Electric power system overall regulation capability adequacy quantitative evaluation method considering day-ahead frequency modulation demand of power grid

Abstract

The invention discloses a quantitative evaluation method for the overall regulation capability adequacy of a power system considering the day-ahead frequency modulation demand of a power grid, and the method comprises the following steps: 1), calculating the power fluctuation in a time period, and predicting the frequency modulation feature demand of the power grid; 2) constructing a quantitative evaluation model for the overall regulation capability adequacy of the power system based on the day-ahead frequency modulation demand relaxation constraint; and 3) solving the quantitative evaluation model for the overall regulation capability adequacy of the power system to obtain the regulation capability adequacy of the power system under the power grid frequency modulation characteristic demand in the step 1). According to the method, whether the overall adjustment capability of the power grid is sufficient or not is reflected through the optimal value of the day-ahead frequency modulation demand relaxation amount, the adjustment capability which can be additionally provided or still lacked by the power grid is quantitatively given, and boundary conditions are provided for inter-provincial transaction of flexible adjustment resources.

Description

Technical Field

[0001] The present invention relates to the field of power systems and automation thereof, and in particular to a method for quantitatively evaluating the adequacy of the overall regulation capability of a power system taking into account the day-ahead frequency regulation demand of a power grid. Background Art

[0002] With the continuous development of new energy, its strong random fluctuation characteristics will bring severe challenges to the safe and stable operation of the power grid. With the gradual increase in the demand for power grid regulation, affected by resource endowment, there will be significant differences in the adequacy of the overall regulation capacity of the power grid between different regions: some provincial power grids have high installed capacity of new energy and relatively insufficient regulation resources, resulting in insufficient adequacy of the overall regulation capacity of the power grid, which restricts the development of new energy; some provinces have small installed capacity of new energy and relatively abundant regulation resources, resulting in insufficient utilization of regulation resources. Under this background, it is necessary to carry out inter-provincial mutual assistance transactions of flexible regulation resources, and promote the consumption and utilization of new energy in a larger range through the wide-area optimization allocation of flexible regulation resources. The prerequisite is to clarify the adequacy of the overall regulation capacity of different provincial power grids, and then provide boundary conditions for inter-provincial transactions of flexible regulation resources.

[0003] Domestic and foreign scholars have conducted relevant research on the assessment of the adequacy of the regulation capacity of the power system. Under the condition of given regulation demand, the adequacy of the overall regulation capacity of the power grid is assessed through indicators such as the probability of insufficient regulation capacity and the expectation of insufficient regulation capacity. However, the existing research mainly reflects the adequacy of the regulation capacity of the power grid under different power supply structures, thereby providing a reference basis for the planning of regulating power sources. In terms of the day-ahead market, it does not directly reflect the additional or insufficient regulation capacity that the provincial power grid can provide in different periods, making it difficult to provide boundary conditions for inter-provincial transactions. In addition, the reasonable quantification of the grid regulation demand is a prerequisite for assessing the adequacy of the overall regulation capacity of the power grid.

[0004] Existing studies mainly consider the demand for frequency regulation capacity, and characterize the system frequency regulation capacity demand based on operating experience, prediction error distribution, data-driven and other methods. However, existing studies mainly consider the deviation between the predicted value and the actual value of renewable energy at the dispatching time point, and do not consider the impact of renewable energy power fluctuations during the dispatching period on the system frequency regulation capacity demand, making it difficult to effectively characterize the frequency regulation capacity demand. In addition, the frequency regulation capacity demand mainly reflects the regulation margin that the power grid needs to reserve to cope with the random fluctuations of renewable energy. However, even if the regulation margin is sufficient, if the overall regulation speed of the power grid is difficult to cope with the large power fluctuations of renewable energy in a short period of time, it will also cause a large grid frequency deviation, threatening the safe operation of the power grid. Therefore, the power grid needs to have not only sufficient regulation margin, but also sufficient regulation speed. The existing regulation capacity sufficiency assessment method mainly considers the ramping ability of regulation resources during the entire dispatching period, and does not consider the demand for resource regulation speed due to power fluctuations during the renewable energy period. Summary of the invention

[0005] The object of the present invention is to provide a method for quantitatively evaluating the margin of the overall regulation capacity of a power system taking into account the day-ahead frequency regulation demand of the power grid, comprising the following steps:

[0006] 1) Taking into account the power fluctuations within the time period, the demand for grid frequency regulation characteristics is predicted;

[0007] 2) Construct a quantitative assessment model for the overall regulation capacity adequacy of the power system based on the relaxation constraint of the day-ahead frequency regulation demand;

[0008] 3) Solve the quantitative evaluation model of the overall regulation capacity margin of the power system to obtain the regulation capacity margin of the power system under the grid frequency regulation characteristic requirements in step 1).

[0009] Furthermore, the steps of predicting the power grid predicted daily frequency regulation demand data include:

[0010] s1) Construct the input characteristics of power grid frequency regulation demand, including new energy fluctuations, load fluctuations and meteorological factors;

[0011] s2) building a DNN deterministic prediction model;

[0012] s3) Inputting the input features of the power grid frequency regulation demand into the DNN deterministic prediction model to obtain the power grid forecast day frequency regulation demand data, including the frequency regulation capacity demand and frequency regulation speed demand for each dispatching period on the forecast day;

[0013] s4) Use the ABKDE method to perform interval prediction on the power grid's predicted daily frequency regulation demand data, obtain the confidence interval of the power grid's predicted daily frequency regulation demand data, and use the upper limit value of the data in the confidence interval as the power grid's frequency regulation characteristic demand to achieve quantification of the frequency regulation capacity and frequency regulation speed requirements.

[0014] Furthermore, the input characteristics of the power grid frequency regulation demand include the fluctuation characteristics of new energy, load and net load in the previous day, as well as the meteorological factors on the forecast day;

[0015] The load fluctuation characteristics include the maximum value, minimum value, mean value and / or median value of the load power in each dispatching period one day before the forecast date;

[0016] The new energy fluctuation characteristics include the maximum value, minimum value, mean value and / or median value of the new energy power in each dispatching period one day before the forecast date, and the maximum value, minimum value, mean value and / or median value of the frequency regulation capacity demand in each dispatching period one day before the forecast date;

[0017] The net load fluctuation characteristics include the maximum value, minimum value, mean value and / or median value of the net load power in each dispatching period one day before the forecast date, and the maximum value, minimum value, mean value and / or median value of the short-term power fluctuation of the net load in each dispatching period one day before the forecast date;

[0018] The meteorological factors on the forecast day include the holiday type on the forecast day, the highest, lowest and / or average temperature in each dispatch period on the forecast day, the highest, lowest and / or average humidity in each dispatch period on the forecast day, and the highest, lowest and / or average rainfall in each dispatch period on the forecast day;

[0019] The holiday type of the prediction day is a 0-1 variable, 1 means the prediction day is a holiday, 0 means it is not a holiday.

[0020] Furthermore, the DNN deterministic prediction model is trained using the power grid day-ahead frequency regulation demand dataset;

[0021] The power grid's day-ahead frequency regulation demand data set includes multiple sets of input and output data; the input data is the input characteristics of the power grid's frequency regulation demand, including renewable energy fluctuations, load fluctuations and meteorological factors; the output data is the power grid's predicted day-ahead frequency regulation demand data, including the frequency regulation capacity demand and frequency regulation speed demand for each dispatching period on the predicted day.

[0022] Furthermore, the steps of using the ABKDE method to perform interval forecasting on the power grid predicted daily frequency regulation demand data include:

[0023] b1) Construct the ABKDE model, namely:

[0024]

[0025] Where h0 is the global baseline bandwidth, λ is the adjustment parameter, g(x) is the density estimate at x, f(x) is the probability density function; x is the daily frequency demand data of the power grid output by the DNN deterministic prediction model; K is the kernel function, n is the number of samples, and x is the probability density function. i is the sample point predicted by DNN; h i is bandwidth;

[0026] b2) Construct the cumulative distribution function F(x), that is:

[0027]

[0028] In the formula, the cumulative distribution function F(x) represents the probability that the random variable X is less than or equal to a specific value x;

[0029] b3) Calculate the predicted value x' with a confidence interval of 95%, using the confidence interval of the power grid's predicted daily frequency regulation demand data;

[0030] The predicted value x' is as follows:

[0031] x'=F -1 (p) (4)

[0032] Where p is a value between 0 and 1.

[0033] Furthermore, the objective function of the quantitative evaluation model of the power system overall regulation capacity adequacy based on the day-ahead frequency regulation demand relaxation constraint is as follows:

[0034]

[0035] In the formula, g and t are the unit number and the dispatch period number respectively; G and T are the number of units and the dispatch time set respectively; Indicates the unit output; Quote energy prices for units; Quote for the up and down frequency regulation capacity of the unit; The up and down frequency regulation capacity of the unit won the bid; 1 / T e is the proportion of the duration of a single scheduling period in the day-ahead scheduling decision model to 1 hour; M and N are the penalty coefficients for the shortage of frequency regulation capacity demand and frequency regulation speed demand; Provides the system with frequency modulation capacity up and down; The relaxation amount required for the up and down frequency modulation speeds.

[0036] Furthermore, the power system regulation capability is improved through the system up and down frequency regulation capacity. And the relaxation of the upper and lower frequency modulation speed requirements Quantification.

[0037] Furthermore, the constraints of the quantitative evaluation model for the adequacy of the overall regulation capacity of the power system based on the relaxation constraint of the day-ahead frequency regulation demand include power balance constraints, frequency regulation capacity constraints, frequency regulation speed limits, unit output limits, and line flow constraints.

[0038] Furthermore, the power balance constraint is as follows:

[0039]

[0040] In the formula, b is the node number; B is the node set; is the net load during the dispatch period.

[0041] Indicates the unit output;

[0042] The frequency regulation capacity constraints are as follows:

[0043]

[0044] In the formula, The system's up and down frequency modulation capacity requirements; The upper and lower frequency regulation capacity of the unit's winning bid; R is the up and down frequency modulation capacity of the system; g,max,UP , R g,max,DN It is the upper limit of the system's up and down frequency modulation capacity;

[0045] The FM speed limits are as follows:

[0046]

[0047] In the formula, It is the climbing capability of the unit during the entire dispatch period; The predicted up and down frequency regulation capacity of adjacent net load; The relaxation amount required for the upper and lower frequency modulation speeds; is the coefficient;

[0048] The unit output limits are as follows:

[0049]

[0050] Where P g,min With P g,max are the maximum and minimum values ​​of the unit output. Formulas (13) and (14) indicate that the unit output should be within a certain range during the day-ahead dispatch decision stage, and the frequency-regulating unit needs to reserve a frequency-regulating capacity that is not less than the bid-winning capacity during the dispatch period.

[0051] The line flow constraints are as follows:

[0052]

[0053] Where: l represents the line number, B represents the node set; H lb A represents the transfer distribution factor matrix element between the injected power and the active power flow; b The matrix element representing the relationship between unit g and node b; P l max Indicates the maximum line transmission capacity limit.

[0054] The technical effect of the present invention is unquestionable. The present invention proposes a quantitative assessment method for the adequacy of the overall regulation capacity of the power system taking into account the day-ahead frequency regulation demand of the power grid. Under the condition of considering the power fluctuation during the new energy dispatch period, the method uses a data-driven model to comprehensively quantify the day-ahead frequency regulation demand including the power grid frequency regulation capacity demand and the frequency regulation speed demand. Based on this, a quantitative assessment model for the adequacy of the overall regulation capacity of the power system based on the relaxation constraint of the day-ahead frequency regulation demand is constructed. The optimal value of the day-ahead frequency regulation demand relaxation reflects whether the overall regulation capacity of the power grid is sufficient, and quantitatively gives the additional regulation capacity that the power grid can provide or the regulation capacity that is still lacking, providing boundary conditions for inter-provincial transactions of flexible regulation resources.

[0055] The beneficial effects of the present invention are as follows:

[0056] 1) A data-driven quantification method for the day-ahead frequency regulation demand of power grids taking into account power fluctuations within a time period is proposed. Based on the fine-grained historical power curves of new energy and loads, feature vectors for frequency regulation capacity demand prediction and frequency regulation speed demand prediction are constructed respectively; a point prediction model for day-ahead frequency regulation demand of power grids based on deep neural network (DNN) is constructed. On this basis, in order to ensure that the quantification of power grid frequency regulation demand is sufficient, an interval prediction model for day-ahead frequency regulation demand of power grids is further constructed based on adaptive bandwidth kernel density estimation (ABKDE), and the upper limit of the prediction interval is used as the final quantification result of frequency regulation demand.

[0057] A quantitative evaluation model for the adequacy of the overall regulation capacity of the power system based on the relaxation constraint of the day-ahead frequency regulation demand 2) A quantitative evaluation model for the adequacy of the overall regulation capacity of the power system based on the relaxation constraint of the day-ahead frequency regulation demand is proposed. Based on the quantified day-ahead frequency regulation capacity demand and frequency regulation speed demand of the power grid, the relaxation constraint of the day-ahead frequency regulation demand is embedded in the day-ahead dispatch decision model of the power grid; the optimal value of the relaxation amount can quantitatively characterize the adequacy of the overall regulation capacity of the power grid (that is, quantitatively give the regulation capacity that the power grid can provide additionally or the regulation capacity that is still lacking), and provide boundary conditions for inter-provincial transactions of flexible regulation resources.

[0058] The method proposed in the present invention can quantitatively evaluate the overall flexible regulation capability of the power grid, provide boundary conditions for inter-provincial transactions of flexible regulation resources, and promote wide-area optimal configuration of flexible regulation resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 Drive the overall process for frequency modulation feature demand data;

[0060] Figure 2 It is the DNN model structure. DETAILED DESCRIPTION

[0061] The present invention is further described below in conjunction with the embodiments, but it should not be understood that the above subject matter of the present invention is limited to the following embodiments. Without departing from the above technical ideas of the present invention, various substitutions and changes are made according to the common technical knowledge and customary means in the art, which should all be included in the protection scope of the present invention.

[0062] Embodiment 1:

[0063] See also Figure 1 to Figure 2 A quantitative evaluation method for the overall regulation capability margin of a power system taking into account the day-ahead frequency regulation demand of the power grid comprises the following steps:

[0064] 1) Taking into account the power fluctuations within the time period, the demand for grid frequency regulation characteristics is predicted;

[0065] 2) Construct a quantitative assessment model for the overall regulation capacity adequacy of the power system based on the relaxation constraint of the day-ahead frequency regulation demand;

[0066] 3) Use MATLAB to solve the quantitative evaluation model of the overall regulation capacity margin of the power system, and obtain the regulation capacity margin of the power system under the grid frequency regulation characteristic requirements in step 1).

[0067] The steps for predicting the power grid forecast daily frequency regulation demand data include:

[0068] s1) Construct the input characteristics of power grid frequency regulation demand, including new energy fluctuations, load fluctuations and meteorological factors;

[0069] s2) building a DNN deterministic prediction model;

[0070] s3) Inputting the input features of the power grid frequency regulation demand into the DNN deterministic prediction model to obtain the power grid forecast day frequency regulation demand data, including the frequency regulation capacity demand and frequency regulation speed demand for each dispatching period on the forecast day;

[0071] s4) Use the ABKDE method to perform interval prediction on the power grid's predicted daily frequency regulation demand data, obtain the confidence interval of the power grid's predicted daily frequency regulation demand data, and use the upper limit value of the data in the confidence interval as the power grid's frequency regulation characteristic demand to achieve quantification of the frequency regulation capacity and frequency regulation speed requirements.

[0072] The input characteristics of the power grid frequency regulation demand include the new energy, load and net load fluctuation characteristics of the previous day, as well as the meteorological factors of the forecast day;

[0073] The load fluctuation characteristics include the maximum value, minimum value, mean value and / or median value of the load power in each dispatching period one day before the forecast date;

[0074] The new energy fluctuation characteristics include the maximum value, minimum value, mean value and / or median value of the new energy power in each dispatching period one day before the forecast date, and the maximum value, minimum value, mean value and / or median value of the frequency regulation capacity demand in each dispatching period one day before the forecast date;

[0075] The net load fluctuation characteristics include the maximum value, minimum value, mean value and / or median value of the net load power in each dispatching period one day before the forecast date, and the maximum value, minimum value, mean value and / or median value of the short-term power fluctuation of the net load in each dispatching period one day before the forecast date;

[0076] The meteorological factors on the forecast day include the holiday type on the forecast day, the highest, lowest and / or average temperature in each dispatch period on the forecast day, the highest, lowest and / or average humidity in each dispatch period on the forecast day, and the highest, lowest and / or average rainfall in each dispatch period on the forecast day;

[0077] The holiday type of the prediction day is a 0-1 variable, 1 means the prediction day is a holiday, 0 means it is not a holiday.

[0078] The DNN deterministic prediction model is trained using the power grid day-ahead frequency regulation demand dataset;

[0079] The power grid's day-ahead frequency regulation demand data set includes multiple sets of input and output data; the input data is the input characteristics of the power grid's frequency regulation demand, including renewable energy fluctuations, load fluctuations and meteorological factors; the output data is the power grid's predicted day-ahead frequency regulation demand data, including the frequency regulation capacity demand and frequency regulation speed demand for each dispatching period on the predicted day.

[0080] The steps of using the ABKDE method to make interval predictions on the power grid predicted daily frequency regulation demand data include:

[0081] b1) Construct the ABKDE model, namely:

[0082]

[0083] Where h0 is the global baseline bandwidth, λ is the adjustment parameter, g(x) is the density estimate at x, f(x) is the probability density function; x is the daily frequency demand data of the power grid output by the DNN deterministic prediction model; K is the kernel function, n is the number of samples, and x is the probability density function. i is the sample point predicted by DNN; h i is bandwidth;

[0084] b2) Construct the cumulative distribution function F(x), that is:

[0085]

[0086] In the formula, the cumulative distribution function F(x) represents the probability that the random variable X is less than or equal to a specific value x;

[0087] b3) Calculate the predicted value x' with a confidence interval of 95%, using the confidence interval of the power grid's predicted daily frequency regulation demand data;

[0088] The predicted value x' is as follows:

[0089] x'=F -1 (p) (4)

[0090] Where p is a value between 0 and 1.

[0091] The objective function of the quantitative evaluation model of the power system overall regulation capacity margin based on the relaxation constraint of the day-ahead frequency regulation demand is as follows:

[0092]

[0093] In the formula, g and t are the unit number and the dispatch period number respectively; G and T are the number of units and the dispatch time set respectively; Indicates the unit output; Quote energy prices for units; Quote for the up and down frequency regulation capacity of the unit; The up and down frequency regulation capacity of the unit won the bid; 1 / T e is the proportion of the duration of a single scheduling period in the day-ahead scheduling decision model to 1 hour; M and N are the penalty coefficients for the shortage of frequency regulation capacity demand and frequency regulation speed demand; Provides the system with frequency modulation capacity up and down; The relaxation amount required for the up and down frequency modulation speeds.

[0094] The power system regulation capability is achieved through the system up and down frequency regulation capacity And the relaxation of the upper and lower frequency modulation speed requirements Quantification.

[0095] The constraints of the quantitative evaluation model for the adequacy of the overall regulation capacity of the power system based on the relaxation constraint of the day-ahead frequency regulation demand include power balance constraint, frequency regulation capacity constraint, frequency regulation speed limit, unit output limit and line flow constraint.

[0096] The power balance constraints are as follows:

[0097]

[0098] In the formula, b is the node number; B is the node set; is the net load during the dispatch period.

[0099] Indicates the unit output;

[0100] The frequency regulation capacity constraints are as follows:

[0101]

[0102] In the formula, The system's up and down frequency modulation capacity requirements; The upper and lower frequency regulation capacity of the unit's winning bid; R is the up and down frequency modulation capacity of the system; g,max,UP , R g,max,DN It is the upper limit of the system's up and down frequency modulation capacity;

[0103] The FM speed limits are as follows:

[0104]

[0105] In the formula, It is the climbing capability of the unit during the entire dispatch period; The predicted up and down frequency regulation capacity of adjacent net load; The relaxation amount required for the upper and lower frequency modulation speeds; is the coefficient;

[0106] The unit output limits are as follows:

[0107]

[0108] Where P g,min With P g,max are the maximum and minimum values ​​of the unit output. Formulas (13) and (14) indicate that the unit output should be within a certain range during the day-ahead dispatch decision stage, and the frequency-regulating unit needs to reserve a frequency-regulating capacity that is not less than the bid-winning capacity during the dispatch period.

[0109] The line flow constraints are as follows:

[0110]

[0111] Where: l represents the line number, B represents the node set; H lb A represents the transfer distribution factor matrix element between the injected power and the active power flow; b The matrix element representing the relationship between unit g and node b; P l max Indicates the maximum line transmission capacity limit.

[0112] Embodiment 2:

[0113] A quantitative evaluation method for the margin of overall regulation capability of a power system taking into account the day-ahead frequency regulation demand of a power grid comprises the following steps:

[0114] 1) Taking into account the power fluctuations within the time period, the demand for grid frequency regulation characteristics is predicted;

[0115] 2) Construct a quantitative assessment model for the overall regulation capacity adequacy of the power system based on the relaxation constraint of the day-ahead frequency regulation demand;

[0116] 3) Solve the quantitative evaluation model of the overall regulation capacity margin of the power system to obtain the regulation capacity margin of the power system under the grid frequency regulation characteristic requirements in step 1).

[0117] Embodiment 3:

[0118] A quantitative evaluation method for the margin of overall regulation capability of a power system taking into account the day-ahead frequency regulation demand of a power grid, wherein the calculation content is the same as that of Example 2, and further, the step of predicting the predicted day-ahead frequency regulation demand data of the power grid comprises:

[0119] s1) Construct the input characteristics of power grid frequency regulation demand, including new energy fluctuations, load fluctuations and meteorological factors;

[0120] s2) building a DNN deterministic prediction model;

[0121] s3) Inputting the input features of the power grid frequency regulation demand into the DNN deterministic prediction model to obtain the power grid forecast day frequency regulation demand data, including the frequency regulation capacity demand and frequency regulation speed demand for each dispatching period on the forecast day;

[0122] s4) Use the ABKDE method to perform interval prediction on the power grid's predicted daily frequency regulation demand data, obtain the confidence interval of the power grid's predicted daily frequency regulation demand data, and use the upper limit value of the data in the confidence interval as the power grid's frequency regulation characteristic demand to achieve quantification of the frequency regulation capacity and frequency regulation speed requirements.

[0123] Embodiment 4:

[0124] A quantitative evaluation method for the adequacy of the overall regulation capacity of a power system taking into account the day-ahead frequency regulation demand of the power grid, wherein the calculation content is the same as any one of Embodiments 2-3, and further, the input characteristics of the power grid frequency regulation demand include the fluctuation characteristics of new energy, load and net load of the previous day, and the meteorological factors of the forecast day;

[0125] The load fluctuation characteristics include the maximum value, minimum value, mean value and / or median value of the load power in each dispatching period one day before the forecast date;

[0126] The new energy fluctuation characteristics include the maximum value, minimum value, mean value and / or median value of the new energy power in each dispatching period one day before the forecast date, and the maximum value, minimum value, mean value and / or median value of the frequency regulation capacity demand in each dispatching period one day before the forecast date;

[0127] The net load fluctuation characteristics include the maximum value, minimum value, mean value and / or median value of the net load power in each dispatching period one day before the forecast date, and the maximum value, minimum value, mean value and / or median value of the short-term power fluctuation of the net load in each dispatching period one day before the forecast date;

[0128] The meteorological factors on the forecast day include the holiday type on the forecast day, the highest, lowest and / or average temperature in each dispatch period on the forecast day, the highest, lowest and / or average humidity in each dispatch period on the forecast day, and the highest, lowest and / or average rainfall in each dispatch period on the forecast day;

[0129] The holiday type of the prediction day is a 0-1 variable, 1 means the prediction day is a holiday, 0 means it is not a holiday.

[0130] Embodiment 5:

[0131] A quantitative evaluation method for the margin of overall regulation capability of a power system taking into account the day-ahead frequency regulation demand of a power grid, wherein the calculation content is the same as any one of Embodiments 2-4, and further, a DNN deterministic prediction model is obtained by training a data set of day-ahead frequency regulation demand of a power grid;

[0132] The power grid's day-ahead frequency regulation demand data set includes multiple sets of input and output data; the input data is the input characteristics of the power grid's frequency regulation demand, including renewable energy fluctuations, load fluctuations and meteorological factors; the output data is the power grid's predicted day-ahead frequency regulation demand data, including the frequency regulation capacity demand and frequency regulation speed demand for each dispatching period on the predicted day.

[0133] Embodiment 6:

[0134] A quantitative evaluation method for the margin of overall regulation capability of a power system taking into account the day-ahead frequency regulation demand of a power grid, wherein the calculation content is the same as any one of Embodiments 2-5, and further, the step of performing interval prediction on the predicted day-ahead frequency regulation demand data of the power grid using the ABKDE method comprises:

[0135] b1) Construct the ABKDE model, namely:

[0136]

[0137] Where h0 is the global baseline bandwidth, λ is the adjustment parameter, g(x) is the density estimate at x, and f(x) is the probability density function;

[0138] b2) Construct the cumulative distribution function F(x), that is:

[0139]

[0140] In the formula, the cumulative distribution function F(x) represents the probability that the random variable X is less than or equal to a specific value x;

[0141] b3) Calculate the predicted value x' with a confidence interval of 95%, using the confidence interval of the power grid's predicted daily frequency regulation demand data;

[0142] The predicted value x' is as follows:

[0143] x'=F -1 (p) (4)

[0144] Where p is a value between 0 and 1.

[0145] Embodiment 7:

[0146] A quantitative evaluation method for the margin of overall regulation capability of a power system taking into account the day-ahead frequency regulation demand of a power grid, wherein the calculation content is the same as any one of Embodiments 2-6, and further, the objective function of the quantitative evaluation model for the margin of overall regulation capability of a power system based on the relaxation constraint of the day-ahead frequency regulation demand is as follows:

[0147]

[0148] In the formula, g and t are the unit number and the dispatch period number respectively; G and T are the number of units and the dispatch time set respectively; Indicates the unit output; Quote energy prices for units; Quote for up / down frequency regulation capacity of the unit; The up / down frequency regulation capacity of the unit won the bid; 1 / T e is the proportion of the duration of a single scheduling period in the day-ahead scheduling decision model to 1 hour; M and N are the penalty coefficients for the shortage of frequency regulation capacity demand and frequency regulation speed demand; ε R,UP / DN , ε H,UP / DN It is the system's up / down frequency regulation capacity and up / down frequency regulation speed demand relaxation.

[0149] Embodiment 8:

[0150] A method for quantitatively evaluating the margin of the overall regulation capability of an electric power system taking into account the day-ahead frequency regulation demand of the power grid, wherein the calculation content is the same as any one of Examples 2-7. Furthermore, the constraints of the quantitative evaluation model for the margin of the overall regulation capability of the electric power system based on the relaxation constraint of the day-ahead frequency regulation demand include power balance constraints, frequency regulation capacity constraints, frequency regulation speed limits, unit output limits, and line flow constraints.

[0151] Embodiment 9:

[0152] A quantitative evaluation method for the margin of overall regulation capability of a power system taking into account the day-ahead frequency regulation demand of the power grid, wherein the calculation content is the same as any one of Embodiments 2-8, and further, the power balance constraint is as follows:

[0153]

[0154] In the formula, b is the node number; B is the node set; is the initial net load during the dispatch period.

[0155] The frequency regulation capacity constraints are as follows:

[0156]

[0157] In the formula, Adjust the system's up / down frequency capacity requirements.

[0158] The FM speed limits are as follows:

[0159]

[0160] In the formula, It is the climbing capability of the unit during the entire dispatch period; The predicted up and down frequency regulation capacity of adjacent net load; The relaxation amount required for the upper and lower frequency modulation speeds; is the coefficient;

[0161] The unit output limits are as follows:

[0162]

[0163] Where P g,min With P g,max are the maximum and minimum values ​​of the unit output. Formulas (13) and (14) indicate that the unit output should be within a certain range during the day-ahead dispatch decision stage, and the frequency-regulating unit needs to reserve a frequency-regulating capacity that is not less than the bid-winning capacity during the dispatch period.

[0164] The line flow constraints are as follows:

[0165]

[0166] Where: l represents the line number, B represents the node set; H lb A represents the transfer distribution factor matrix element between the injected power and the active power flow; b The matrix element representing the relationship between unit g and node b; P l max Indicates the maximum line transmission capacity limit.

[0167] Embodiment 10:

[0168] A quantitative evaluation method for the overall regulation capacity adequacy of a power system taking into account the day-ahead frequency regulation demand of the power grid, wherein the calculation content is the same as any one of Embodiments 2-9, and further, the regulation capacity of the power system is calculated by the system up / down frequency regulation capacity ε R,UP / DN , Up / down frequency modulation speed demand relaxation εH,UP / DN Quantification.

[0169] Embodiment 11:

[0170] A quantitative evaluation method for the overall regulation capacity adequacy of the power system taking into account the day-ahead frequency regulation demand of the power grid is as follows:

[0171] Part 1: A quantification technology for system frequency regulation characteristic demand is proposed (including data-driven point prediction model and interval prediction model). Part 2: The frequency regulation characteristic demand is embedded as a boundary condition in the overall regulation capacity adequacy assessment model of the power system, and the system flexibility regulation capability is characterized by constructing a dispatch decision model that relaxes the frequency regulation characteristic demand constraints.

[0172] The specific steps are as follows:

[0173] (1) Data-driven quantification method for power grid day-ahead frequency regulation demand considering intra-period power fluctuations

[0174] The characteristics of the system frequency regulation demand during the dispatch period can be decomposed into the frequency regulation capacity and frequency regulation speed requirements at the time point. Among them, the frequency regulation capacity requirement characterizes the regulation margin that the system needs to reserve to cope with random fluctuations in net load, which is reflected as the maximum gap between the power after system regulation and the dispatch plan; the frequency regulation speed requirement characterizes the climbing ability required by the system to cope with short-term power fluctuations in net load, which is reflected as the maximum short-term power fluctuation of net load during the dispatch period. Through the reasonable quantification of the two types of frequency regulation requirements, it can be ensured that the system has the regulation margin and regulation speed to cope with random fluctuations in net load. To this end, the present invention proposes a data-driven quantification method for the day-ahead frequency regulation demand of the power grid taking into account power fluctuations within the time period, which is specifically described as follows:

[0175] ① Overall process of data-driven quantification method for power grid day-ahead frequency regulation demand considering power fluctuations within a time period

[0176] 1) Feature vector construction: Based on data such as renewable energy fluctuations, load fluctuations and meteorological factors, the input features of the power grid frequency regulation demand are constructed, with a total of 3170 dimensions; the output features are the power grid frequency regulation demand characteristics of the forecast day, with a total of 96 dimensions (i.e., the power grid frequency regulation demand characteristics of each time period are predicted separately).

[0177] 2) DNN deterministic prediction model: The present invention utilizes the powerful nonlinear feature extraction capability of DNN, inputs the sample data and input features after the second and third steps into the DNN deterministic prediction model, and obtains the point prediction value of the system's power grid frequency regulation demand characteristics on the prediction day.

[0178] 3) ABKDE interval prediction model: Based on the deterministic prediction results of DNN, the present invention adopts the ABKDE method to perform interval prediction and obtain the confidence interval of the prediction results of the power grid frequency regulation demand characteristics.

[0179] 4) Quantification of system frequency regulation demand: The upper limit value of the interval prediction obtained by the ABKDE method is finally used as the system's power grid frequency regulation characteristic demand, and then the frequency regulation capacity and frequency regulation speed requirements are quantified. Figure 1 The following is the overall flow chart.

[0180] ② Construction of the day-ahead frequency regulation demand characteristics of the power grid based on historical data and meteorological factors

[0181] This section aims to predict the frequency regulation capacity or frequency regulation speed of the system in the next day. The input features with a granularity of 15 minutes are constructed based on the data with a fine granularity of 1 minute. The constructed input and output features are shown in Table 1. The input feature vector is the fluctuation characteristics of new energy, load and net load in the previous day of history, as well as the meteorological factors on the forecast day, with a total of 3170 dimensions; the output feature vector is the frequency regulation speed / frequency regulation capacity / frequency regulation speed of the system on the forecast day, with a total of 96 dimensions.

[0182] Table 1 Input and output feature vectors of load forecasting model

[0183]

[0184]

[0185] ③ DNN-based power grid day-ahead frequency regulation demand point prediction model

[0186] The training samples formed above can be used for training the frequency modulation speed and frequency modulation capacity point prediction model.

[0187] Deep neural networks (DNNs) have excellent predictive capabilities due to their superiority in extracting nonlinear features, processing large-scale data, and dealing with noisy data (see Figure 2 ). Therefore, the present invention chooses to use DNN to construct a prediction model for the maximum fluctuation point of net load short-term power. However, the selection of a specific data-driven model is not the core issue of the present invention. Other models such as extreme learning machines, decision trees, long short-term memory networks, and convolutional neural networks can also be used to predict the maximum fluctuation value of net load short-term power. After training, DNN can be used to predict the maximum fluctuation of the system net load in each scheduling period of the operating day.

[0188] ④ ABKDE-based power grid day-ahead frequency regulation demand interval prediction model

[0189] For the point prediction of frequency modulation capacity and frequency modulation speed, the main consideration is: due to the existence of errors, if the predicted frequency modulation capacity and frequency modulation speed are lower than the actual demand values, it may lead to insufficient reserved capacity for system auxiliary services, affecting the stability of frequency. In order to reduce this risk, the present invention further adopts an integrated method based on adaptive bandwidth kernel density estimation (ABKDE) to compensate for errors. When the upper limit of the interval prediction is used as the demand for frequency modulation capacity and frequency modulation speed (that is, the maximum possible frequency modulation capacity and speed), the possibility of the predicted value being less than the true value can be significantly reduced, thereby improving the frequency performance of the system. Therefore, the prediction results of DNN are input into ABKDE for further processing, and the specific process is as follows.

[0190] Kernel density estimation (KDE) is a nonparametric probability density estimation method that mainly smoothes sample points to make them affect the surrounding space, and uses kernel functions to quantify this effect. Its mathematical expression is shown in formula (1).

[0191]

[0192] In the formula, K is the kernel function, n is the number of samples, and x i is the sample point predicted by DNN, and h is the bandwidth.

[0193] Although kernel density estimation (KDE) can estimate the probability density of samples to a certain extent, when the data distribution is uneven, the fixed bandwidth h may lead to problems such as over-smoothing or under-fitting in the estimation process. To solve these problems, adaptive bandwidth kernel density estimation (ABKDE) introduces an adaptive bandwidth method, which allows the bandwidth to be adjusted according to the local data density. In areas with sparse data, the bandwidth is increased to reduce the variance of the estimate; while in areas with dense data, the bandwidth is reduced to reduce the estimation bias. The ABKDE model is shown in equations (2) and (3).

[0194]

[0195] Where h0 is the global baseline bandwidth, λ is the adjustment parameter, g(x) is the density estimate at x, and f(x) is the probability density function.

[0196] On this basis, in order to ensure that the predicted value is slightly larger than the actual value of the actual operation scheduling point, the present invention introduces the Cumulative Distribution Function (CDF) and the Percent Point Function (PPF).

[0197] The cumulative distribution function F(x) represents the probability that the random variable X is less than or equal to a specific value x. The specific formula of the cumulative distribution function F(x) is shown in (4).

[0198]

[0199] CDF is a non-decreasing function, and its value is between 0 and 1. Then, the PPF function is obtained according to the inverse function of the cumulative distribution function F(x). The PPF function is shown in formula (5).

[0200] x=F -1 (p) (5)

[0201] In the formula, p is usually a value between 0 and 1. Taking the confidence interval of 95% as an example, the value of p is 0.975; x represents the predicted value with a confidence interval of 95%.

[0202] ⑤ Quantification of power grid frequency regulation demand on the day before

[0203] After processing based on the DNN point prediction model and the ABKDE interval prediction model, the present invention selects the upper limit value of the prediction result of the ABKDE interval prediction model as the system frequency regulation demand, ensuring that the obtained frequency regulation demand can include the actual operation status of the system, and embeds the system frequency regulation demand into the subsequent quantitative evaluation model of the adequacy of the overall regulation capacity of the power system based on the relaxation constraint of the day-ahead frequency regulation demand.

[0204] (2) Quantitative evaluation model of the overall regulation capacity of the power system based on the relaxation constraint of the day-ahead frequency regulation demand

[0205] The overall regulation capacity margin of the power system can be characterized as the system's ability to respond to the power grid's day-ahead frequency regulation demand. To this end, this section builds a quantitative assessment model for the overall regulation capacity margin of the power system based on the day-ahead frequency regulation demand relaxation constraint based on the power grid's day-ahead frequency regulation capacity and frequency regulation speed requirements obtained in the previous section, and characterizes the overall regulation capacity margin of the system by evaluating the degree to which the system meets the day-ahead frequency regulation demand.

[0206] Objective Function

[0207] In order to improve the rationality and accuracy of the day-ahead dispatch decision model and accurately characterize the regulation capability of the power system, this section introduces the cost or benefit of the power system participating in the transaction while considering the energy cost of the unit and the frequency regulation capacity cost. The objective function is still to minimize the system operation cost, as shown in formula (6).

[0208]

[0209] In the formula, g and t are the unit number and the dispatch period number respectively; G and T are the number of units and the dispatch time set respectively; Indicates the unit output; is the unit energy quote ($ / MWh), Quote the up / down frequency regulation capacity of the unit ($ / MWh); The up / down frequency regulation capacity of the unit won the bid; 1 / T e is the proportion of a single scheduling period duration in the day-ahead scheduling decision model to 1 hour. The scheduling period of the present invention is 15 minutes, then T e =4. Where M / N is the penalty coefficient of the frequency regulation capacity demand / frequency regulation speed demand shortfall, ε R,UP / DN , ε H,UP / DN It is the relaxation of the system's up / down frequency regulation capacity and up / down frequency regulation speed requirements, and represents the shortage of the system's up / down frequency regulation capacity requirements and up / down frequency regulation speed requirements.

[0210] 2) Operational constraints

[0211] The operating constraints of the traditional day-ahead scheduling decision model are introduced in detail as follows.

[0212] i. Power balance constraints

[0213] At the initial moment of each dispatch period, the unit output must be equal to the load, as shown in formula (7).

[0214]

[0215] In the formula, b is the node number; B is the node set; is the initial net load during the dispatch period.

[0216] ii. Frequency regulation capacity constraints

[0217] In the traditional day-ahead dispatch decision model, the frequency regulation capacity provided by the unit is limited by the maximum frequency regulation capacity reported by it, and the winning frequency regulation capacity should not be less than 0. At the same time, in order to accurately characterize the flexibility regulation capability of the power system, the present invention embeds slack in the frequency regulation capacity constraint to characterize the regulation capability of the power system that can participate in the transaction. The unit frequency regulation capacity constraint is shown in equations (8)-(11).

[0218]

[0219] In the formula, Adjust the system's up / down frequency capacity requirements.

[0220] iii. Frequency Modulation Speed ​​Limit

[0221] With the large-scale grid connection of renewable energy, the uncertainty and volatility of the power system have also increased, resulting in a sharp increase or decrease in the net load of the system during the dispatch period, which in turn makes the ramping capability of the unit unable to respond to the system frequency regulation demand in time. Therefore, in order to enable the dispatch decision model to more effectively cope with the rapid changes in the net load of the system, this section introduces the frequency regulation speed constraint, as shown in equations (12) and (13).

[0222]

[0223] In the formula, It is the climbing capability of the unit during the entire dispatch period; The purpose of adding system frequency regulation speed constraints in the present invention is to ensure that the units can follow the system net load fluctuations during the dispatch period, maintain the supply and demand matching within the period, and at the same time, through the relaxation variable ε R,UP / DN , ε H,UP / DN Accurately quantify the flexibility regulation capabilities of power systems.

[0224] iiii. Unit output limitation

[0225] The output of the frequency-regulated unit is shown in formula (14), and the output of the non-frequency-regulated unit is shown in formula (15).

[0226]

[0227] Where P g,min With P g,max are the maximum and minimum values ​​of the unit output. Formulas (14) and (15) indicate that the unit output should be within a certain range during the day-ahead dispatch decision stage, and the frequency-regulating unit needs to reserve a frequency-regulating capacity that is not less than the bid-winning capacity during the dispatch period.

[0228] iiiii. Line power flow constraints

[0229] The line flow constraint ensures that the net load fluctuation does not exceed the maximum carrying capacity of the transmission line, thereby preventing line failures and system stability problems caused by overload. Therefore, the line flow constraint is established for each scheduling period, as shown in formula (16).

[0230]

[0231] Where: l represents the line number, B represents the node set; H lb A represents the transfer distribution factor matrix element between the injected power and the active power flow; b The matrix element representing the relationship between unit g and node b; P l max Indicates the maximum line transmission capacity limit.

[0232] Embodiment 12:

[0233] The verification of a quantitative evaluation method for the adequacy of the overall regulation capacity of the power system taking into account the day-ahead frequency regulation demand of the power grid is as follows:

[0234] (1) Data Collection

[0235] In order to verify the effectiveness of the quantitative evaluation method for the overall regulation capacity of the power system taking into account the day-ahead frequency regulation demand of the power grid proposed in this invention, the actual operation data of the load and renewable energy of a power grid in a certain province of China from January 2023 to December 2023 were collected. The net load data under high renewable energy penetration was obtained by scaling up the renewable energy data. At the same time, in order to consider the impact of meteorological conditions on the system frequency regulation characteristics, various meteorological data in the region were collected.

[0236] (2) Sample feature vector construction

[0237] The input feature vector is the fluctuation characteristics of new energy, load, net load, etc. of the previous day, as well as the meteorological factors on the forecast day, totaling 3170 dimensions; the output feature vector is the maximum fluctuation of the frequency modulation characteristics on the forecast day, totaling 96 dimensions.

[0238] (3) Sample pretreatment

[0239] The meteorological factors, new energy fluctuation characteristics, load fluctuation characteristics and frequency modulation characteristics contained in the sample have different dimensions and large numerical differences, which is not conducive to the training of the DNN model. Therefore, the sample needs to be normalized and preprocessed. The present invention uses the z-score method to normalize the input data and output data of the sample, as shown in formula (1). The z-score standardization method uses the sample mean and standard deviation for normalization processing. The data processed by z-score satisfies the mean of 0 and the standard deviation of 1.

[0240]

[0241] In the formula, x * is the normalized sample value, x is the sample value to be normalized, μ is the sample mean, and σ is the sample standard deviation.

[0242] (4) Training samples and input feature screening

[0243] The prediction accuracy of the maximum fluctuation prediction model of frequency modulation features is closely dependent on the similarity between the training samples and the samples on the prediction day, as well as the matching degree between the input features of the training samples and the samples on the prediction day. Therefore, this section first quantifies the matching degree of sample features and selects samples with high matching degree as training samples; then, quantifies the matching degree between the input features of the training samples and the samples on the prediction day, and selects features with high matching degree as the final sample input features.

[0244] (5) Prediction of the maximum fluctuation point of frequency modulation characteristics

[0245] Deep neural networks (DNNs) have shown excellent prediction performance due to their powerful ability in nonlinear feature extraction, efficient processing of large-scale data, and robustness in noisy data environments, and their computing time can meet scheduling requirements. Therefore, the present invention chooses to use DNN to construct a prediction model that takes into account the maximum fluctuation of power grid frequency regulation characteristics. Since the data-driven prediction model is not the core part of the present invention, traditional machine learning algorithms (such as LSTM, decision trees, etc.) can also be used to predict the maximum fluctuation of power grid frequency regulation characteristics as an alternative data-driven model.

[0246] (6) Frequency modulation feature interval prediction

[0247] A point prediction is made for the maximum fluctuation of the frequency modulation characteristics during the scheduling period. However, due to the existence of data-driven errors, the predicted maximum fluctuation of the frequency modulation characteristics may be smaller than the true value, resulting in insufficient reserved capacity for system auxiliary services, affecting the regulation performance of the system. To avoid this situation, the present invention adopts an integrated method based on ABKDE for interval prediction. By selecting the upper limit of the interval prediction as the maximum fluctuation value of the frequency modulation characteristics, the risk of the predicted value being smaller than the true value can be avoided to a great extent. The present invention uses the point prediction result of DNN as input and further uses ABKDE for interval prediction.

[0248] (7) Quantitative evaluation model of the overall regulation capacity of the power system based on the relaxation constraint of the day-ahead frequency regulation demand

[0249] In the traditional day-ahead dispatch decision model, the frequency regulation capacity and frequency regulation speed are embedded using the upper limit of the frequency regulation characteristic interval prediction as the boundary condition of the dispatch decision model. On the basis of satisfying power balance, the model evaluates the adequacy of the power system regulation capacity through the frequency regulation capacity relaxation constraint.

[0250] (8) Solve the quantitative evaluation model of the overall regulation capacity of the power system taking into account the power grid's day-ahead frequency regulation demand

[0251] The predicted upper limit value of the system's maximum short-term power fluctuation through the ABKDE interval is taken as the system frequency modulation speed demand, thereby the frequency modulation speed demand value is quantified.

[0252] The specific simulation results are as follows:

[0253] 1) Example description

[0254] This section will use the power grid operation data of a province in my country (based on historical meteorological information, etc.) to verify the effectiveness of the proposed point prediction model and interval prediction model. This study uses the sliding window method for load forecasting, that is, for each date to be predicted, a corresponding load forecasting model is constructed and trained, and the training sample data remains fixed. For example: when predicting the load data on February 11, the input samples used include all data from February 1 to February 10. If you want to predict the data on February 12, use the data from February 2 to February 11, and so on.

[0255] The unit parameters used in the example are as follows:

[0256] The IEEE30 node includes 6 units, and the energy quotation and frequency regulation capacity quotation of each unit, the maximum / minimum frequency regulation capacity of each unit, the maximum / minimum power and the unit climbing rate in a short period of time are listed, as shown in Table 2.

[0257] Table 2 Unit parameters

[0258]

[0259] 2) Comparative analysis of simulation results

[0260] ① Verification of the effectiveness of the data-driven quantification method for the power grid’s day-ahead frequency regulation capacity demand considering power fluctuations within the time period

[0261] i. Comparison of training methods

[0262] The present invention will compare the following methods:

[0263] M1: Set the frequency regulation capacity demand to a percentage of the load forecast: during peak load periods, set the frequency regulation capacity demand to 3% of the predicted peak load; during valley load periods, set the frequency regulation capacity demand to 3% of the predicted valley load;

[0264] M2: Set the frequency regulation capacity requirement to three times the standard deviation of the net load forecast error;

[0265] M3: Only DNN data-driven prediction model is used;

[0266] M4: Adds interval prediction model based on M3.

[0267] ii. Scheduling results analysis

[0268] The scheduling decision model before the festival will obtain the scheduling results for the next day. The scheduling period is 15 minutes, and one scheduling period contains 15 short periods, that is, each short period is 1 minute. After obtaining the scheduling results for the whole day, the following effect of different methods for the net load in actual operation in each scheduling period is simulated. If the net load cannot be followed, a penalty cost needs to be borne to evaluate the system operation performance under different scheduling decision methods.

[0269] In order to verify the effectiveness of the method M4 proposed in the present invention, the present invention selects the operating data on February 11, 2023 to test the operating performance of the system under methods M1-M4. The comparison indicators include frequency regulation demand satisfaction rate and economic cost (energy cost, capacity cost, penalty cost). Among them, the penalty cost coefficient is set to 100. The indicator values ​​of different methods are shown in Table 3.

[0270] Table 3 M1-M4 frequency regulation capacity demand satisfaction rate and economic cost

[0271] Frequency modulation demand satisfaction rate (%) Energy cost ($) Capacity cost ($) Penalty cost ($) Total cost ($) M1 100％ 213289.6 9818.9 15454.6 238563.1 M2 100％ 213289.6 17856.2 3711.6 234857.4 M3 57.3％ 213289.6 5423.6 33784.6 252497.8 M4 100％ 213289.6 15854.9 0 229144.5

[0272] As shown in Table 3, compared with the data obtained by methods M1-M4, since methods M1-M2 determine the system frequency regulation capacity demand according to the load peak of 3% during the dispatch period, the frequency regulation capacity cost of method M1 is higher than that of M3; since methods M1-M2 do not consider the short-term ramping capability of the unit during the dispatch period, it is impossible to take into account the sudden increase or decrease of the system in a short period of time, resulting in the inability to effectively follow the net load fluctuations in the real-time dispatch stage of the system, and thus methods M1-M2 demand to bear more penalty costs. Method M3 only predicts the frequency regulation capacity demand, but due to the existence of point prediction errors, the prediction is inaccurate, affecting the dispatch results, and thus generating greater penalty costs. On the other hand, since method M7 can effectively follow the net load power fluctuations in the actual dispatch stage, it does not need to bear penalty costs. The capacity cost and penalty cost of M1-M2 are higher than that of M4, mainly because M1-M4 does not fully consider the actual fluctuations of the power grid. At some dispatching points, reserving more frequency regulation capacity leads to waste, while at other dispatching points, the reserved frequency regulation capacity is insufficient to meet the system frequency regulation capacity demand, resulting in the inability to effectively follow the net load fluctuation, which in turn increases the penalty cost. Finally, the total cost borne by method M4 is 96.06%, 97.57%, and 90.76% of M1-M3, respectively.

[0273] Based on the actual power grid data, the present invention uses the new energy data and load data to calculate the net load, thereby determining the frequency regulation capacity demand of the system. The calculation results show that the proposed DNN-based point prediction model and ABKDE interval prediction quantification method can ensure that the upper limit of the prediction interval effectively covers the actual demand value of the power grid frequency regulation capacity demand, while ensuring the system security and economy.

[0274] ②Verification of the effectiveness of the data-driven quantification method for the day-ahead frequency regulation speed demand of the power grid considering power fluctuations within the time period

[0275] The present invention adopts the evaluation index R 2 , MAE, RMSE and MAPE are used to discuss the accuracy of the point prediction model, and PICP and PINAW are used to discuss the interval prediction model.

[0276] (1) Coefficient of Determination (R 2 )

[0277]

[0278] In the formula, y i is the true value, is the model prediction value, is the average of the true values. 2 The value range is 0-1, and the larger the value, the stronger the explanatory power of the model.

[0279] (2) Mean Absolute Error (MAE)

[0280]

[0281] MAE is the average of the absolute values ​​of the differences between the predicted values ​​and the true values. The smaller the value, the better.

[0282] (3) Root Mean Square Error (RMSE)

[0283]

[0284] RMSE is a standardized measure of error. The square term will penalize the prediction error more heavily, and the smaller the value, the better.

[0285] (4) Mean Absolute Percentage Error (MAPE)

[0286]

[0287] MAPE is the average value of the ratio of the absolute value of the prediction error to the true value. It is used to measure the accuracy of the prediction. The smaller the value, the better.

[0288] (5) Prediction Interval Coverage Probability (PICP)

[0289]

[0290] Where, L n and U n are the upper and lower bounds of the nth predicted value respectively. I is the indicator function, which takes a value between 0% and 100%. The larger the value, the better.

[0291] (6) Prediction Interval Normalized Average Width (PINAW)

[0292]

[0293] Where Y max ,Y min are the maximum and minimum values ​​of the true value, respectively. PINAW is an indicator to measure the width of the prediction interval, which aims to balance the accuracy and reliability of the prediction. The smaller the value, the better.

[0294] In order to effectively prove the effectiveness of the data-driven quantification method for the day-ahead frequency regulation speed demand of the power grid taking into account the power fluctuation within the time period proposed in the present invention, this section conducts simulation verification on the data-driven quantification method for the frequency regulation speed demand. The data used are the data from February 1 to May 30, 2023 to construct the training samples and the data on May 30 to construct the input features. The average R2, MAE, RMSE, MAPE and coverage rate (whether it can effectively cover the system frequency regulation speed demand) of method M3 on May 31, 2023 are shown in Table 4.

[0295] Table 4 Method M3 frequency modulation speed point prediction index

[0296]

[0297] As shown in Table 4, the point prediction model can predict the system frequency modulation speed demand relatively accurately. However, due to the existence of prediction errors, it cannot completely cover the system frequency modulation speed demand and ensure the system frequency demand. Therefore, it is necessary to perform interval prediction. Then, the ABKDE interval prediction model is introduced. The PINAW and PICP values ​​of method M4 are shown in Table 5.

[0298] Table 5 Method M4 interval prediction indicators

[0299]

[0300] Compared with method M3, method M4 reserves more frequency modulation speed requirements, which can ensure that the short-term fluctuation characteristics of the net load can be effectively dealt with during the scheduling period, and finally the frequency modulation speed requirements of the period can be met, ensuring the safety and frequency performance of the system, which can further verify the importance of interval prediction.

[0301] ③Analysis of quantitative evaluation results of the overall regulation capacity adequacy of the power system based on the relaxation constraint of the day-ahead frequency regulation demand

[0302] The system frequency regulation capacity demand and frequency regulation speed demand obtained by the "Validation of the Data-driven Quantification Method for the Day-ahead Frequency Regulation Demand of the Power Grid Considering the Power Fluctuation in the Time Period" technology are used as boundary conditions and embedded in the day-ahead dispatch decision model that accurately characterizes the flexibility regulation capability of the power system. The regulation capability of the power system participating in transactions at different times is analyzed. The adequacy of the power system's flexibility regulation resources and the power system's regulation demand at the first 10 moments of dispatch is shown in Table 5 (a positive adequacy indicates that additional regulation capability can still be provided under the premise of satisfying the constraints of the system's own power balance; a negative adequacy indicates that the system's own flexibility regulation resources are insufficient to meet the system's regulation needs, and regulation capability needs to be purchased from other regions).

[0303] Table 5 Power system flexibility regulation capability and system regulation demand margin

[0304]

[0305] As shown in Table 4, in the first 10 time periods, the regulation capacity margins in the 1st, 3rd, 5th, 6th, 7th, 8th, 9th, and 10th time periods are positive, indicating that the system has sufficient regulation capacity in these time periods and can provide regulation support for other regions. In the 4th and 10th time periods, the regulation capacity is negative, indicating that the system has insufficient regulation capacity in these time periods, and the power supply may not meet the demand, and the regulation capacity needs to be obtained from other regions. For the regulation speed, the margins are positive in the 1st, 2nd, 3rd, 6th, 7th, 8th, and 9th time periods, indicating that the system can respond quickly and meet the demand in these time periods. However, in the 4th, 5th, and 10th time periods, the regulation speed margins are negative, indicating that the system has insufficient response speed in these time periods and may not be able to cope with the short-term fluctuations of the system, resulting in untimely regulation. Especially in the 4th time period, although the system's flexible regulation resources can meet the demand for frequency regulation capacity, the ramp rate of the regulation resources is insufficient due to the drastic fluctuations in a short period of time, so the system needs to rely on the regulation capabilities of other regions. In general, positive values ​​of sufficiency indicate that the system can provide regulatory support to external regions, while negative values ​​mean that external regions are required to provide regulation to make up for the shortfall.

[0306] The present invention proposes a method for evaluating the adequacy of the overall regulation capacity of the power system taking into account the day-ahead frequency regulation demand of the power grid, so as to promote accurate transactions of regulation capacity between different regions. The present invention first quantifies the system frequency regulation speed and frequency regulation capacity requirements based on the system frequency regulation demand data that takes into account short-term power fluctuations within the time period. Finally, the frequency regulation speed requirement and the frequency regulation capacity requirement are embedded in a quantitative evaluation model of the adequacy of the overall regulation capacity of the power system based on the relaxation constraint of the day-ahead frequency regulation demand, and the adequacy of the regulation capacity of the system flexibility regulation resources and the system regulation capacity requirement are obtained, and the regulation capacity of the power system participating in transactions in different regions is accurately quantified. The simulation results of the example show that the method obtained by the present invention can effectively cover the maximum value of the actual power grid frequency regulation demand, ensure the adequacy of the system set demand, and thus lay the foundation for the evaluation of the adequacy of the overall regulation capacity of the power system.

Claims

1. A quantitative evaluation method for the overall regulation capacity adequacy of a power system taking into account the day-ahead frequency regulation demand of the power grid, characterized in that: The following steps are involved: 1) Taking into account the power fluctuations within the time period, the demand for grid frequency regulation characteristics is predicted; 2) Construct a quantitative assessment model for the overall regulation capacity adequacy of the power system based on the relaxation constraint of the day-ahead frequency regulation demand; 3) Solve the quantitative evaluation model of the overall regulation capacity margin of the power system to obtain the regulation capacity margin of the power system under the grid frequency regulation characteristic requirements in step 1).

2. According to claim 1, a quantitative evaluation method for the overall regulation capacity adequacy of a power system taking into account the day-ahead frequency regulation demand of the power grid is characterized in that: The steps for predicting the power grid forecast daily frequency regulation demand data include: 1) Construct the input characteristics of power grid frequency regulation demand, including new energy fluctuations, load fluctuations and meteorological factors; 2) Build a DNN deterministic prediction model; 3) Input the input features of the power grid frequency regulation demand into the DNN deterministic prediction model to obtain the power grid forecast day frequency regulation demand data, including the frequency regulation capacity demand and frequency regulation speed demand for each dispatching period on the forecast day; 4) The ABKDE method is used to perform interval prediction on the power grid's predicted daily frequency regulation demand data, and the confidence interval of the power grid's predicted daily frequency regulation demand data is obtained. The upper limit value of the data in the confidence interval is used as the power grid's frequency regulation characteristic demand to achieve quantification of the frequency regulation capacity and frequency regulation speed requirements.

3. A quantitative evaluation method for the adequacy of the overall regulation capacity of the power system taking into account the day-ahead frequency regulation demand of the power grid according to claim 2, characterized in that: The input characteristics of the power grid frequency regulation demand include the new energy, load and net load fluctuation characteristics of the previous day, as well as the meteorological factors of the forecast day; The load fluctuation characteristics include the maximum value, minimum value, mean value and / or median value of the load power in each dispatching period one day before the forecast date; The new energy fluctuation characteristics include the maximum value, minimum value, mean value and / or median value of the new energy power in each dispatching period one day before the forecast date, and the maximum value, minimum value, mean value and / or median value of the frequency regulation capacity demand in each dispatching period one day before the forecast date; The net load fluctuation characteristics include the maximum value, minimum value, mean value and / or median value of the net load power in each dispatching period one day before the forecast date, and the maximum value, minimum value, mean value and / or median value of the short-term power fluctuation of the net load in each dispatching period one day before the forecast date; The meteorological factors on the forecast day include the holiday type on the forecast day, the highest, lowest and / or average temperature in each dispatch period on the forecast day, the highest, lowest and / or average humidity in each dispatch period on the forecast day, and the highest, lowest and / or average rainfall in each dispatch period on the forecast day; The holiday type of the prediction day is a 0-1 variable, 1 means the prediction day is a holiday, 0 means it is not a holiday.

4. The method for quantitatively evaluating the adequacy of the overall regulation capacity of the power system taking into account the day-ahead frequency regulation demand of the power grid according to claim 2 is characterized in that: The DNN deterministic prediction model is trained using the power grid day-ahead frequency regulation demand dataset; The power grid's day-ahead frequency regulation demand data set includes multiple sets of input and output data; the input data is the input characteristics of the power grid's frequency regulation demand, including renewable energy fluctuations, load fluctuations and meteorological factors; the output data is the power grid's predicted day-ahead frequency regulation demand data, including the frequency regulation capacity demand and frequency regulation speed demand for each dispatching period on the predicted day.

5. The method for quantitatively evaluating the adequacy of the overall regulation capacity of the power system taking into account the day-ahead frequency regulation demand of the power grid according to claim 2 is characterized in that: The steps of using the ABKDE method to make interval predictions on the power grid predicted daily frequency regulation demand data include: 1) Construct the ABKDE model, namely: Where h0 is the global baseline bandwidth, λ is the adjustment parameter, g(x) is the density estimate at x, f(x) is the probability density function; x is the daily frequency demand data of the power grid output by the DNN deterministic prediction model; K is the kernel function, n is the number of samples, and x is the probability density function. i is the sample point predicted by DNN; h i is bandwidth; 2) Construct the cumulative distribution function F(x), that is: In the formula, the cumulative distribution function F(x) represents the probability that the random variable X is less than or equal to a specific value x; 3) Calculate the predicted value x' with a 95% confidence interval, using the confidence interval of the power grid's predicted daily frequency regulation demand data; The predicted value x' is as follows: x'=F -1 (p) (4) Where p is a value between 0 and 1.

6. A quantitative evaluation method for the overall regulation capability adequacy of a power system taking into account the day-ahead frequency regulation demand of the power grid according to claim 1, characterized in that: The objective function of the quantitative evaluation model of the power system overall regulation capacity margin based on the relaxation constraint of the day-ahead frequency regulation demand is as follows: In the formula, g and t are the unit number and the dispatch period number respectively; G and T are the number of units and the dispatch time set respectively; Indicates the unit output; Quote energy prices for units; Quote for the up and down frequency regulation capacity of the unit; The up and down frequency regulation capacity of the unit won the bid; 1 / T e It is the proportion of the duration of a single scheduling period in the day-ahead scheduling decision model to 1 hour; M and N are the penalty coefficients for the shortage of frequency regulation capacity demand and frequency regulation speed demand; Provides the system with frequency modulation capacity up and down; It is the relaxation amount required for the up and down frequency modulation speeds.

7. A quantitative evaluation method for the adequacy of the overall regulation capacity of the power system taking into account the day-ahead frequency regulation demand of the power grid according to claim 1, characterized in that: The power system regulation capability is achieved through the system up and down frequency regulation capacity And the relaxation of the upper and lower frequency modulation speed requirements Quantification.

8. The method for quantitatively evaluating the adequacy of the overall regulation capability of the power system taking into account the day-ahead frequency regulation demand of the power grid according to claim 1, characterized in that: The constraints of the quantitative evaluation model for the adequacy of the overall regulation capacity of the power system based on the relaxation constraint of the day-ahead frequency regulation demand include power balance constraint, frequency regulation capacity constraint, frequency regulation speed limit, unit output limit and line flow constraint.

9. A quantitative evaluation method for the adequacy of the overall regulation capacity of the power system taking into account the day-ahead frequency regulation demand of the power grid according to claim 8, characterized in that: The power balance constraints are as follows: In the formula, b is the node number; B is the node set; is the net load during the dispatch period. Indicates the unit output; The frequency regulation capacity constraints are as follows: In the formula, The system's up and down frequency modulation capacity requirements; The upper and lower frequency regulation capacity of the unit's winning bid; R is the up and down frequency modulation capacity of the system; g,max,UP , R g,max,DN It is the upper limit of the system's up and down frequency modulation capacity; The FM speed limits are as follows: In the formula, It is the climbing capability of the unit during the entire dispatch period; The predicted up and down frequency regulation capacity of adjacent net load; The relaxation amount required for the upper and lower frequency modulation speeds; is the coefficient; The unit output limits are as follows: Where P g,min With P g,max It is the maximum and minimum output of the unit. The line flow constraints are as follows: Where: l represents the line number, B represents the node set; H lb A represents the transfer distribution factor matrix element between the injected power and the active power flow; b The matrix element representing the relationship between unit g and node b; Indicates the maximum line transmission capacity limit.

Images