A method for determining reserve capacity and allocating its cost considering the random fluctuation characteristics of renewable energy

Abstract

The present invention belongs to the field of power systems and automation thereof, and particularly relates to a method for determining reserve capacity and allocating its cost taking into account the random fluctuation characteristics of renewable energy. The method comprises the following steps: utilizing historical load / renewable energy forecast data and fitting the probability density curve of each load / renewable energy source based on an uncertainty estimation method of non-parametric kernel density; determining the total reserve capacity by simulating and generating a reserve demand scenario based on the Monte Carlo method; constructing a market clearing model with minimizing the unit operating cost and the rotating reserve cost as the objective function; solving the clearing model with a quadratic programming algorithm to obtain the energy price and the reserve price, and determining the total amount of reserve auxiliary funds to be allocated; calculating the replacement value of each load / renewable energy source based on a VCG mechanism, and allocating the reserve auxiliary fees according to the relative proportion of the value. The present invention can reasonably determine the total reserve capacity and effectively avoid the problem of imbalance between revenue and expenditure, and can be widely applied to the power spot market of the power system.

Description

Technical Field

[0001] The present invention belongs to the field of power systems and automation thereof, and in particular relates to a method for determining standby capacity and allocating its cost taking into account the random fluctuation characteristics of new energy. Background Art

[0002] The power system operates under various uncertainties, such as the randomness of load demand and the volatility of renewable energy output. To ensure real-time balance between sources and loads, reserving reserve capacity is a necessary measure to address the significant uncertainty in the power system. This is particularly true in the context of the "dual carbon" initiative, where the large-scale and high-proportion integration of renewable energy introduces even greater uncertainty. As the power market gradually advances, electricity will become a commodity, and reserving reserve capacity will become an ancillary service that should be paid for. Determining reserve capacity and appropriately allocating it to responsible parties is key to building a scientific and rational reserve ancillary service market.

[0003] Numerous studies have addressed the aforementioned issues. First, regarding reserve capacity determination, current research primarily assumes a known total reserve demand and focuses on determining the optimal reserve capacity for each unit. However, such deterministic approaches fail to account for the randomness of load and renewable energy. As the scale of renewable energy integration increases, these approaches may be overly conservative or under-reserved, directly impacting the economics and security of the power grid. Regarding reserve cost allocation, this patent focuses on the locational marginal pricing (LMP) approach. This is primarily due to the LMP pricing mechanism's clear physical significance and widespread adoption in my country and around the world. Currently, power markets in the United States, including PJM, Texas, California, New York, and New England, have all adopted LMP. Among my country's first eight electricity spot market pilot programs, markets in Guangdong and Shandong also adopted the LMP mechanism. Under the LMP pricing mechanism, using Lagrange multiplier analysis, the reserve price can be defined as the change in total system cost associated with an increase in reserve capacity by one unit. From this, we can conclude that the reserve cost provided by each unit is the reserve price multiplied by the reserve capacity provided by the unit. However, who will share this reserve cost is a pressing issue in the reserve ancillary service industry. Failure to reasonably and fairly allocate reserve costs among market participants will inevitably dampen their enthusiasm for competing in the reserve market, ultimately jeopardizing the safe and economical operation of the power grid.

[0004] Scholars have conducted some research to address these issues, and their solutions can be broadly categorized into two types. One approach assumes that reserve demand is a linear function of load or renewable energy. Using the LMP electricity pricing mechanism, the reserve costs borne by load and renewable energy can be directly derived. This approach proposes a pricing mechanism with clear physical meaning and favorable properties, such as incentive compatibility and individual rationality. This characteristic stems from the fact that the assumed linear function explicitly captures the contribution of load / renewable energy to reserve. However, in reality, reserve capacity is reserved to address grid uncertainty. Therefore, the size of reserve capacity is not directly related to the size of load / renewable energy, but rather to the magnitude of its forecast error. Therefore, while these approaches can produce pricing methods that incorporate reserve costs, they rely on assumptions that are difficult to replicate in reality. Another approach assumes that load / renewable energy follows a specific distribution and determines reserve cost allocation through analogies like value-at-risk or opportunity constraints. These approaches offer promising ideas for effectively allocating reserve costs and have considerable practical application. However, their implementation relies on certain assumptions that differ significantly from actual operating conditions, making the accuracy and rationality of the calculated results difficult to guarantee.

[0005] In summary, in order to ensure the healthy and rapid development of the electricity market, it is urgent to study a method for determining reserve capacity and its cost allocation that takes into account the random fluctuation characteristics of renewable energy to ensure the rationality and fairness of the market. Summary of the Invention

[0006] In order to provide technical support for the scientific and rational construction of a reserve ancillary service market and ensure the rationality and fairness of the market, the present invention provides a method for determining reserve capacity and allocating its costs, taking into account the random fluctuation characteristics of renewable energy. The method specifically includes the following steps:

[0007] Through the historical forecast data and measured data of load / new energy, the random fluctuation characteristics of load / new energy are analyzed and characterized based on the non-parametric kernel density estimation method, and the probability density curve of each load / new energy is established;

[0008] Perform scenario simulations based on random sampling of the probability density curves of each uncertainty, and calculate the corresponding backup demand based on each scenario;

[0009] The backup capacity required for each scenario is sorted according to the uncertainty ratio that needs to be addressed, which is generally set to above 80%. The total backup capacity is determined according to the quantile of the backup capacity distribution after sorting. That is, if there are n samples (corresponding to n backup requirements), they are sorted in descending order according to the quantile of the backup capacity distribution after sorting, and the top α% (uncertainty ratio) is taken as the backup capacity.

[0010] Taking the minimization of unit operating cost and spinning reserve cost as the objective function, a market clearing model is constructed;

[0011] A quadratic programming algorithm is used to solve the clearing model, obtain the energy price and reserve price, and determine the total amount of reserve assistance that needs to be allocated;

[0012] The replacement value of each load / new energy source is calculated according to the VCG mechanism, and the standby auxiliary costs are allocated according to the relative proportion of the value.

[0013] Furthermore, the probability density curve of load / new energy is expressed as:

[0014]

[0015] Among them, f t (x) is the probability density function of load demand / new energy output x at time t; C is the number of all available historical data at time t, x i is the i-th sample of historical data, is the kernel function, and h is the bandwidth.

[0016] Furthermore, the kernel function adopts Gaussian kernel function, then the kernel function Expressed as:

[0017]

[0018] Furthermore, with the minimum unit operating cost and spinning reserve cost as the objective function, the market clearing model is constructed, including:

[0019] Objective function:

[0020]

[0021] Constraints on system power balance:

[0022]

[0023] Constraints on thermal stability of system lines:

[0024]

[0025] Constraints on the maximum upper and lower spinning reserves of system units:

[0026]

[0027] Maximum and minimum output constraints of thermal power units:

[0028]

[0029] Maximum upper and lower spinning reserve constraints of the unit:

[0030]

[0031] Output constraints of wind turbines:

[0032]

[0033] in, is the fuel cost of the unit, is the upper spinning reserve cost of the unit, is the lower spinning reserve cost of the unit; T is the time set; R is the set of units, R={RG, RW}, RG represents the set of thermal power units, RW represents the set of wind power units; P g,t is the output of thermal power unit g at time t; P w,t is the output of wind turbine w at time t; D d,t represents the demand of load d at time t, and RD represents the load set; is the upper limit of branch transmission power; PTDF is the branch power; USR g,t The upper spinning reserve provided by thermal power unit g at time t, F UR,t is the total demand for spinning reserve; DSR g,t is the lower spinning reserve provided by thermal power unit g at time t, F DR,t is the total demand for spinning reserve; P g,t,max is the maximum output of the unit at time t; U g,t The start and stop status of the thermal power unit g at time t. When its value is 1, it indicates the start state, and when its value is 0, it indicates the stop state; is the maximum output of thermal power unit g; R U P is the maximum drop rate for shutdown and the maximum rise rate for startup; g,t,min is the minimum output of the unit at time t; κ is a parameter ranging from 0 to 1; P w,f The maximum output of the wind turbine.

[0034] Furthermore, the allocation of standby auxiliary expenses according to the relative proportion of value includes:

[0035]

[0036] in, represents the standby ancillary service fee paid by the system operator to unit i; P is the incremental cost of the system for each additional unit of reserve demand; t i is the output of unit i at time t; is the value of unit i calculated by the replacement cost of other units at time t; RD represents the load set, and RW represents the wind turbine.

[0037] Furthermore, if The value of generator set i is calculated by the substitution cost of other generator sets at any time and is expressed as:

[0038]

[0039] in, is the optimal dispatch plan for other generators when generator i does not participate in the market, is the optimal dispatch plan for other generators when generator i participates in the market; is the quotation set of other generators when generator i does not participate in the market, is the quote set of all generators when generator i participates in the market; is the cost of the optimal dispatch plan for other generators when generator i does not participate in the market; R ＞0,-i is the spinning reserve demand when generator set i does not participate in the market; R ＜0,-i The spinning reserve demand when generator set i does not participate in the market; is the system cost when generator set i does not participate in the market, is the system cost when generator i participates in the market. The system cost is calculated according to the objective function of the clearing model; is π i The elements in represent the value of unit i calculated by the replacement cost of other units at time t.

[0040] Furthermore, scenario simulation is performed by randomly sampling the probability density curves of each uncertainty, and the corresponding backup demand is calculated according to each scenario, including the following steps:

[0041] Get the total forecast error set R under the load and new energy combination mode at time t, expressed as:

[0042]

[0043] The total forecast error set R is divided into upper spinning reserve and lower spinning reserve. The upper spinning reserve corresponds to the state of sudden load increase or renewable energy output lower than expected, that is, the reserve demand when the value of the total forecast error set R is greater than zero, which is recorded as R ＞0 The lower spinning reserve corresponds to the state of sudden load reduction or higher-than-expected renewable energy output, that is, the reserve demand when the value of the total forecast error set R is less than zero, denoted as R ＜0 .

[0044] Furthermore, the required backup capacity for each scenario is ranked, and the total backup capacity is determined according to the quantile of the ranked backup capacity distribution based on the proportion of uncertainty to be addressed. The total backup capacity needs to meet the following requirements:

[0045]

[0046] in, is the ceiling function; represents the total reserve demand to cope with the first α% of the total forecast error of load and renewable energy, and n is the number of samples of historical data.

[0047] Furthermore, the process of selecting the number of samples of historical data is as follows:

[0048]

[0049]

[0050] in, To calculate the spare capacity F by Monte Carlo sampling R,t The result obtained by approximating the mean of V(F R,t ) is the spare capacity F R,t variance; for The variance of .

[0051] The present invention has the following beneficial effects:

[0052] 1) The present invention can effectively determine the reserve capacity based on the fluctuation characteristics of uncertainty, and reasonably allocate the reserve cost to the load / new energy users according to the principle of "whoever causes it, bears it";

[0053] 2) The scenario simulation-based method for determining the reserve capacity proposed in this invention can simulate the distribution characteristics of various complex and uncertain factors and reasonably determine the total reserve capacity;

[0054] 3) The proposed VCG-based backup cost allocation method uses the load / renewable energy substitution benefits to reasonably and fairly allocate costs, making it more reasonable and economical than simply adding up each user's backup requirements. Furthermore, the cost calculated by the VCG mechanism is positively correlated with the user's substitution benefits (value), effectively avoiding revenue and expenditure imbalances.

[0055] 4) The present invention can be widely applied to the power spot market of the power system, and is particularly suitable for the market settlement of new energy participating in the standby auxiliary market. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1The probability density fitting result of the output prediction error of wind turbine 1 at 13:00 in the embodiment of the present invention is shown;

[0057] Figure 2 This is the probability density fitting result of the output prediction error of wind turbine 2 at 16:00 in an embodiment of the present invention;

[0058] Figure 3 1. A graph showing a trend of the spinning reserve demand variance coefficient changing with the number of sampling samples at different times in an embodiment of the present invention;

[0059] Figure 4 1 is a graph showing a trend of the spinning reserve demand variance coefficient changing with the number of sampling samples at different times in an embodiment of the present invention;

[0060] Figure 5 A graph showing a change trend of the spinning reserve capacity at different times as the proportion of uncertainty that can be addressed is shown in an embodiment of the present invention;

[0061] Figure 6 The total upper spinning reserve demand to cope with 99% uncertainty at different times in the embodiment of the present invention;

[0062] Figure 7 1. A graph showing a change trend of the spinning reserve capacity as a function of the proportion of uncertainty that can be addressed at different times in an embodiment of the present invention;

[0063] Figure 8 is the total spinning reserve demand to cope with 99% uncertainty at different times in the embodiment of the present invention;

[0064] Figure 9 is the sum of the backup demands of each user at different times in the embodiment of the present invention;

[0065] Figure 10 The value ratio of each user at different times in the embodiment of the present invention;

[0066] Figure 11 is the variance of the prediction error of each user at different times in the embodiment of the present invention;

[0067] Figure 12 The standby auxiliary service fee payable by each user at different times in the embodiment of the present invention;

[0068] Figure 13 The standby auxiliary service fees that different users need to pay in the embodiments of the present invention;

[0069] Figure 14 This is a flow chart of a method for determining reserve capacity and allocating its costs taking into account the random fluctuation characteristics of new energy sources according to the present invention. DETAILED DESCRIPTION

[0070] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0071] The present invention provides a method for determining the reserve capacity and its cost allocation taking into account the random fluctuation characteristics of new energy. Figure 14 , specifically including the following steps:

[0072] Through the historical forecast data and measured data of load / new energy, the random fluctuation characteristics of load / new energy are analyzed and characterized based on the non-parametric kernel density estimation method, and the probability density curve of each load / new energy is established;

[0073] Perform scenario simulations based on random sampling of the probability density curves of each uncertainty, and calculate the corresponding backup demand based on each scenario;

[0074] Sort the backup capacity required for each scenario, and determine the total backup capacity based on the quantile of the backup capacity distribution after sorting, based on the proportion of uncertainty that needs to be addressed.

[0075] Taking the minimization of unit operating cost and spinning reserve cost as the objective function, a market clearing model is constructed;

[0076] A quadratic programming algorithm is used to solve the clearing model, obtain the energy price and reserve price, and determine the total amount of reserve assistance that needs to be allocated;

[0077] The replacement value of each load / new energy source is calculated according to the VCG mechanism, and the standby auxiliary costs are allocated according to the relative proportion of the value.

[0078] In this embodiment, firstly, using historical load / new energy forecast data, an uncertainty estimation method based on non-parametric kernel density is proposed, without assuming that the load / new energy follows a fixed distribution; on this basis, a backup demand scenario simulation generation method based on the Monte Carlo method is proposed to determine the total backup capacity. Secondly, according to the Vickrey-Clarke-Groves (VCG) theory, the value of a unit is defined as the substitution benefit of the unit to other units, and the backup cost is proportionally allocated to the load / new energy according to the value. Finally, a modified IEEE 30-node system is used for simulation analysis to verify the effectiveness of the proposed method. This embodiment describes the solution of the present invention in detail from the above three aspects.

[0079] (1) Method for determining reserve capacity based on scenario simulation

[0080] This embodiment divides the process of determining the reserve capacity into two parts. One part is to calculate the probability density curve of the load demand / new energy output at time k based on the forecast error sequence of the historical load / new energy at each moment through an uncertainty distribution fitting method based on non-parametric kernel density. The other part is to determine the total reserve capacity based on the obtained probability density curve through Monte Carlo simulation.

[0081] 1) Uncertainty distribution fitting method based on nonparametric kernel density

[0082] The probability density curve can effectively characterize the fluctuation characteristics of the prediction error of uncertain factors such as new energy. How to estimate it through the historical data of load / new energy prediction errors is the difficulty in characterizing the fluctuation characteristics of new energy and other factors. Existing probability density estimation methods can be divided into parameter estimation methods and non-parametric estimation methods. The parameter estimation method usually assumes that the uncertainty factors obey a certain specific distribution. For the actual complex and changeable operating conditions, the above method is difficult to apply. Among the non-parametric estimation methods, the kernel density estimation method can start from the characteristics of the data itself without making any prior assumptions about the data distribution, and has a wide range of applications. The probability density function f of the load demand / new energy output x at time t is estimated by the non-parametric kernel density method. t (x), can be expressed as:

[0083]

[0084] Among them, C is the number of all available historical data at time t, x i is the i-th sample, is the kernel function, and h is the bandwidth.

[0085] Kernel functions include Gaussian kernel function, Triangle kernel function, Uniform kernel function, and Epannechnikov kernel function. Since the Gaussian kernel function is relatively easy to calculate and has the excellent property of being continuously differentiable at any order, this embodiment selects the Gaussian kernel function as the kernel function K for non-parametric density estimation. Its calculation formula is:

[0086]

[0087] Therefore, based on the historical load / renewable energy forecast error sequence at each moment, the probability density curve of each load demand / renewable energy output at time k can be obtained based on formulas (1)-(2). After calculating the probability density function, the distribution of the grinding error of this function is used to sample and calculate the backup demand corresponding to each sample according to this distribution. The demands are sorted, and the total demand is the first α% of the sorted backup demand set.

[0088] 2) Method for determining total reserve capacity based on Monte Carlo simulation

[0089] At a certain time t, the total prediction error set R under the combined mode of positioning load and new energy is:

[0090]

[0091] Among them, F D and F G are the prediction error vectors corresponding to all loads and new energy sources, respectively, and X is the distribution space of load and new energy prediction errors; is the backup demand corresponding to the tth moment of the i-th sample set, n is the number of samples that can effectively cover the prediction error sample space; {R i | i＝1:n} represents a set of alternative requirements.

[0092] In the auxiliary service reserve, there are two types of auxiliary services: upper spinning reserve and lower spinning reserve. Upper spinning reserve is used to deal with sudden load increase or renewable energy output lower than expected, while lower spinning reserve is the opposite.

[0093] R={R ＞0 ,R ＜0} (4)

[0094] Among them, R ＞0 For upper spinning reserve, R ＜0 It is worth noting that when using formula (3) to calculate the reserve demand, it is necessary to divide it according to whether the reserve demand is greater than zero, that is, when the value of R is greater than 0, it is the upper spin reserve R. ＞0 When the value of R is less than 0, it is the backspin standby R ＜0 Calculate the upper spinning reserve or lower spinning reserve R according to the actual situation ＜0 The following is a further explanation using the spinning reserve as an example.

[0095] Arrange the spare requirement sizes of all sample sets in ascending order to obtain a new set As shown below:

[0096]

[0097] In order to cope with the load and total forecast error of new energy in the first α% of the X distribution space, the total reserve demand should meet the following requirements:

[0098]

[0099] in, is the ceiling function. Therefore, It can be defined as the total reserve demand to cope with the first α% (uncertainty ratio) of the total forecast error of load and new energy. However, it is worth noting that the sample space of the load and new energy forecast error is generally very large, and the number of enumerations tends to infinity. How to determine the finite number of samples n is the key to determining the total demand for reserve capacity. Therefore, this embodiment is based on the Monte Carlo method to perform theoretical derivation to establish the relationship between the number of sampling samples n and the sample space approximation accuracy β, and to guide how to determine the effective number of samples n in actual engineering applications. The core idea of ​​Monte Carlo is to establish a stationary distribution P and a priori probability distribution of the system. Therefore, the reserve capacity F is determined by Monte Carlo sampling. R,t Mean E(F R,t ) to approximate:

[0100]

[0101] According to the law of large numbers, when n→∞, the spare capacity F R,t The approximate mean of Converges to the mathematical expectation E(F with probability 1 R,t ).at the same time, The variance of can be expressed as:

[0102]

[0103] Among them, V(F R,t ) is estimated to be:

[0104]

[0105] According to the definition of t distribution:

[0106]

[0107] where t(n) is the t distribution with n degrees of freedom.

[0108] It can be seen that when the number of sampling points n is sufficiently large, It approximately obeys the t distribution with n degrees of freedom. Then at a certain confidence level, we have:

[0109]

[0110] When n approaches infinity, the t distribution approaches the normal distribution. At this time, the convergence of the Monte Carlo method depends on The variance of the estimate. The error of the estimate is expressed using the coefficient of variance:

[0111]

[0112] Combining formula (8) and formula (12), we can get:

[0113]

[0114] From formula (13), we can get that the number of sampling n and the variance coefficient β 2 Inversely proportional to the actual prediction process, the variance coefficient can be determined according to the accuracy requirements and the number of sampling n can be determined. Moreover, from formula (13), it can be seen that its convergence condition is almost unaffected by the scale and complexity of the system, and it has good conditions for engineering application and promotion.

[0115] (2) Backup cost allocation method based on VCG mechanism

[0116] This embodiment introduces the market clearing model of standby services and defines the total price of standby services. On this basis, a standby cost allocation method based on VCG theory is proposed, which allocates the cost in equal proportion according to the replacement cost of each responsible party.

[0117] 1) Market Clearing Model with Backup Service Demand

[0118] The market clearing model usually adopts the DC power flow model, with the minimum unit operating cost and spinning reserve cost as the objective function. It must meet the load balance, unit operating characteristics, thermal stability constraints, etc. The market clearing model is expressed as:

[0119] Objective function:

[0120]

[0121] Where, and They correspond to the unit's fuel cost, upper spinning reserve cost, and lower spinning reserve cost respectively.

[0122] System constraints:

[0123]

[0124] Where, P g,t is the output of thermal power unit g at time t, P w,t is the output of wind turbine w at time t, D d,t is the demand of load d at time t; USR g,t 、DSR g,t They are the upper and lower spinning reserves provided by thermal power unit g at time t; F UR,t and F DR,t are the total upper and lower spinning reserve demands at time t, respectively.

[0125] Unit operating constraints:

[0126]

[0127]

[0128] Among them, S D and S U The maximum down-rate at shutdown and the maximum up-rate at startup, the present invention simplifies the up and down climbing rates to be consistent, that is, R U ;U g,t The start and stop status of the thermal power unit g at time t. When its value is 1, it indicates the start state, and when its value is 0, it indicates the stop state.

[0129] Formula (15) is the system power balance constraint; Formula (16) is the line thermal stability constraint; Formulas (17) and (18) are the spinning reserve capacity constraints; Formulas (19)-(21) are the maximum and minimum output constraints of thermal power units; Formulas (22)-(25) are the maximum up / down spinning reserve constraints of the units; Formula (26) is the output constraint of the wind turbine unit.

[0130] The energy price is the locational marginal pricing (LMP) at each bus. The LMP at node i represents the additional system cost (also known as the system incremental cost) required to meet the load demand of one unit at node i. Therefore, according to formulas (14)-(26), the node electricity price LMP is t i Expressed as:

[0131]

[0132] Where L is the Lagrangian function of the clearing model (14)-(26), is the active power demand of load i at time t.

[0133] The energy cost paid by the Independent System Operator (ISO) to market member (thermal power unit and wind power unit) i is for:

[0134]

[0135] The energy cost W charged by ISO from load i i t,LMP for:

[0136]

[0137] Since LMP is essentially derived from the system increment required to meet load demand, the incremental cost (i.e., the amount paid to market members) will be less than or equal to the cost required to meet load demand, thus ensuring the ISO's energy cost balance.

[0138] According to the node electricity price mechanism, the node price of the backup cost is is the incremental cost of the system for each additional unit of reserve demand, that is:

[0139]

[0140] Where L is the Lagrangian function of the clearing model formulas (14)-(26).

[0141] Then, the total amount of reserve auxiliary service that needs to be allocated, that is, the reserve auxiliary service fee paid by the ISO to thermal power unit i, is:

[0142]

[0143] 2) Backup cost sharing principle based on VCG mechanism

[0144] Unlike energy costs, ancillary service costs are paid by the party responsible for generating backup demand. This includes not only the load side but also renewable energy generators. However, the domestic power market only charges for backup ancillary services based on loads, and this collection method is relatively simplistic. This long-term trend will seriously impact the healthy development of the power market. To address this, this embodiment proposes a backup cost principle based on VCG theory.

[0145] The VCG mechanism is a mechanism design theory that incentivizes market participants to report truthful information and achieve incentive compatibility. It offers advantages such as incentive compatibility, individual rationality, and optimal social welfare, but it cannot achieve a balanced budget. Currently, scholars commonly use VCG theory to study pricing design in different markets. This invention combines the marginal price mechanism with the VCG mechanism to achieve complementary advantages. It uses marginal prices to define the payment costs received by thermal power units and VCG theory to define the standby replacement benefits of each responsible party. Standby costs are then allocated proportionally to achieve a balanced budget.

[0146] According to the VCG theory, the value of a generator set is the substitution benefit of the generator set to other generator sets, that is, the change in the total power generation cost of other generator sets before and after the generator set participates in the market, which can be expressed as formula (32):

[0147]

[0148] Where, is the optimal dispatch plan for other generators when generator i does not participate in the market; is the optimal dispatching plan of other generators when generator i participates in the market; R ＞0,-i With R ＞0,-i are the upper spinning reserve demand and the lower spinning reserve demand when generator set i does not participate in the market; represents the set of bids of generator units participating in the market and the bid of generator unit i is not included in the set; is the set of bids of generator units participating in the market, and the set includes the bid of generator unit i; Indicates a generator set The quotation for generator set i is given in this embodiment. The bold letters in this embodiment represent a set of data in the scheme, and the non-bold letters represent an element in the set, i.e. express The price of generator set i in brackets Indicates a quotation in a certain scheduling plan, and in some cases, no additional explanation is given in brackets; is the system cost function when generator i does not participate in the market; is the shorthand representation of the objective function of the electricity spot market clearing model, that is, the shorthand representation of formula (14). Therefore, the first term The first term represents the total cost of other generators when generator i does not participate in the market, and the second term represents the total cost of other generators when generator i participates in the market, that is, the total system cost minus the cost of generator i. i It is the value of unit i calculated by the replacement cost of other units.

[0149] According to the above theory, the present invention regards the load as a negative generator, and the standby auxiliary service cost is settled according to the substitution benefit of load / new energy, as shown in formula (33):

[0150]

[0151] (3) Simulation analysis

[0152] This embodiment uses an IEEE 30-node system connected to new energy sources as the basis for simulation analysis to verify the effectiveness of the method proposed in the present invention. A wind farm with a capacity of no more than 100MW is connected to nodes 5 and 20, respectively. The prediction error of the wind farm is scaled according to the real data provided in the literature. The load forecast output is the given value of the IEEE 30-node system, and its prediction error is assumed to be normally distributed (with a mean of a given value and a variance of 10% of the predicted value). All simulations are performed on a PC equipped with an Intel(R) Core(TM) i7-7500U CPU @ 2.70GHz 32GB RAM.

[0153] 1) Validation of the method for determining the reserve capacity

[0154] This embodiment first verifies the effectiveness of the proposed uncertainty probability distribution feature fitting based on non-parametric kernel density; on this basis, the effectiveness of the proposed backup capacity determination method is verified.

[0155] Figure 1 and Figure 2 The following are the probability density fitting results of the output prediction errors of wind turbines 1 and 2 at 13:00 and 16:00, respectively. Among them, the rectangle is the probability density diagram obtained based on actual data statistics, and the solid line is the probability density curve fitting result of the method proposed in the present invention. It can be seen from the figure that the proposed method can well follow the distribution changes of actual data and has a good fit with the actual data distribution. At the same time, it can also be found that there is a large gap between the wind power output prediction error and the normal distribution, and it is difficult to effectively approximate the complex probability distribution characteristics of real data using parameter estimation methods. Therefore, the proposed non-parametric kernel density estimation method can better fit the probability distribution curve of the data and effectively reflect the fluctuation characteristics of new energy.

[0156] Figure 3 and Figure 4 The following are the trend diagrams of the variance coefficient of the upper / lower rotating standby demand changing with the number of sampling samples at different times. As can be seen from the figure, for the upper / lower rotating standby demand, when the number of samples reaches more than 300, the corresponding variance coefficient has become flat and can reach below 5%. This phenomenon shows that the distribution characteristics of the current sampled samples basically no longer change, and sampling can be stopped. It can also be seen from the figure that the variance coefficient of the upper rotating standby demand has greater jitter in the early stage, and the number of samples required to converge to the same variance coefficient is slightly more. Through detailed observation of the data, we found that the main reason is that the uncertainty factors of the upper rotating standby demand are more than those of the lower rotating standby demand, and its change pattern is more complicated. Therefore, more samples are needed to effectively reflect the distribution characteristics of the upper rotating standby demand.

[0157] After sampling the samples that can effectively reflect the fluctuation characteristics of various uncertainties through the Monte Carlo method, the upper spinning reserve and lower spinning reserve capacities calculated according to the proportion of uncertainties that can be dealt with are as follows: Figure 5 and Figure 7 As shown in the figure, different curves represent the backup demand at different times. As can be seen from the figure, no matter at any time, Figure 5 and Figure 7 The curve of reserve capacity demand as the proportion of uncertainty addressed shows a clear turning point. This means that to cope with greater uncertainty, the required reserve capacity will increase dramatically, and the corresponding costs will explode. Therefore, choosing the turning point as the final total reserve demand will provide a more reasonable overall cost and benefit.

[0158] Figure 6 and Figure 8These are the upper and lower spinning reserve requirements at each turning point. As can be seen from the figure, the upper and lower spinning reserve requirements vary at different times, and their changing trends are also distinct. Different uncertainty factors contribute differently to reserve capacity and correspond to different replacement costs. The next section will discuss how to rationally allocate costs based on their value (i.e., replacement costs).

[0159] 2) Verification of the effectiveness of the alternative cost allocation method

[0160] If the spatiotemporal correlation between the uncertainties is not considered, the total reserve demand for upward rotation and downward rotation obtained by directly adding the reserve demand of each load / new energy source is as follows: Figure 9 As shown in the figure, the upper spinning reserve capacity ranges from 60MW to 89MW, while the lower spinning reserve ranges from 61MW to 86MW. This difference is more than double the reserve capacity determined by the method proposed in this patent, resulting in a significant increase in economic costs, a scarcity of reserve resources, and an increase in reserve prices.

[0161] Through the VCG mechanism, the substitution benefits of each user at each moment can be obtained, which account for Figure 10 Among them, at most moments, the proportion of certain users is relatively high, and only at time 1 (corresponding to the line with the highest proportion in the figure) and time 3 (corresponding to the line with the lowest proportion in the figure) does a large difference appear. Figure 11 The variance of the prediction error for different users at different times (user 4 has the largest variance in the figure). The relative size of these variances indicates that they are inconsistent with the value contribution of each user. Therefore, the existing method of directly using variance to determine user backup costs is clearly unreasonable. Figure 12 is the standby ancillary service fee that each user should pay at different times. Since the total standby ancillary service fee at each time varies greatly, although the value proportion of most users at each time is similar, the standby ancillary service price to be paid at each time shows obvious differences, which shows that the proposed method can effectively reflect the differences in standby ancillary services of each user. And because the proposed method calculates the payment fee of each user in proportion to the total standby ancillary service fee, there is no imbalance between income and expenditure. Finally, Figure 13 The standby auxiliary service fees payable by different users are shown in Figure 3, with User 3 paying the highest fee, reaching $14. User 23 (i.e., wind turbine 2) pays the lowest fee, only $3, less than 22% of User 3's. This shows that standby auxiliary service fees are not simply affected by a single factor, but rather exhibit complex characteristics influenced by multiple factors. The present invention utilizes the substitution benefits of other units to effectively characterize the value of each user's standby consumption, thereby achieving reasonable allocation of standby auxiliary service fees according to the principle of "whoever causes it, pays for it."

[0162] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A method for determining reserve capacity and allocating its cost taking into account the random fluctuation characteristics of new energy, characterized by: The following steps are involved: Through the historical forecast data and measured data of load demand and renewable energy output, the random fluctuation characteristics of load demand and renewable energy output are analyzed and characterized based on the non-parametric kernel density estimation method, and the probability density curves of each load demand and renewable energy output are established; Perform scenario simulations based on random sampling of the probability density curves of each uncertainty, and calculate the corresponding backup demand based on each scenario; Sort the backup capacity required for each scenario, and determine the total backup capacity based on the quantile of the backup capacity distribution after sorting, based on the proportion of uncertainty that needs to be addressed. Taking the minimization of unit operating cost and spinning reserve cost as the objective function, a market clearing model is constructed; A quadratic programming algorithm is used to solve the clearing model and determine the total amount of reserve assistance that needs to be allocated; The VCG mechanism calculates the replacement value of each load demand and renewable energy output, and allocates the standby auxiliary costs according to the relative proportion of the value, including: in, represents the standby ancillary service fee paid by the system operator to unit i; is the incremental cost of the system for each additional unit of reserve demand; is the output of unit i at time t; is the value of unit i calculated by the replacement cost of other units at time t; RD represents the load set, and RW represents the wind turbine.

2. The method for determining reserve capacity and allocating its cost taking into account the random fluctuation characteristics of new energy according to claim 1, characterized in that: The probability density curve of load demand and renewable energy output is expressed as: Among them, f t (x) is the probability density function of load demand / new energy output x at time t; C is the number of all available historical data at time t, x i is the i-th sample of historical data, is the kernel function, and h is the bandwidth.

3. The method for determining reserve capacity and allocating its cost taking into account the random fluctuation characteristics of new energy according to claim 2, characterized in that: The kernel function uses the Gaussian kernel function, then the kernel function Expressed as:

4. The method for determining reserve capacity and allocating its cost taking into account the random fluctuation characteristics of new energy according to claim 1, characterized in that: Taking the minimization of unit operating cost and spinning reserve cost as the objective function, the market clearing model is constructed including: Objective function: Constraints on system power balance: Constraints on thermal stability of system lines: Constraints on the maximum upper and lower spinning reserves of system units: Maximum and minimum output constraints of thermal power units: Maximum upper and lower spinning reserve constraints of the unit: Output constraints of wind turbines: in, is the fuel cost of the unit, is the upper spinning reserve cost of the unit, is the lower spinning reserve cost of the unit; T is the time set; R is the set of units, R={RG, RW}, RG represents the set of thermal power units; P g,t is the output of thermal power unit g at time t; P w,t is the output of wind turbine w at time t; D d,t represents the demand of load d at time t, and RD represents the load set; is the upper limit of branch transmission power; PTDF is the branch power; USR g,t The upper spinning reserve provided by thermal power unit g at time t, F UR,t is the total demand for spinning reserve; DSR g,t is the lower spinning reserve provided by thermal power unit g at time t, F DR,t is the total demand for spinning reserve; P g,t,max is the maximum output of the unit at time t; U g,t The start and stop status of the thermal power unit g at time t. When its value is 1, it indicates the start state, and when its value is 0, it indicates the stop state; is the maximum output of thermal power unit g; R U P is the maximum drop rate for shutdown and the maximum rise rate for startup; g,t,min is the minimum output of the unit at time t; κ is a parameter ranging from 0 to 1; P w,f The maximum output of the wind turbine.

5. The method for determining reserve capacity and allocating its cost taking into account the random fluctuation characteristics of new energy according to claim 1, characterized in that: like The value of generator set i is calculated by the substitution cost of other generator sets at any time and is expressed as: in, is the optimal dispatch plan for other generators when generator i does not participate in the market, is the cost of the optimal dispatch plan for other generators when generator i does not participate in the market; R ＞0,-i is the spinning reserve demand when generator set i does not participate in the market; R ＜0,-i The spinning reserve demand when generator set i does not participate in the market; is the system cost when generator set i does not participate in the market, is the system cost when generator i participates in the market. The system cost is calculated according to the objective function of the clearing model; is π i The elements in represent the value of unit i calculated by the replacement cost of other units at time t.

6. The method for determining reserve capacity and allocating its cost taking into account the random fluctuation characteristics of new energy according to claim 1, characterized in that: Scenario simulation is performed based on random sampling of the probability density curves of each uncertainty. Calculating the corresponding backup demand for each scenario includes the following steps: Get the total forecast error set R under the load and new energy combination mode at time t, expressed as: The total forecast error set R is divided into upper spinning reserve and lower spinning reserve. The upper spinning reserve corresponds to the state of sudden load increase or renewable energy output lower than expected, that is, the reserve demand when the value of the total forecast error set R is greater than zero, which is recorded as R ＞0 The lower spinning reserve corresponds to the state of sudden load reduction or renewable energy output higher than expected, that is, the reserve demand when the value of the total prediction error set R is less than zero, denoted as R ＜0 .

7. The method for determining reserve capacity and allocating its cost taking into account the random fluctuation characteristics of new energy according to claim 1, characterized in that: Sort the backup capacity required for each scenario. Based on the proportion of uncertainty to be addressed and the quantile of the sorted backup capacity distribution, determine the total backup capacity size. The total backup capacity size needs to meet the following requirements: in, is the ceiling function; represents the total reserve demand to cope with the first α% of the total forecast error of load and renewable energy, and n is the number of samples of historical data.

8. The method for determining reserve capacity and allocating its cost taking into account the random fluctuation characteristics of new energy according to claim 1, characterized in that: The process of selecting the number of samples of historical data is as follows: in, To calculate the spare capacity F by Monte Carlo sampling R,t The result obtained by approximating the mean of V(F R,t ) is the spare capacity F R,t variance; for The variance of .

Images