The timing resampling rules preset in S2 include: (1) According to the 1-hour time resolution, the scheduling period set is decomposed into S scheduling period subsets that are associated with the scheduling time resolution and do not overlap with each other, where The first sample Schedule period subset The following scheduling periods are included: in, is the resampling interval, , is the scheduling time resolution, is the scheduling period set, is the subset symbol, is any symbol, Set for scheduling period The subscripts represent the scheduling period set No. , , , Until the scheduling periods; (2) Count the maximum and minimum values of the system net load every hour and form four scheduling period subsets in a certain order. The specific steps include the following: a1: Schedule period set Group by hour, The scheduling period included in the hour for: in, Set for scheduling period No. Scheduling period, subscript The value range is based on Hours decide, ; a2: Statistics Sample No. The periods of maximum and minimum system net load values in all hourly dispatch periods; a3: Based on the time periods of the maximum and minimum system net loads, four scheduling period subsets are formed in the order of maximum-minimum, minimum-maximum, maximum-maximum-minimum, and minimum-minimum-maximum: in, For the The maximum-minimum scheduling period subset in samples, For the The subset of the minimum-maximum scheduling periods in the samples, For the The subset of the maximum-maximum-minimum scheduling periods in the samples, For the The subset of the minimum-minimum-maximum scheduling periods in the samples, and Respectively Sample No. The period of maximum and minimum system net load in all scheduling periods of the hour, subscript and Respectively represent Hour and Hour; Obtained by the timing resampling rule A subset of scheduling periods.

3. The method for enhancing and rapidly generating training data for the unit commitment optimization problem according to claim 2, characterized in that: The S3 includes the following steps: S31: For each scheduling period subset, extract the renewable energy output data and load forecast data for the corresponding period in the data set before the first day to form 1-hour resolution data; The new energy output data includes the new energy station's day-ahead output data. According to the high time resolution new energy station day-ahead output data of the sample Schedule period subset Extract data slices during the included scheduling period: in, For the Sample new energy stations In the Schedule period subset The day-ahead forecast output is For the Sample new energy stations Gather during the scheduled period Time The day-ahead forecast output is It is a collection of new energy stations. For the a subset of scheduling periods; The load forecast data includes load day forecast data, which is obtained by The schedule period subset contains the schedule period extraction data slice: in, Indicates the Sample nodes In the Schedule period subset The load forecast value of the day before is Indicates the Sample nodes Gather during the scheduled period Next period The load forecast value of the day before is is the set of load nodes; S32: Each scheduling period subset on each day constitutes an independent sample, and together constitutes the data set before the next day.

4. The method for enhancing and rapidly generating training data for the unit commitment optimization problem according to claim 3, characterized in that: The S4 includes the following steps: S41: Constructing a set of proportional adjustment factors , the formula is: in, is the preset adjustment factor limit, To adjust the step size, the expansion factor is ; S42: The sample 1-hour time resolution new energy output data and load forecast data in the second day data set, which are expanded from the scheduling period subset of the first day data set, are scaled and expanded according to the proportional adjustment factor: in, and Respectively represent Sample No. New energy stations in the sample expanded from the scheduling period subset In the period Day-ahead forecast output and nodes In the period The load forecast value of the day before is and On this basis, according to New samples scaled by the scaling factor, Indicates the Scaling factors; S43: Each scheduling period subset of each day After scaling and expansion by the proportional adjustment factors, an independent sample is formed, which together constitutes the data set before the third day.

5. The method for enhancing and rapidly generating training data for the unit commitment optimization problem according to claim 4, characterized in that: The objective function of the unit commitment model with a set time resolution that is consistent with the number of samples in the dataset before the third day constructed in S5 is: Among them, in Sample No. The scheduling period subset is The samples are scaled by the scaling factor. is the total operating cost, To participate in the optimization of the collection of units, For the A subset of scheduling periods, 、 and Respectively for units Time The binary variables of the running status, whether the startup event occurs, and whether the shutdown event occurs, 、 and Respectively for units Time No-load costs, startup costs and shutdown costs, For the crew Time Lower and output level The associated variable cost function for electricity generation.

6. A training data enhancement and rapid generation device for unit commitment optimization problems, characterized in that: include: Memory, used to store computer programs; A processor, configured to execute the computer program to implement the method for enhancing and rapidly generating training data for the unit commitment optimization problem according to any one of claims 1 to 5.

Images