Infeasible boundary calculation method and system based on machine learning and operation planning optimization

Abstract

The invention discloses an infeasible boundary calculation method and system based on machine learning and operation planning optimization, and the method comprises the steps: obtaining the current operation boundary data of a to-be-dispatched power grid when a security constraint unit combination is infeasible in the operation process of the to-be-dispatched power grid; inputting the current operation boundary data into a pre-trained machine learning model to obtain a conflict constraint type output by the machine learning model; the machine learning model is obtained based on historical clearing data training, and the historical clearing data is historical data which is infeasible when security constraint unit combination occurs in power grid operation; calculating a minimum infeasible conflict set based on the conflict constraint type; and calculating an infeasible boundary adjustment amount based on the minimum infeasible conflict set and the current operation boundary data, the infeasible boundary adjustment amount being used for adjusting the scheduling scheme of the to-be-scheduled power grid, so that the infeasible boundary in the security constraint unit combination can be quickly and accurately positioned, and the operation boundary parameters in the scheduling scheme can be correspondingly adjusted.

Description

Technical Field

[0001] The present invention relates to the field of power systems, and in particular to an infeasible boundary calculation method and system based on machine learning and operations optimization. Background Art

[0002] In power system dispatch, security-constrained unit commitment (SCUC) refers to determining the combination of generators in the power system while considering various security constraints to ensure the safe and reliable operation of the system. SCUC is usually modeled as a mixed-integer linear programming problem (MILP).

[0003] In actual power grid operation, the constructed SCUC model is usually huge in scale, and may contain millions of variables and constraints. At the same time, there are many types of constraints, and there is a strong coupling between them in time and space. Therefore, when unreasonable power grid operation boundary data appears and makes the SCUC model infeasible, how to quickly and accurately locate the data boundary problem has become a major difficulty in power dispatching today, which makes it difficult to adjust the parameters of the power grid dispatching plan when the SCUC model is infeasible. Summary of the invention

[0004] In order to solve the above technical problems, the embodiments of the present invention propose an infeasible boundary calculation method and system based on machine learning and operations optimization, which can quickly and accurately locate the infeasible boundary in the safety constrained unit combination, and adjust the operating boundary parameters in the scheduling plan accordingly to achieve more efficient and accurate power grid scheduling.

[0005] In order to achieve the above object, an embodiment of the present invention provides an infeasible boundary calculation method based on machine learning and operations optimization, comprising: When the safety constraint unit combination is not feasible during the operation of the power grid to be dispatched, obtaining the current operation boundary data of the power grid to be dispatched; Inputting the current operation boundary data into a pre-trained machine learning model to obtain the conflict constraint type output by the machine learning model; wherein the machine learning model is trained based on historical clearing data, and the historical clearing data is historical data of infeasible safety constraint unit combination during power grid operation; Based on the conflict constraint type, a minimum infeasible conflict set is calculated and obtained; Based on the minimum infeasible conflict set and the current operating boundary data, an infeasible boundary adjustment amount is calculated and obtained, wherein the infeasible boundary adjustment amount is used to adjust the operating boundary parameters in the scheduling plan of the power grid to be scheduled.

[0006] Furthermore, the machine learning model is trained through the following steps: Acquiring the historical clearing data; wherein the historical clearing data includes historical grid boundary data and its corresponding historical conflict constraints; Perform simulation based on the historical clearing data to generate simulation data; Based on the simulation data and the historical clearing data, obtaining a sample data set; Performing feature engineering processing on the sample data set to obtain sample features; The machine learning model to be trained is trained using the sample features to obtain the machine learning model.

[0007] Further, the historical grid boundary data includes historical unit output information, and the simulation data includes unit simulation data; Then, the simulation is performed based on the historical clearing data to generate simulation data, including: Adjusting the unit parameters of the historical unit output information; The unit operation is simulated based on the adjusted historical unit output information to generate unit simulation data.

[0008] Furthermore, obtaining a sample data set based on the simulation data and the historical clearing data includes: Calculating a first empirical distribution function of the historical clearing data, and calculating a second empirical distribution function of the simulation data; Calculating a maximum difference between the first empirical distribution function and the second empirical distribution function; If the maximum difference is greater than a preset significance level critical value, the simulation data and the historical clearing data are combined to obtain a sample data set.

[0009] Furthermore, the performing feature engineering processing on the sample data set to obtain sample features includes: Extracting sample key information from the sample data set as training input features in the sample features; Extracting sample conflict constraint types from the sample data set as training output features in the sample features; The sample key information includes at least one of load quantity and unit operation status, and the sample conflict constraint type includes at least one of power generation constraint and load supply and demand constraint.

[0010] Furthermore, the step of calculating and obtaining a minimum infeasible conflict set based on the conflict constraint type includes: Based on the conflict constraint types, a minimum infeasible conflict set is calculated by a pre-configured integer programming solver; wherein the minimum infeasible conflict set includes multiple types of objective conflict constraints that make the safety constraint unit combination infeasible.

[0011] Furthermore, the current operation boundary data includes operation information corresponding to the multiple types of target conflict constraints; The calculating and obtaining the infeasible boundary adjustment amount based on the minimum infeasible conflict set and the current operation boundary data includes: Perform relaxation calculation on each type of target conflict constraint to obtain the relaxation amount corresponding to each type of target conflict constraint; Based on the relaxation amount corresponding to each type of target conflict constraint and the operation information, the infeasible boundary adjustment amount is calculated.

[0012] The embodiment of the present invention further provides an infeasible boundary computing system based on machine learning and operations optimization, comprising: A data acquisition module, used for acquiring current operation boundary data of the power grid to be dispatched when the safety constraint unit combination is not feasible during the operation of the power grid to be dispatched; A conflict constraint type determination module is used to input the current operation boundary data into a pre-trained machine learning model to obtain the conflict constraint type output by the machine learning model; wherein the machine learning model is trained based on historical clearing data, and the historical clearing data is historical data in which safety constraint unit combination is not feasible during power grid operation; A minimum infeasible conflict set calculation module, used to calculate and obtain a minimum infeasible conflict set based on the conflict constraint type; An infeasible boundary adjustment amount calculation module is used to calculate and obtain an infeasible boundary adjustment amount based on the minimum infeasible conflict set and the current operating boundary data, wherein the infeasible boundary adjustment amount is used to adjust the operating boundary parameters in the scheduling plan of the power grid to be scheduled.

[0013] Furthermore, the machine learning model is trained through the following steps: Acquiring the historical clearing data; wherein the historical clearing data includes historical grid boundary data and its corresponding historical conflict constraints; Perform simulation based on the historical clearing data to generate simulation data; Based on the simulation data and the historical clearing data, obtaining a sample data set; Performing feature engineering processing on the sample data set to obtain sample features; The machine learning model to be trained is trained using the sample features to obtain the machine learning model.

[0014] Further, the historical grid boundary data includes historical unit output information, and the simulation data includes unit simulation data; Then, the simulation is performed based on the historical clearing data to generate simulation data, including: Adjusting the unit parameters of the historical unit output information; The unit operation is simulated based on the adjusted historical unit output information to generate unit simulation data.

[0015] In summary, the present invention has the following beneficial effects: According to an embodiment of the present invention, when a safety-constrained unit combination is infeasible during the operation of the power grid to be dispatched, the current operating boundary data of the power grid to be dispatched is obtained; the current operating boundary data is input into a pre-trained machine learning model to obtain a conflict constraint type output by the machine learning model; wherein the machine learning model is trained based on historical clearing data, and the historical clearing data is historical data of infeasible safety-constrained unit combinations during power grid operation; based on the conflict constraint type, a minimum infeasible conflict set is calculated; based on the minimum infeasible conflict set and the current operating boundary data, an infeasible boundary adjustment amount is calculated, wherein the infeasible boundary adjustment amount is used to adjust the operating boundary parameters in the dispatching scheme of the power grid to be dispatched, so as to quickly and accurately locate the infeasible boundary in the safety-constrained unit combination, and adjust the operating boundary parameters in the dispatching scheme accordingly to achieve more efficient and accurate power grid dispatching. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a flow chart of an embodiment of an infeasible boundary calculation method based on machine learning and operations optimization provided by the present invention; Figure 2 It is a structural schematic diagram of an embodiment of an infeasible boundary computing system based on machine learning and operations optimization provided by the present invention. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0018] In the description of this application, the terms "first", "second", "third", etc. are used for descriptive purposes only and are not to be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first", "second", "third", etc. may explicitly or implicitly include one or more of the feature. In the description of this application, unless otherwise specified, "plurality" means two or more.

[0019] In the description of this application, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0020] In the description of this application, it should be noted that, unless otherwise defined, all technical and scientific terms used in the present invention have the same meaning as those commonly understood by those skilled in the art. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood by specific circumstances.

[0021] See also Figure 1 , is a flow chart of an embodiment of an infeasible boundary calculation method based on machine learning and operations optimization provided by the present invention, the method includes steps S1-S4, which are as follows: S1, when the safety constraint unit combination is not feasible during the operation of the power grid to be dispatched, obtaining the current operation boundary data of the power grid to be dispatched; It should be noted that since the security-constrained unit commitment (SCUC) is usually modeled as a mixed-integer linear programming problem (MILP), the objective function can be to minimize the total cost of power dispatch (including unit power generation cost, unit start-up and shutdown cost, transmission cost and network security cost), and the constraints can be unit output, unit ramping, minimum unit start-up and shutdown time and other unit combination constraints, as well as system balance, system reserve, network flow and other extended constraints. The infeasibility of the security-constrained unit commitment in the operation of the power grid to be dispatched generally refers to the infeasibility of the model caused by a certain type of constraint.

[0022] S2, inputting the current operation boundary data into a pre-trained machine learning model to obtain the conflict constraint type output by the machine learning model; wherein the machine learning model is trained based on historical clearing data, and the historical clearing data is historical data of infeasible safety constraint unit combination in power grid operation; S3, based on the conflict constraint type, calculating and obtaining the minimum infeasible conflict set; S4, based on the minimum infeasible conflict set and the current operating boundary data, calculating and obtaining an infeasible boundary adjustment amount, wherein the infeasible boundary adjustment amount is used to adjust the operating boundary parameters in the scheduling plan of the power grid to be scheduled.

[0023] In an optional implementation, the machine learning model is trained by the following steps: Acquiring the historical clearing data; wherein the historical clearing data includes historical grid boundary data and its corresponding historical conflict constraints; Perform simulation based on the historical clearing data to generate simulation data; Based on the simulation data and the historical clearing data, obtaining a sample data set; Performing feature engineering processing on the sample data set to obtain sample features; The machine learning model to be trained is trained using the sample features to obtain the machine learning model.

[0024] Exemplarily, the obtaining of the historical clearing data includes: collecting historical clearing data with boundary problems, including supply and demand information at the user end and the power generation end, unit output status and output range, new energy predicted output, topological structure and other boundary data, and at the same time collecting conflicting constraints corresponding to each boundary data that make the model infeasible, such as load balance, unit power, power plant power, etc., and cleaning, processing and verifying the data to ensure its reliability in model training and analysis.

[0025] It should be noted that the training of the machine learning model to be trained can be performed by adjusting parameters using models such as neural networks and support vector machines.

[0026] In an optional implementation, the historical grid boundary data includes historical unit output information, and the simulation data includes unit simulation data; Then, the simulation is performed based on the historical clearing data to generate simulation data, including: Adjust the unit parameters of the historical unit output information; wherein the unit parameters may include but are not limited to one or more of the adjustable unit output range and the unit ramp rate, and may also include parameters related to boundary data such as system load; The unit operation is simulated based on the adjusted historical unit output information to generate unit simulation data.

[0027] For example, the unit parameters such as the unit output range and the state can be adjusted according to the generator technical parameters and performance curves in the historical unit output information to make it more consistent with the actual operation. The example is as follows: Assuming that the minimum adjustable output of the unit is 30MW, the maximum adjustable output is 200MW, and the climbing rate is 5MW / min, the first method of generating unit simulation data is to modify the adjustable output range, for example, the minimum adjustable output value is increased by 0% to 50% in intervals of 10%, and 5 sets of simulation data with minimum adjustable outputs of 30MW, 33MW, 36MW, 42MW, and 45MW are obtained respectively; the second method is to modify the unit climbing rate, and the corresponding simulation data generation idea is the same as above.

[0028] In an optional implementation, the acquiring a sample data set based on the simulation data and the historical clearing data includes: Calculating a first empirical distribution function of the historical clearing data, and calculating a second empirical distribution function of the simulation data; Calculating a maximum difference between the first empirical distribution function and the second empirical distribution function; If the maximum difference is greater than a preset significance level critical value, the simulation data and the historical clearing data are combined to obtain a sample data set.

[0029] For example, by verifying the distribution characteristics of the simulation data, it is ensured that the training effect of the subsequent model will not be affected while enriching the sample data set. For example, the KS test method can be used to verify whether the simulation data and the historical clearing data as the original data come from the same probability distribution, thereby ensuring the rationality of the generated data, as shown in formula (1): ; (1) in, is the statistic of the KS test, and are the first empirical distribution function of historical clearing data and the second empirical distribution function of simulation data respectively; Determined by calculating the maximum difference between the distribution functions , and finally by comparing The significance level critical value corresponding to the given significance level is used to determine whether the two come from the same distribution.

[0030] In this embodiment, after the above-mentioned data generation and verification process, the simulation data is finally fused with the real historical data to construct a more diverse and comprehensive data set to expand the training data of the machine learning model and improve its generalization ability in diverse situations.

[0031] In an optional implementation, performing feature engineering processing on the sample data set to obtain sample features includes: Extracting sample key information from the sample data set as training input features in the sample features; Extracting sample conflict constraint types from the sample data set as training output features in the sample features; The sample key information includes at least one of load quantity and unit operation status, and the sample conflict constraint type includes at least one of power generation constraint and load supply and demand constraint.

[0032] It should be noted that the extracted sample conflict constraint types may include the power plant power lower limit being too high, the power plant power upper limit being too high, the unit power lower limit being too high, the unit power upper limit being too high, and load supply and demand imbalance.

[0033] In an optional implementation, the calculating and obtaining a minimum infeasible conflict set based on the conflict constraint type includes: Based on the conflict constraint types, a minimum infeasible conflict set is calculated by a pre-configured integer programming solver; wherein the minimum infeasible conflict set includes multiple types of objective conflict constraints that make the safety constraint unit combination infeasible.

[0034] Specifically, in this embodiment, before calculating the minimum infeasible conflict set, the integer programming solver can be configured so that the calculated minimum infeasible conflict set must include the conflict constraint type. For example, in Gurobi (a new generation of large-scale mathematical programming optimizer), the Force parameter can be set to 1 so that the solver focuses on such constraints and boundary variables, thereby accelerating the calculation process of the minimum infeasible conflict set in the integer programming solver.

[0035] In an optional implementation, the current operation boundary data includes operation information corresponding to the multiple types of target conflict constraints; The calculating and obtaining the infeasible boundary adjustment amount based on the minimum infeasible conflict set and the current operation boundary data includes: Perform relaxation calculation on each type of target conflict constraint to obtain the relaxation amount corresponding to each type of target conflict constraint; Based on the relaxation amount corresponding to each type of target conflict constraint and the operation information, the infeasible boundary adjustment amount is calculated.

[0036] For example, in the provincial load balance constraint, in theory, the sum of the output of all generators in a certain electrical province, the amount of electricity delivered by the province, the amount of electricity received by the province, and the electricity load of the province should satisfy the corresponding balance relationship. However, when the boundary data is unreasonable and causes load imbalance, the corresponding imbalance is often calculated by relaxing the constraint, as shown in formula (2), ; (2) in, Indicates the unit In time The output power, Indicates contact line In time The output power, Ministry of Electricity In time The power load, Indicates province In time The load positive relaxation, Indicates province In time The negative relaxation of the load, Indicates province The unit id in Indicates the sending end contact line id, Indicates the receiving end contact line id; Finally, the model is solved and The specific load relaxation amount is obtained, and then the corresponding boundary data is adjusted.

[0037] Preferably, the method further comprises: Whenever the safety-constrained unit combination of the power grid to be dispatched is not feasible during operation, the acquired current operating boundary data of the power grid to be dispatched is added to the sample data set to achieve continuous training and improvement of the machine learning model.

[0038] Correspondingly, an embodiment of the present invention also provides an infeasible boundary calculation system based on machine learning and operations optimization, which can implement all processes of the infeasible boundary calculation method based on machine learning and operations optimization provided in the above embodiments.

[0039] See also Figure 2 , is a structural diagram of an embodiment of an infeasible boundary computing system based on machine learning and operations optimization provided by the present invention, including: The data acquisition module 101 is used to acquire the current operation boundary data of the power grid to be dispatched when the safety constraint unit combination is not feasible during the operation of the power grid to be dispatched; A conflict constraint type determination module 102 is used to input the current operation boundary data into a pre-trained machine learning model to obtain the conflict constraint type output by the machine learning model; wherein the machine learning model is trained based on historical clearing data, and the historical clearing data is historical data of infeasible safety constraint unit combination during power grid operation; A minimum infeasible conflict set calculation module 103, used to calculate and obtain a minimum infeasible conflict set based on the conflict constraint type; The infeasible boundary adjustment amount calculation module 104 is used to calculate and obtain the infeasible boundary adjustment amount based on the minimum infeasible conflict set and the current operating boundary data, wherein the infeasible boundary adjustment amount is used to adjust the operating boundary parameters in the scheduling plan of the power grid to be scheduled.

[0040] In an optional implementation, the machine learning model is trained by the following steps: Acquiring the historical clearing data; wherein the historical clearing data includes historical grid boundary data and its corresponding historical conflict constraints; Perform simulation based on the historical clearing data to generate simulation data; Based on the simulation data and the historical clearing data, obtaining a sample data set; Performing feature engineering processing on the sample data set to obtain sample features; The machine learning model to be trained is trained using the sample features to obtain the machine learning model.

[0041] In an optional implementation, the historical grid boundary data includes historical unit output information, and the simulation data includes unit simulation data; Then, the simulation is performed based on the historical clearing data to generate simulation data, including: Adjusting the unit parameters of the historical unit output information; The unit operation is simulated based on the adjusted historical unit output information to generate unit simulation data.

[0042] In an optional implementation, the acquiring a sample data set based on the simulation data and the historical clearing data includes: Calculating a first empirical distribution function of the historical clearing data, and calculating a second empirical distribution function of the simulation data; Calculating a maximum difference between the first empirical distribution function and the second empirical distribution function; If the maximum difference is greater than a preset significance level critical value, the simulation data and the historical clearing data are combined to obtain a sample data set.

[0043] In an optional implementation, performing feature engineering processing on the sample data set to obtain sample features includes: Extracting sample key information from the sample data set as training input features in the sample features; Extracting sample conflict constraint types from the sample data set as training output features in the sample features; The sample key information includes at least one of load quantity and unit operation status, and the sample conflict constraint type includes at least one of power generation constraint and load supply and demand constraint.

[0044] In an optional implementation, the calculating and obtaining a minimum infeasible conflict set based on the conflict constraint type includes: Based on the conflict constraint types, a minimum infeasible conflict set is calculated by a pre-configured integer programming solver; wherein the minimum infeasible conflict set includes multiple types of objective conflict constraints that make the safety constraint unit combination infeasible.

[0045] In an optional implementation, the current operation boundary data includes operation information corresponding to the multiple types of target conflict constraints; The calculating and obtaining the infeasible boundary adjustment amount based on the minimum infeasible conflict set and the current operation boundary data includes: Perform relaxation calculation on each type of target conflict constraint to obtain the relaxation amount corresponding to each type of target conflict constraint; Based on the relaxation amount corresponding to each type of target conflict constraint and the operation information, the infeasible boundary adjustment amount is calculated.

[0046] In summary, the present invention has the following beneficial effects: According to an embodiment of the present invention, when a safety-constrained unit combination is infeasible during the operation of the power grid to be dispatched, the current operating boundary data of the power grid to be dispatched is obtained; the current operating boundary data is input into a pre-trained machine learning model to obtain a conflict constraint type output by the machine learning model; wherein the machine learning model is trained based on historical clearing data, and the historical clearing data is historical data of infeasible safety-constrained unit combinations during power grid operation; based on the conflict constraint type, a minimum infeasible conflict set is calculated; based on the minimum infeasible conflict set and the current operating boundary data, an infeasible boundary adjustment amount is calculated, wherein the infeasible boundary adjustment amount is used to adjust the operating boundary parameters in the dispatching scheme of the power grid to be dispatched, so as to quickly and accurately locate the infeasible boundary in the safety-constrained unit combination, and adjust the operating boundary parameters in the dispatching scheme accordingly to achieve more efficient and accurate power grid dispatching.

[0047] Through the description of the above implementation methods, those skilled in the art can clearly understand that the present invention can be implemented by means of software plus a necessary hardware platform, and of course can also be implemented entirely by hardware. Based on such an understanding, all or part of the contribution of the technical solution of the present invention to the background technology can be embodied in the form of a software product, and the computer software product can be stored in a storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., including a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments of the present invention or certain parts of the embodiments.

[0048] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for calculating the infeasible boundary based on machine learning and operations optimization, characterized in that: include: When the safety constraint unit combination is not feasible during the operation of the power grid to be dispatched, obtaining the current operation boundary data of the power grid to be dispatched; Inputting the current operation boundary data into a pre-trained machine learning model to obtain the conflict constraint type output by the machine learning model; wherein the machine learning model is trained based on historical clearing data, and the historical clearing data is historical data of infeasible safety constraint unit combination during power grid operation; Based on the conflict constraint type, a minimum infeasible conflict set is calculated and obtained; Based on the minimum infeasible conflict set and the current operating boundary data, an infeasible boundary adjustment amount is calculated and obtained, wherein the infeasible boundary adjustment amount is used to adjust the operating boundary parameters in the scheduling plan of the power grid to be scheduled.

2. The infeasible boundary calculation method based on machine learning and operations optimization according to claim 1, characterized in that: The machine learning model is trained through the following steps: Acquiring the historical clearing data; wherein the historical clearing data includes historical grid boundary data and its corresponding historical conflict constraints; Perform simulation based on the historical clearing data to generate simulation data; Based on the simulation data and the historical clearing data, obtaining a sample data set; Performing feature engineering processing on the sample data set to obtain sample features; The machine learning model to be trained is trained using the sample features to obtain the machine learning model.

3. The infeasible boundary calculation method based on machine learning and operations optimization according to claim 2, characterized in that: The historical grid boundary data includes historical unit output information, and the simulation data includes unit simulation data; Then, the simulation is performed based on the historical clearing data to generate simulation data, including: Adjusting the unit parameters of the historical unit output information; The unit operation is simulated based on the adjusted historical unit output information to generate unit simulation data.

4. The infeasible boundary calculation method based on machine learning and operations optimization according to claim 2, characterized in that: The acquiring of a sample data set based on the simulation data and the historical clearing data includes: Calculating a first empirical distribution function of the historical clearing data, and calculating a second empirical distribution function of the simulation data; Calculating a maximum difference between the first empirical distribution function and the second empirical distribution function; If the maximum difference is greater than a preset significance level critical value, the simulation data and the historical clearing data are combined to obtain a sample data set.

5. The infeasible boundary calculation method based on machine learning and operations optimization according to claim 2, characterized in that: The performing feature engineering processing on the sample data set to obtain sample features includes: Extracting sample key information from the sample data set as training input features in the sample features; Extracting sample conflict constraint types from the sample data set as training output features in the sample features; The sample key information includes at least one of load quantity and unit operation status, and the sample conflict constraint type includes at least one of power generation constraint and load supply and demand constraint.

6. The infeasible boundary calculation method based on machine learning and operations optimization according to claim 1, characterized in that: The step of calculating and obtaining a minimum infeasible conflict set based on the conflict constraint type includes: Based on the conflict constraint types, a minimum infeasible conflict set is calculated by a pre-configured integer programming solver; wherein the minimum infeasible conflict set includes multiple types of objective conflict constraints that make the safety constraint unit combination infeasible.

7. The infeasible boundary calculation method based on machine learning and operations optimization according to claim 6, characterized in that: The current operation boundary data includes operation information corresponding to the multiple types of target conflict constraints; The calculating and obtaining the infeasible boundary adjustment amount based on the minimum infeasible conflict set and the current operation boundary data includes: Perform relaxation calculation on each type of target conflict constraint to obtain the relaxation amount corresponding to each type of target conflict constraint; Based on the relaxation amount corresponding to each type of target conflict constraint and the operation information, the infeasible boundary adjustment amount is calculated.

8. An infeasible boundary computing system based on machine learning and operations optimization, characterized in that: include: A data acquisition module, used for acquiring current operation boundary data of the power grid to be dispatched when the safety constraint unit combination is not feasible during the operation of the power grid to be dispatched; A conflict constraint type determination module is used to input the current operation boundary data into a pre-trained machine learning model to obtain the conflict constraint type output by the machine learning model; wherein the machine learning model is trained based on historical clearing data, and the historical clearing data is historical data in which safety constraint unit combination is not feasible during power grid operation; A minimum infeasible conflict set calculation module, used to calculate and obtain a minimum infeasible conflict set based on the conflict constraint type; An infeasible boundary adjustment amount calculation module is used to calculate and obtain an infeasible boundary adjustment amount based on the minimum infeasible conflict set and the current operating boundary data, wherein the infeasible boundary adjustment amount is used to adjust the operating boundary parameters in the scheduling plan of the power grid to be scheduled.

9. The infeasible boundary computing system based on machine learning and operations optimization as claimed in claim 8, characterized in that: The machine learning model is trained through the following steps: Acquiring the historical clearing data; wherein the historical clearing data includes historical grid boundary data and its corresponding historical conflict constraints; Perform simulation based on the historical clearing data to generate simulation data; Based on the simulation data and the historical clearing data, obtaining a sample data set; Performing feature engineering processing on the sample data set to obtain sample features; The machine learning model to be trained is trained using the sample features to obtain the machine learning model.

10. The infeasible boundary computing system based on machine learning and operations optimization according to claim 9, characterized in that: The historical grid boundary data includes historical unit output information, and the simulation data includes unit simulation data; Then, the simulation is performed based on the historical clearing data to generate simulation data, including: Adjusting the unit parameters of the historical unit output information; The unit operation is simulated based on the adjusted historical unit output information to generate unit simulation data.

Images