Inertia demand evaluation method and device based on data driving and unit combination method and device containing inertia constraint

Abstract

The invention provides an inertia demand evaluation method and device based on data driving and an inertia constraint-containing unit combination method and device. The method comprises the following steps: establishing and solving a unit commitment model containing a frequency security constraint according to a frequency change rate constraint and a maximum frequency deviation constraint, and determining a system minimum inertia demand of each time period by using a system minimum inertia demand calculation model according to a unit commitment result obtained by solving so as to construct a system inertia demand data set; training a machine learning model based on the system inertia demand data set, predicting the minimum inertia demand of the system by adopting the trained machine learning model, introducing the obtained prediction result as the minimum inertia demand constraint into the unit commitment model to replace the frequency safety constraint, and constructing the unit commitment model containing the minimum inertia constraint; and solving the unit commitment model containing the minimum inertia constraint to obtain an optimized unit commitment result. According to the method, unit combination problem calculation efficiency and power system frequency safety are both considered.

Description

TECHNICAL FIELD

[0001] The embodiment of the present application relates to the technical field of power system optimization scheduling, and particularly relates to a data-driven inertia demand evaluation method and a unit commitment method with inertia constraints. BACKGROUND

[0002] With large-scale access of new energy such as wind power and photovoltaic power, the output proportion of traditional synchronous generators in the power system is reduced, the equivalent inertia level of the system is significantly reduced, and the frequency stability problem is increasingly prominent. In frequency safety analysis, the frequency response performance is usually characterized by the frequency change rate (RoCoF) and the frequency nadir (Nadir), and the minimum inertia demand is derived to a certain extent. However, the existing methods mostly rely on strict physical modeling and nonlinear constraint expression, and the optimization problem is large in scale and high in difficulty. For example, existing researches propose to explicitly introduce frequency safety constraints in the unit commitment model, but this will bring non-convex and mixed integer modeling form, resulting in too long calculation time and unable to meet the requirement of fast result in actual scheduling. Another type of method is to evaluate the frequency safety after solving the unit commitment, but this post-checking method cannot directly guarantee the safety in the optimization process, and is easy to lead to the infeasibility of the optimization result.

[0003] In summary, the existing methods have the following main shortcomings: (1) The explicit modeling of frequency constraints is complex, increases the number of integer variables, and makes the unit commitment problem difficult to be solved within a limited time; (2) The simulation-based post-checking method lacks feedforward, and the optimization result cannot guarantee frequency safety; (3) The existing scheduling data and prediction information are not effectively utilized for data-driven modeling, which limits the generalization and expansion ability of the method. SUMMARY

[0004] Therefore, the embodiment of the present application provides a data-driven inertia demand evaluation method and a unit commitment method with inertia constraints, which introduces a machine learning model to predict the minimum inertia demand, replaces the complex frequency safety constraints with the minimum inertia demand as constraints, and realizes the balance between safety and calculation efficiency.

[0005] According to a first aspect of an embodiment of the present application, a data-driven inertia demand evaluation and unit commitment method with inertia constraint is provided, comprising: establishing a unit commitment model with frequency security constraints and solving the model according to frequency rate of change constraints and maximum frequency deviation constraints to obtain a unit commitment result; determining system minimum inertia demand of each period by using a system minimum inertia demand calculation model according to the unit commitment result, to construct a system inertia demand data set; training a machine learning model based on the system inertia demand data set to obtain a trained machine learning model; predicting the system minimum inertia demand by using the trained machine learning model, and introducing the obtained prediction result as minimum inertia demand constraint into the unit commitment model to replace the frequency rate of change constraints and the maximum frequency deviation constraints, to construct a unit commitment model with minimum inertia constraint; and solving the unit commitment model with minimum inertia constraint to obtain an optimized unit commitment result.

[0006] In an implementation manner, the unit commitment model with frequency security constraints takes minimization of total operation cost of thermal power units and maximization of new energy consumption as an optimization objective, and an expression thereof is as follows:

[0007] wherein, C is total cost of unit commitment operation; T is total number of unit commitment operation time periods; Ω G is all thermal power unit set; a g , b g , c g respectively represent quadratic, linear and constant terms of the thermal power unit g generation cost curve; P g,t represents output of the thermal power unit g at time point t ; u g,t represents operation state of the thermal power unit g at time point t , u g,t =1 represents that the thermal power unit g is in start-up state at time point t , otherwise, the thermal power unit is in shutdown state; v g,t and w g,t are start-stop variables of the thermal power unit, v g,t =1 represents that the thermal power unit g is in start-up state at time point tTransitioning from shutdown to startup. v g,t =1 indicates a thermal power unit g At any moment t Transitioning from shutdown to startup. w g,t =1 indicates a thermal power unit g At any moment t From power-on state to shutdown state; and thermal power units g Start-stop fees; C PFR It is the standby cost coefficient for primary frequency regulation; For the unit i At any moment t Primary frequency regulation reserve capacity; Ω PFR It is the collection of all primary frequency regulating units, including thermal power units and wind power units.

[0008] In another implementation, the constraints of the unit combination model with frequency security constraints include power balance constraints, thermal power unit output limit constraints, thermal power unit minimum start-up and shutdown time constraints, thermal power unit ramping constraints, thermal power unit operating state logic constraints, new energy unit output limit constraints, frequency change rate constraints, maximum frequency deviation constraints, and network power flow constraints.

[0009] In another implementation, the calculation model for the minimum inertia requirement of the system is as follows:

[0010]

[0011]

[0012]

[0013]

[0014] in, H min To meet the minimum inertia requirement of the system, It is the minimum inertia required under the constraint of the rate of change of frequency. It is the minimum inertia required under the maximum frequency deviation constraint.

[0015] According to a second aspect of the present invention, a data-driven inertia demand assessment and unit combination device with inertia constraints is provided, comprising: a unit combination model construction module with frequency safety constraints, configured to establish and solve a unit combination model with frequency safety constraints based on frequency change rate constraints and maximum frequency deviation constraints to obtain unit combination results; a dataset construction module, configured to determine the system minimum inertia demand for each time period based on the unit combination results using a system minimum inertia demand calculation model to construct a system inertia demand dataset; a training module, configured to train a machine learning model based on the system inertia demand dataset to obtain a trained machine learning model; a unit combination model construction module with minimum inertia constraints, configured to predict the system minimum inertia demand using the trained machine learning model, and introduce the obtained prediction results as minimum inertia demand constraints into the unit combination model to replace the frequency change rate constraints and the maximum frequency deviation constraints to construct a unit combination model with minimum inertia constraints; and a solution module, configured to solve the unit combination model with minimum inertia constraints to obtain optimized unit combination results.

[0016] In one implementation, the unit combination model with frequency security constraints aims to minimize the total operating cost of thermal power units while maximizing the absorption of new energy sources, and its expression is:

[0017] in, C It is the total cost of operating the units in combination; T It represents the total number of time periods during which the generating units operate in combination; Ω G It is the collection of all thermal power units; a g , b g , c g They represent thermal power units g The quadratic, linear, and constant terms of the power generation cost curve; P g,t Indicates thermal power unit g At any moment t contribution; u g,t Indicates thermal power unit g At any moment t The running status, u g,t =1 indicates a thermal power unit g At any moment t It is powered on; otherwise, it is powered off. v g,t and w g,t These are the start-up and shutdown variables for thermal power units.v g,t =1 indicates a thermal power unit g At any moment t Transitioning from shutdown to startup. v g,t =1 indicates a thermal power unit g At any moment t Transitioning from shutdown to startup. w g,t =1 indicates a thermal power unit g At any moment t From power-on state to shutdown state; and thermal power units g Start-stop fees; C PFR It is the standby cost coefficient for primary frequency regulation; For the unit i At any moment t Primary frequency regulation reserve capacity; Ω PFR It is the collection of all primary frequency regulating units, including thermal power units and wind power units.

[0018] In another implementation, the constraints of the unit combination model with frequency security constraints include power balance constraints, thermal power unit output limit constraints, thermal power unit minimum start-up and shutdown time constraints, thermal power unit ramping constraints, thermal power unit operating state logic constraints, new energy unit output limit constraints, frequency change rate constraints, maximum frequency deviation constraints, and network power flow constraints.

[0019] In another implementation, the calculation model for the minimum inertia requirement of the system is as follows:

[0020]

[0021]

[0022]

[0023]

[0024] in, H min To meet the minimum inertia requirement of the system, It is the minimum inertia required under the constraint of the rate of change of frequency. It is the minimum inertia required under the maximum frequency deviation constraint.

[0025] According to a third aspect of the present invention, an electronic device is provided, including a processor and a memory storing a program. The program includes instructions that, when executed by the processor, cause the processor to perform the steps performed by the method of the first aspect described above.

[0026] According to a fourth aspect of the present invention, a computer storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the method of the first aspect described above.

[0027] Compared with existing schemes that rely on explicit frequency constraints or post-simulation verification, the present invention has the following advantages: (1) This invention significantly reduces model complexity, avoids the introduction of nonlinear constraints and large-scale integer variables, and greatly improves the computational efficiency of unit combination problems; (2) This invention utilizes a machine learning model to achieve rapid prediction of minimum inertia requirements, which can adapt to the uncertainty of new energy sources and load forecasting and improve the real-time performance of the method. (3) By extending feature construction and data-driven modeling, the model of this invention has stronger generalization ability and robustness; (4) This invention improves the utilization rate of new energy sources and the economy of power system dispatch while ensuring frequency security. Attached Figure Description

[0028] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0029] Figure 1 This is a flowchart of the steps of the method of the present invention. Detailed Implementation

[0030] To provide a clearer understanding of the technical features, objectives, and effects of the embodiments of the present invention, specific implementation methods of the embodiments of the present invention will now be described with reference to the accompanying drawings.

[0031] In this document, “exemplary” means “serving as an example, illustration or description”, and any illustrations or implementations described herein as “exemplary” should not be construed as a more preferred or advantageous technical solution.

[0032] To enable those skilled in the art to better understand the technical solutions in the embodiments of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art should fall within the protection scope of the present invention.

[0033] The specific implementation of the embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0034] See Figure 1 The present invention provides a data-driven method for inertia demand assessment and unit combination with inertia constraints, which mainly includes the following steps: Step S1: Based on the frequency change rate constraint and the maximum frequency deviation constraint, establish and solve the unit combination model with frequency safety constraints to obtain the unit combination results; Step S2: Based on the unit combination results, determine the minimum inertia requirement of the system for each time period using the system minimum inertia requirement calculation model, so as to construct a system inertia requirement dataset; Step S3: Train a machine learning model based on the system inertia requirement dataset to obtain a trained machine learning model; Step S4: Use the trained machine learning model to predict the minimum inertia requirement of the system, and introduce the prediction result as the minimum inertia requirement constraint into the unit combination model to replace the frequency change rate constraint and the maximum frequency deviation constraint, so as to construct a unit combination model with minimum inertia constraint. Step S5: Solve the unit combination model with minimum inertia constraint to obtain the optimized unit combination result.

[0035] In one implementation, the unit combination model with frequency security constraints aims to minimize the total operating cost of thermal power units while maximizing the absorption of new energy sources, and its expression is:

[0036] in, C It is the total cost of operating the units in combination; T It represents the total number of time periods during which the generating units operate in combination; Ω G It is the collection of all thermal power units; a g , b g , c g They represent thermal power units g The quadratic, linear, and constant terms of the power generation cost curve; Pg,t Indicates thermal power unit g At any moment t contribution; u g,t Indicates thermal power unit g At any moment t The running status, u g,t =1 indicates a thermal power unit g At any moment t It is powered on; otherwise, it is powered off. v g,t and w g,t These are the start-up and shutdown variables for thermal power units. v g,t =1 indicates a thermal power unit g At any moment t Transitioning from shutdown to startup. v g,t =1 indicates a thermal power unit g At any moment t Transitioning from shutdown to startup. w g,t =1 indicates a thermal power unit g At any moment t From power-on state to shutdown state; and thermal power units g Start-stop fees; C PFR It is the standby cost coefficient for primary frequency regulation; For the unit i At any moment t Primary frequency regulation reserve capacity; Ω PFR It is the collection of all primary frequency regulating units, including thermal power units and wind power units.

[0037] In another implementation, the constraints of the unit combination model with frequency security constraints include power balance constraints, thermal power unit output limit constraints, thermal power unit minimum start-up and shutdown time constraints, thermal power unit ramping constraints, thermal power unit operating state logic constraints, new energy unit output limit constraints, frequency change rate constraints, maximum frequency deviation constraints, and network power flow constraints.

[0038] In another implementation, the calculation model for the minimum inertia requirement of the system is as follows:

[0039]

[0040]

[0041]

[0042]

[0043] in, H min To meet the minimum inertia requirement of the system, It is the minimum inertia required under the constraint of the rate of change of frequency. It is the minimum inertia required under the maximum frequency deviation constraint.

[0044] Specifically, the solution of the present invention is further described according to the following embodiments: Existing unit combination studies generally fail to adequately consider frequency security issues, or only verify them through post-hoc simulations. This results in solutions that cannot guarantee the security of system inertia levels under conditions of large-scale renewable energy integration. Even when some studies introduce frequency constraints, the modeling process is usually complex and computationally intensive, making it difficult to solve the optimization problem quickly and hindering scheduling decisions in practical engineering. Therefore, the technical problem this invention aims to solve is: how to quickly and accurately obtain the minimum inertia requirement of the system while ensuring system frequency security, and efficiently incorporate it into the unit combination optimization model, thereby improving the scheduling efficiency and security of power systems with a high proportion of renewable energy.

[0045] This invention provides a data-driven method for inertia demand assessment and unit combination with inertia constraints. First, a unit combination model with frequency safety constraints is established. By introducing inertia-related constraints, such as frequency change rate constraints and maximum frequency deviation constraints, into the unit combination model, the optimized result, i.e., the unit combination result, meets the requirements for safe system frequency operation. Based on the unit combination result, the corresponding minimum inertia demand is further derived using a system minimum inertia demand calculation model, thereby forming training samples and constructing a system inertia demand dataset. Finally, a machine learning model is trained based on the constructed dataset to achieve rapid prediction of the system minimum inertia demand. The prediction results can be fed back into the unit combination model to establish a unit combination model with minimum inertia constraints, thereby improving the model's computational efficiency and real-time performance.

[0046] The effect of system active power regulation on frequency change can be described by the rotor motion equation: (1) In the formula, H sys It is the total inertia of the system, including the rotational inertia of the thermal power unit and the virtual inertia of the wind power unit; f N It is the system's rated frequency; Δ f ( t ) is a moment tThe system's frequency deviation value; D t It is the system damping, for time... t The product of the system load and the load damping coefficient; It is a moment t Frequency modulation power provided by all system resources; Δ P DIS It is the disturbance power experienced by the system.

[0047] In frequency security analysis, two inertia-related metrics are typically considered: the rate of change of frequency (RoCoF) and the maximum frequency deviation (Nadir Frequency).

[0048] At the instant a fault occurs, the rate of frequency change is at its maximum, and its calculation formula is: (2) In the formula, This represents the system's maximum frequency change rate.

[0049] When the upper limit of the given system frequency change rate index is RoCoF max When the frequency change rate constraint is met, the minimum inertia required by the system is obtained. This constraint ensures that the frequency of the system does not drop too quickly in the initial stage after a disturbance. The calculation formula is as follows: (3) When the frequency drops to its lowest point, the system enters a transient extreme point, at which point the rate of change of frequency satisfies: (4) Substituting equation (4) into equation (1), we obtain the formula for calculating the maximum frequency deviation: (5) In the formula, It's an FM dead zone. T d This is the dynamic time constant for primary frequency modulation.

[0050] Given the maximum system frequency deviation Δ f max When the minimum inertia required by the system under the maximum frequency deviation constraint is obtained, this constraint ensures that the lowest frequency point of the system is still above the safety threshold. The calculation formula is as follows: (6) Substituting equations (3) and (6) as frequency security constraints into the unit combination model yields a unit combination model with frequency security constraints. Its optimization objective is to minimize the total operating cost of thermal power units while maximizing the absorption of new energy sources. The specific expression is: (7) In the formula, C It is the total cost of operating the units in combination; T It represents the total number of time periods during which the generating units operate in combination; Ω G It is the collection of all thermal power units; a g , b g , c g They represent thermal power units g The quadratic, linear, and constant terms of the power generation cost curve; P g,t Indicates thermal power unit g At any moment t contribution; u g,t Indicates thermal power unit g At any moment t The running status, u g,t =1 indicates a thermal power unit g At any moment t It is powered on; otherwise, it is powered off. v g,t and w g,t These are the start-up and shutdown variables for thermal power units. v g,t =1 indicates a thermal power unit g At any moment t Transitioning from shutdown to startup. v g,t =1 indicates a thermal power unit g At any moment t Transitioning from shutdown to startup. w g,t =1 indicates a thermal power unit g At any moment t From power-on state to shutdown state; and thermal power units g Start-stop fees; C PFR It is the standby cost coefficient for primary frequency regulation; For the unit i At any moment t Primary frequency regulation reserve capacity; Ω PFR It is the collection of all primary frequency regulating units, including thermal power units and wind power units.

[0051] The constraints of the unit combination model with frequency safety constraints include power balance constraints, thermal power unit output limit constraints, thermal power unit minimum start-up and shutdown time constraints, thermal power unit ramping constraints, thermal power unit operating state logic constraints, new energy unit output limit constraints, frequency change rate constraints, maximum frequency deviation constraints, and network power flow constraints. 1) Power balance constraints: (8) In the formula, P w,t Indicates wind turbine w At any moment t of efforts, P s,t Indicates photovoltaic unit s At any moment t of efforts, P d,t Indicates load d At any moment t The load demand, Ω W Ω S Ω D These are collections of all wind turbines, photovoltaic units, and loads.

[0052] 2) Output limit constraints of thermal power units: (9) In the formula, , They represent thermal power units g The upper and lower limits of output.

[0053] 3) Minimum start-up and shutdown time constraints for thermal power units: (10) (11) In the formula, , They represent thermal power units g The duration of power-on and power-off, , They represent thermal power units g Minimum startup and shutdown times.

[0054] 4) Gradient constraints for thermal power units: (12) (13) In the formula, , They represent thermal power units g Maximum upward and downward climbing power.

[0055] 5) Logical constraints on the operating status of thermal power units: (14) (15) 6) Output limit constraints of new energy units: (16) (17) In the formula, Indicates wind turbine w At any moment t contribution; Indicates wind turbine w At any moment t The predicted output, considering the participation of wind turbine units in primary frequency regulation, requires the reserve capacity of primary frequency regulation. Indicates photovoltaic unit s At any moment t contribution; Indicates photovoltaic unit s At any moment t The predicted output.

[0056] 7) Run standby constraints: (18) (19) (20) In the formula, α The percentage of load that is in reserve for operation; u w,t Indicates wind turbine w At any moment t The running status, u w,t =1 indicates a wind turbine generator. w At any moment t It is powered on; otherwise, it is powered off. Indicates wind turbine w The predicted output.

[0057] 8) Frequency security constraints: (twenty one) (twenty two) (twenty three) In the formula, H g , H w They represent thermal power units g Wind turbinew The inertial time constant, Indicates wind turbine w The installed capacity.

[0058] 9) Network power flow constraints: Considering safety and stability factors such as line thermal limits, at any time t, the power flow corresponding to the sum of the electrical quantities flowing through each branch should not exceed its available transmission capacity, that is: (twenty four) In the formula, branch road l The limits of current transmission; , , , These represent the nodes from thermal power units, wind power units, photovoltaic units, and loads to the branch lines, respectively. l The power transfer distribution factor.

[0059] The optimization results obtained from the unit combination model with frequency safety constraints include the start-up and shutdown status and output level of the units in each time period. From this, the primary frequency regulation capacity that the system units can provide in each time period can be calculated. Then, the minimum inertia requirement of the system is further calculated using the system minimum inertia requirement calculation model. Specifically, the required minimum inertia is calculated using the aforementioned frequency change rate constraint and maximum frequency deviation constraint respectively. The larger value of the required inertia under the two constraints should be selected as the system minimum inertia requirement. The system minimum inertia requirement calculation model is as follows: (25) (26) (27) (28) (29) In the formula, H min To meet the minimum inertia requirement of the system, , These are the minimum inertia required under the system frequency change rate constraint and the maximum frequency deviation constraint, respectively.

[0060] Based on the optimization results and the calculation of the system's minimum inertia requirement, a dataset is further constructed for machine learning modeling. Load forecasts, wind power forecasts, and photovoltaic forecasts for each time period are extracted from the forecast data as input features, and the calculated system minimum inertia requirement for each time period is used as the prediction target value. The input features are mapped to the prediction target values ​​to obtain a dataset covering multiple typical days and all time periods.

[0061] This invention employs an eXtreme Gradient Boosting (XGBoost) model. This model iteratively trains multiple regression trees and minimizes the second-order approximation of the objective function in each iteration, thereby ensuring prediction accuracy while exhibiting strong generalization ability. Its basic idea is to... t In the next iteration, the predicted value Expression before t The prediction results for -1 tree and the current newly added tree The superposition of, that is: (30) In the formula, The function space representing the regression tree. The objective function of XGBoost consists of the training error and a regularization term: (31) In the formula, This is the loss function, used to measure the error between the predicted value and the true value. This is a regularization term used to control model complexity. Through second-order Taylor expansion, XGBoost transforms the above optimization problem into a weighted minimization problem of the first-order and second-order gradients, thereby significantly improving training efficiency.

[0062] In traditional XGBoost models, the loss function typically employs a symmetric loss function such as mean squared error (MSE). However, when the predicted system inertia demand is lower than the actual demand, it poses a risk to system frequency security. To address this issue, this invention improves the loss function of the XGBoost model. The core idea of ​​the improved model is to enhance its sensitivity to inertia underestimation by amplifying the gradient of underestimated samples and increasing the penalty for the number of underestimations, thereby strengthening system frequency security.

[0063] Specifically, a weighting coefficient is introduced based on the squared loss. α When the predicted value is less than the actual value, i.e., inertia is underestimated, the gradient of the underestimated samples is amplified, making the model pay more attention to underestimation and increasing the learning intensity on underestimated samples, thereby reducing the probability and severity of underestimation. Simultaneously, a penalty term for the number of underestimations is added. β This penalizes the number of underestimations, further strengthening the model's learning of underestimation scenarios. The specific weighted asymmetric loss function expression is as follows: (32) In the formula, The function is an indicator function; its value is 1 when the condition within the parentheses is true, and 0 otherwise. In this way, the value of the loss function increases significantly when inertia is underestimated, prompting the model to work harder during training to accurately predict inertia requirements and avoid underestimation.

[0064] In this invention, the training process of the machine learning model can be described as the following steps: S1: Calculation Example Generation. Based on hourly forecast data of wind power, photovoltaic power, and load, the previously established unit combination optimization model with frequency safety constraints is used to solve the problem, obtaining the unit operating status and output level for each typical day and time period. The optimization results are then input into the system minimum inertia requirement calculation model to obtain the system's minimum inertia requirement at each time point.

[0065] S2: Dataset Construction. The time-by-time data for each typical day is processed to form a sample set containing input features and output target values. Input features are divided into basic features and derived features. Basic features include load forecast values. Wind power forecast Photovoltaic forecast values The output target is the minimum inertia requirement at the corresponding time. H min,t Building upon the basic features, to enhance the model's ability to characterize system operation, several derived features are introduced, one of which is the penetration rate of new energy sources. ,in For a moment t The sum of wind and solar power output reflects the impact of the proportion of new energy sources on system inertia demand; secondly, time-series information is introduced to reflect the intraday periodicity of system load and new energy output, based on a time index. t Characteristics of sine and cosine functions , ,in t The first is the intraday time period number, with a value range of 1 to 96; the second is the volatility index, used to characterize the changing trends of new energy sources and load, defined as the difference between adjacent time periods. , The output target value is the system. t Minimum inertia requirement at any given time.

[0066] S3: Data Preprocessing and Splitting. First, feature filtering is performed, removing redundant or low-information features and retaining only those highly relevant to the minimum inertia requirement. The processed dataset is then divided into training, validation, and test sets in a 6:2:2 ratio to ensure the model's generalization ability across different datasets.

[0067] S4: Model Training and Hyperparameter Tuning. During training, the improved XGBoost model is iteratively trained using the training set, while the validation set is used as the performance evaluation basis. Cross-validation is employed to adjust key hyperparameters, including the maximum tree depth, learning rate, subsampling ratio, and regularization parameters. Parameter tuning is automated using methods such as Bayesian optimization or grid search to balance model complexity and prediction accuracy.

[0068] S5: Model Evaluation and Testing. The trained model is evaluated using an independent test set, employing mean absolute error (MAE), root mean square error (RMSE), and coefficient of determination (CDO). R 2 The prediction accuracy is quantitatively evaluated using indicators such as [insert indicators here]. The adaptability and robustness of the model under different operating conditions are verified by comparing the results with actual minimum inertia requirements.

[0069] S6: Model Application. The final trained model, with input consisting only of load and renewable energy forecast data and their extended features, can quickly predict the system's minimum inertia requirement. H min The prediction result can be directly introduced into the unit combination model as a minimum inertia requirement constraint, replacing the original frequency safety constraint based on RoCoF and Nadir indices. The minimum inertia requirement constraint is: (33) By replacing equations (22)-(23) in the aforementioned unit combination model with frequency security constraints with equation (32), a unit combination model with minimum inertia constraints is obtained. This substitution significantly reduces the number of integer variables and the complexity of constraints, thereby improving the solvability and speed of the unit combination problem. While ensuring system frequency security, this method can achieve a rapid and approximately accurate characterization of the minimum inertia requirement, providing a more efficient technical path for unit combination optimization under the background of large-scale new energy access.

[0070] In summary, this invention first proposes a minimum inertia requirement calculation model, which calculates the system's minimum inertia requirement based on the unit combination optimization results with frequency safety constraints, and constructs a system minimum inertia requirement dataset for machine learning modeling. This invention also proposes a data-driven system minimum inertia requirement prediction model, employing an improved gradient boosting tree model to predict minimum inertia requirements, thereby enhancing prediction safety, and improving prediction accuracy and robustness through feature expansion. Furthermore, this invention constructs a unit combination model with minimum inertia constraints, introducing the minimum inertia requirement predicted by machine learning as a constraint condition into the unit combination model, thus replacing the traditional complex frequency constraints, simplifying the optimization problem and accelerating the solution.

[0071] As another example, the present invention provides a data-driven inertia demand assessment and unit combination device with inertia constraints, comprising: The module for building a unit combination model with frequency safety constraints is used to establish and solve a unit combination model with frequency safety constraints based on the frequency change rate constraint and the maximum frequency deviation constraint, so as to obtain the unit combination result. The dataset construction module is used to determine the minimum inertia requirement of the system for each time period based on the unit combination results and the system minimum inertia requirement calculation model, so as to construct the system inertia requirement dataset. The training module is used to train a machine learning model based on the system inertia requirement dataset to obtain a trained machine learning model. A unit combination model construction module with minimum inertia constraint is used to predict the minimum inertia requirement of the system using the trained machine learning model, and to introduce the prediction result as the minimum inertia requirement constraint into the unit combination model to replace the frequency change rate constraint and the maximum frequency deviation constraint, so as to construct a unit combination model with minimum inertia constraint. The solver module is used to solve the unit combination model with minimum inertia constraints to obtain the optimized unit combination results.

[0072] In one implementation, the unit combination model with frequency security constraints aims to minimize the total operating cost of thermal power units while maximizing the absorption of new energy sources, and its expression is:

[0073] in, C It is the total cost of operating the units in combination; T It represents the total number of time periods during which the generating units operate in combination; Ω G It is the collection of all thermal power units; a g , b g , c g They represent thermal power units g The quadratic, linear, and constant terms of the power generation cost curve; P g,t Indicates thermal power unit g At any moment t contribution; u g,t Indicates thermal power unit g At any moment t The running status, u g,t =1 indicates a thermal power unit g At any moment t It is powered on; otherwise, it is powered off.v g,t and w g,t These are the start-up and shutdown variables for thermal power units. v g,t =1 indicates a thermal power unit g At any moment t Transitioning from shutdown to startup. v g,t =1 indicates a thermal power unit g At any moment t Transitioning from shutdown to startup. w g,t =1 indicates a thermal power unit g At any moment t From power-on state to shutdown state; and thermal power units g Start-stop fees; C PFR It is the standby cost coefficient for primary frequency regulation; For the unit i At any moment t Primary frequency regulation reserve capacity; Ω PFR It is the collection of all primary frequency regulating units, including thermal power units and wind power units.

[0074] In another implementation, the constraints of the unit combination model with frequency security constraints include power balance constraints, thermal power unit output limit constraints, thermal power unit minimum start-up and shutdown time constraints, thermal power unit ramping constraints, thermal power unit operating state logic constraints, new energy unit output limit constraints, frequency change rate constraints, maximum frequency deviation constraints, and network power flow constraints.

[0075] In another implementation, the calculation model for the minimum inertia requirement of the system is as follows:

[0076]

[0077]

[0078]

[0079]

[0080] in, H min To meet the minimum inertia requirement of the system, It is the minimum inertia required under the constraint of the rate of change of frequency. It is the minimum inertia required under the maximum frequency deviation constraint.

[0081] The data-driven inertia requirement assessment and unit combination device with inertia constraints in this embodiment are used to implement the corresponding methods in the aforementioned multiple method embodiments, and have the beneficial effects of the corresponding method embodiments.

[0082] As another example, the present invention also provides an electronic device, which will now be described as an example of a hardware device that can be applied to various aspects of the present invention, serving as a server or client of the invention. The term "electronic device" is intended to represent various forms of digital electronic computer devices, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0083] The electronic device may include a processor, a communications interface, memory, and a communications bus.

[0084] The processor, communication interface, and memory communicate with each other via a communication interface. The communication interface is used to communicate with other electronic devices or servers.

[0085] The processor is used to execute programs, specifically the relevant steps in the above method embodiments.

[0086] Specifically, the program may include program code, which includes computer operation instructions.

[0087] The processor may be a CPU, an Application Specific Integrated Circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention. The one or more processors included in a smart device may be of the same type, such as one or more CPUs; or they may be of different types, such as one or more CPUs and one or more ASICs.

[0088] Memory is used to store programs. Memory may include high-speed RAM, and may also include non-volatile memory, such as at least one disk drive.

[0089] When executed by a processor, the program is used to cause an electronic device to perform the methods described in this invention.

[0090] Furthermore, the specific implementation of each step in the program can be found in the corresponding descriptions of the steps and units in the above method embodiments, and will not be repeated here. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices and modules described above can be referred to the corresponding process descriptions in the foregoing method embodiments, and will not be repeated here.

[0091] An exemplary embodiment of the present invention also provides a computer storage medium storing a computer program, wherein when the computer program is executed by a processor, it implements the methods of the various embodiments of the present invention. The corresponding process descriptions in the foregoing method embodiments can be referred to, and will not be repeated here.

[0092] The methods described above according to embodiments of the present invention can be implemented in hardware, firmware, or as software or computer code that can be stored in a recording medium (such as a CD-ROM, RAM, floppy disk, hard disk, or magneto-optical disk), or as computer code originally stored on a remote recording medium or a non-transitory machine-readable medium and subsequently stored on a local recording medium, downloaded via a network. Thus, the methods described herein can be processed by software stored on a recording medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware (such as an ASIC or FPGA). It is understood that the computer, processor, microprocessor controller, or programmable hardware includes storage components (e.g., RAM, ROM, flash memory, etc.) capable of storing or receiving software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the methods described herein. Furthermore, when a general-purpose computer accesses code used to implement the methods shown herein, the execution of the code transforms the general-purpose computer into a dedicated computer for executing the methods shown herein.

[0093] Specific embodiments of the invention have now been described. Other embodiments are within the scope of the appended claims. In some cases, the actions described in the claims can be performed in a different order and still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing can be advantageous.

[0094] It should be understood that although this specification is described according to various embodiments, not every embodiment contains only one independent technical solution. This way of describing the specification is only for clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

[0095] Finally, it should be noted that the above embodiments are only used to illustrate the embodiments of the present invention, and are not intended to limit the embodiments of the present invention. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the embodiments of the present invention. Therefore, all equivalent technical solutions also fall within the scope of the embodiments of the present invention, and the patent protection scope of the embodiments of the present invention should be defined by the claims.

Claims

1. A data-driven inertia demand assessment and unit commitment method with inertia constraints, characterized in that, The method comprises the following steps: According to the frequency change rate constraint and the maximum frequency deviation constraint, a unit commitment model containing frequency safety constraints is established and solved to obtain a unit commitment result; According to the unit commitment result, the system minimum inertia demand calculation model is used to determine the system minimum inertia demand of each period to construct a system inertia demand data set; Based on the system inertia demand data set, a machine learning model is trained to obtain a trained machine learning model; The trained machine learning model is used to predict the system minimum inertia demand, and the obtained prediction result is introduced into the unit commitment model as a minimum inertia demand constraint to replace the frequency change rate constraint and the maximum frequency deviation constraint to construct a unit commitment model containing minimum inertia constraints; The unit commitment model containing minimum inertia constraints is solved to obtain an optimized unit commitment result.

2. The method of claim 1, wherein, The unit commitment model containing frequency safety constraints takes the minimization of the total cost of thermal power unit operation and the maximization of new energy consumption as the optimization objective, and its expression is: wherein, C is the total cost of unit commitment operation; T is the total number of unit commitment operation time periods;Ω G is the set of all thermal power units; a g , b g , c g respectively represent the quadratic, linear and constant terms of the thermal power unit g generation cost curve; P g,t represent the output of the thermal power unit g at time t ; u g,t represent the operating state of the thermal power unit g at time t , u g,t =1 represents that the thermal power unit g is in the start-up state at time t , otherwise it is in the shutdown state; v g,t and w g,t are respectively the start-stop variables of the thermal power unit, v g,t =1 represents that the thermal power unit g enters the start-up state from the shutdown state at time t , v g,t =1 represents that the thermal power unit g enters the start-up state from the shutdown state at time t , w g,t =1 represents that the thermal power unit g enters the shutdown state from the start-up state at time t ; and are respectively the start-stop cost of the thermal power unit g ; C PFR is the primary frequency regulation reserve cost coefficient; is the primary frequency regulation reserve capacity of the unit i at time t ;Ω PFR is the set of all primary frequency regulation units, including thermal power units and wind power units.

3. The method of claim 2, wherein, The constraint conditions of the unit commitment model containing frequency safety constraints include power balance constraint, thermal power unit output limit constraint, thermal power unit minimum start-stop time constraint, thermal power unit ramp constraint, thermal power unit operation state logic constraint, new energy unit output limit constraint, frequency change rate constraint, maximum frequency deviation constraint and network power flow constraint.

4. The method of claim 1, wherein, The system minimum inertia demand calculation model is: wherein, H min Jmin is the minimum inertia requirement for the system, Jmin is the minimum inertia requirement for the system under the rate of frequency change constraint, Jmin is the minimum inertia requirement for the system under the maximum frequency deviation constraint.

5. A data-driven inertia demand evaluation and unit commitment device with inertia constraint, characterized in that, The method comprises the following steps: A unit commitment model containing frequency safety constraints is established and solved according to the frequency change rate constraint and the maximum frequency deviation constraint to obtain a unit commitment result; According to the unit commitment result, the system minimum inertia demand calculation model is used to determine the system minimum inertia demand of each period to construct a system inertia demand data set; Based on the system inertia demand data set, a machine learning model is trained to obtain a trained machine learning model; The trained machine learning model is used to predict the system minimum inertia demand, and the obtained prediction result is introduced into the unit commitment model as a minimum inertia demand constraint to replace the frequency change rate constraint and the maximum frequency deviation constraint to construct a unit commitment model containing minimum inertia constraints; The unit commitment model containing minimum inertia constraints is solved to obtain an optimized unit commitment result.

6. The apparatus of claim 5, wherein, The unit commitment model containing frequency safety constraints takes the minimization of the total cost of thermal power unit operation and the maximization of new energy consumption as the optimization objective, and its expression is: wherein, C is the total cost of unit commitment operation; T is the total number of unit commitment operation time periods;Ω G is the set of all thermal power units; a g , b g , c g respectively represent the quadratic, linear and constant terms of the generation cost curve of the thermal power unit g . P g,t represent the output of the thermal power unit g at time t . u g,t represent the operating state of the thermal power unit g at time t , u g,t =1 indicates that the thermal power unit g is in the start-up state at time t , otherwise it is in the shutdown state; v g,t and w g,t are respectively the start-stop variables of the thermal power unit v g,t =1 indicates that the thermal power unit g enters the start-up state from the shutdown state at time t , v g,t =1 indicates that the thermal power unit g enters the start-up state from the shutdown state at time t , w g,t =1 indicates that the thermal power unit g enters the shutdown state from the start-up state at time t . and are respectively the start-stop cost of the thermal power unit g . C PFR is the primary frequency regulation reserve cost coefficient; is the primary frequency regulation reserve capacity of the unit i at time t ;Ω PFR is the set of all primary frequency regulation units, including thermal power units and wind power units.

7. The apparatus of claim 6, wherein, The constraint conditions of the unit commitment model containing frequency safety constraints include power balance constraint, thermal power unit output limit constraint, thermal power unit minimum start-stop time constraint, thermal power unit ramp constraint, thermal power unit operation state logic constraint, new energy unit output limit constraint, frequency change rate constraint, maximum frequency deviation constraint and network power flow constraint.

8. The apparatus of claim 5, wherein, The system minimum inertia demand calculation model is: wherein, H min is the minimum inertia requirement for the system, is the minimum inertia required under the rate of frequency change constraint, is the minimum inertia required under the maximum frequency deviation constraint.

9. An electronic device, comprising: The method comprises the following steps: A processor; A memory for storing programs; The program comprises instructions which, when executed by the processor, cause the processor to perform the steps of the method according to any one of claims 1-4.

10. A computer storage medium, characterized in that, A computer program product comprising a computer program stored on a computer readable medium, the program being executable by a processor to perform the method according to any one of claims 1-4.

Images