A method and device for determining future annual load time series scenarios

By constructing a quadratic planning mathematical model, combining the randomness and correlation of load characteristic indicators in the future and historical years, multiple load time series scenarios are generated, and the problem of single and high similarity of load time series scenarios in the existing technology is solved. More accurate forecasts of future annual load changes and calculations of new energy consumption capacity have been achieved.

CN108805388BActive Publication Date: 2024-11-12CHINA ELECTRIC POWER RESEARCH INSTITUTE CO LTD +2
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN201810311225.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2018-04-09
Publication Date
2024-11-12
Estimated Expiration
2038-04-09

AI Technical Summary

Technical Problem

The existing method of generating time series scenarios for future annual load is too single, and the randomness and correlation of historical load characteristic indicators are not fully considered, resulting in the generated load time series scenarios that are too similar to historical loads and cannot fully reflect future annual loads. The changes are difficult to meet the needs of new energy consumption capacity calculation.

Method used

By determining the correlation coefficient matrix of the edge probability distribution of daily load characteristic indexes for the future year and the error of daily load characteristic indexes in history, a quadratic planning mathematical model is constructed to generate a future annual load time series scenario that meets the daily load characteristic law, considering The randomness and correlation factors of historical load characteristic indicators generate multiple different load time series scenarios.

Benefits of technology

The generated future annual load time series scenarios can more accurately reflect the load changes in the future year, adapt to the needs of new energy consumption capacity calculation, avoid the problem of excessive similarity with historical load scenarios, and provide more reasonable load prediction.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN108805388B_ABST
    Figure CN108805388B_ABST
Patent Text Reader

Abstract

The present invention provides a method and device for determining a future year load time series scenario. First, the marginal probability distribution of daily load characteristic indicators for each day of the future year, the correlation coefficient matrix of the errors of daily load characteristic indicators in historical years, and the daily load characteristic indicators for each day of the future year are determined; then the daily load time series scenario for each day of the future year is determined, and finally the load time series scenario for the future year is determined. The present invention takes into account the randomness and correlation factors of the historical load characteristic indicators, generates a series of different future year load time series scenarios, and can meet the needs of calculating the new energy absorption capacity. The future year daily load time series scenarios obtained by the present invention based on the daily load characteristic indicators for each day of the future year have similar daily load characteristic indicators, and their rest day curves and working day curves are consistent with the corresponding historical loads, ensuring the rationality of the generated future year load time series scenarios.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of power grid planning, and in particular to a method and device for determining a future annual load time series scenario. Background Art

[0002] In recent years, the problem of new energy consumption has become increasingly prominent. Quantitative analysis of the new energy consumption capacity of the power grid in future years has important guiding significance for the rational planning of new energy installation in the power grid. Time series production simulation technology is an important technical means to quantitatively analyze the new energy consumption capacity in future years. It can fully consider the operation constraints of various power sources, as well as the random fluctuation characteristics of load and new energy output. By performing hourly production simulation calculations on the power system throughout the year, the power generation of new energy and conventional power sources in each period can be obtained. The annual load time series in the future is the key input for carrying out time series production simulation calculations, which directly determines the accuracy of the calculation results of new energy consumption capacity. Due to the random fluctuation characteristics of load, a single load time series scenario is difficult to fully describe the changes in load in the future. Therefore, in order to meet the needs of new energy consumption capacity calculation, it is necessary to consider the random changes and distribution laws of load characteristics such as daily load power consumption and daily load peak-to-valley difference rate, and generate a large number of load time series scenarios through random simulation. Therefore, in order to meet the needs of new energy consumption production simulation calculations, it is urgent to propose a way to determine the annual load time series scenario, and use cluster analysis, mathematical statistics and optimization methods to generate future annual load time series scenarios that meet the daily load characteristic laws.

[0003] The existing future annual load time series scenarios are mainly based on given load rate, peak-to-valley difference rate and other future annual daily load characteristic indicators, which are obtained by scaling the historical load data. The existing methods do not consider the randomness and correlation factors of the historical load characteristic indicators. The generated load time series scenarios are too simple and too similar to the historical load scenarios. They cannot fully reflect the changes in future annual loads and are difficult to meet the needs of calculating the new energy absorption capacity. Summary of the invention

[0004] In order to overcome the deficiency of the above-mentioned prior art that the generated load time series scenario is too single and too similar to the historical load scenario, resulting in difficulty in adapting to the calculation needs of new energy absorption capacity, the present invention provides a method and device for determining the load time series scenario of future years. First, the marginal probability distribution of daily load characteristic indicators of each day in the future year and the correlation coefficient matrix of the errors of daily load characteristic indicators of historical years are determined, and the daily load characteristic indicators of each day in the future year are determined; then the daily load time series scenario of each day in the future year is determined, and finally the load time series scenario of the future year is determined. The present invention takes into account the randomness and correlation factors of the historical load characteristic indicators, generates a series of different load time series scenarios of the future years, avoids the situation where the similarity with the historical load scenarios is too high, and can adapt to the calculation needs of new energy absorption capacity.

[0005] In order to achieve the above-mentioned object of the invention, the present invention adopts the following technical scheme:

[0006] In one aspect, the present invention provides a method for determining a future annual load time series scenario, comprising:

[0007] Determine the marginal probability distribution of the daily load characteristic indicators of the future years and the correlation coefficient matrix of the daily load characteristic indicator errors of the historical years according to the daily load characteristic indicators of the historical years, and determine the daily load characteristic indicators of the future years according to the marginal probability distribution of the daily load characteristic indicators of the future years and the correlation coefficient matrix of the daily load characteristic indicator errors of the historical years;

[0008] Determine the daily load time series scenarios for future years based on the pre-built quadratic programming mathematical model;

[0009] Determine the load time series scenario for the future year based on the daily load time series scenario for each day of the future year;

[0010] The quadratic programming mathematical model is constructed based on the daily load characteristic indicators of each day in the future years.

[0011] Before determining the marginal probability distribution and correlation coefficient matrix of the daily load characteristic index errors of each day in the historical year according to the daily load characteristic index of each day in the historical year, the method includes:

[0012] Determine the daily load characteristic index error of each day in the historical year according to the following formula:

[0013]

[0014]

[0015]

[0016]

[0017]

[0018]

[0019] in, represents the daily load rate error of the mth working day in the historical year, It represents the daily peak-to-valley rate error of the mth working day in the historical year. Indicates the maximum daily load error of the mth working day in the historical year; M work Indicates the total number of working days in the historical year; γ work (m) represents the maximum daily load on the mth working day of the historical year, and γ work (m) = max{l t (m)},l t (m) represents the load time series at time t on the mth working day of the historical year; α work (m) represents the daily load rate of the mth working day in the historical year, and β work (m) represents the maximum daily load on the mth working day of the historical year, and β work (m) = (γ work (m)-min{l t (m)}) / γ work (m); Indicates the daily load rate error of the nth rest day in the historical year, It represents the daily peak-to-valley rate error of the nth rest day in the historical year. Indicates the daily maximum load error of the nth rest day in the historical year; N off Indicates the total number of rest days in a historical year; γ off (n) represents the maximum daily load of the nth rest day in the historical year, and γ off (n) = max{l t (n)},l t (n) represents the load time series at time t on the nth rest day of the historical year; α off (n) represents the daily load rate of the nth rest day in the historical year, and β off (n) represents the peak-to-valley difference rate of the nth rest day in the historical year, and β off (n) = (γ off (n)-min{l t (n)}) / γ off (n).

[0020] The marginal probability distribution of the daily load characteristic index of the future year is as follows:

[0021]

[0022]

[0023] Among them, x represents the independent variable, represents the marginal probability distribution of daily load rate on future working days, Represents the marginal probability distribution of the daily peak-to-valley difference rate of future working days, Represents the marginal probability distribution of the daily maximum load on future working days;

[0024] express Pan The probability density function after It represents the base value of daily load factor for future working days. represents the probability density function of the daily load factor error on working days in the historical year, and h represents the kernel function bandwidth; express Pan The probability density function after Indicates the benchmark value of the daily peak-to-valley difference rate for future working days. represents the probability density function of the daily peak-to-valley rate error of the historical working days, and express Pan The probability density function after Indicates the daily maximum load reference value for future working days. represents the probability density function of the daily maximum load error on working days in the historical year, and express Pan The probability density function after Indicates the base value of daily load factor for rest days in the future. represents the probability density function of the daily load factor error on rest days in historical years, and express Pan The probability density function after The benchmark value of the daily peak-to-valley difference rate for rest days in the future years, represents the probability density function of the daily peak-to-valley rate error of the historical rest days, and express Pan The probability density function after Indicates the daily maximum load reference value for the rest days in the future. represents the probability density function of the daily maximum load error on rest days in historical years, and

[0025] The correlation coefficient matrix of the historical annual daily load characteristic index error is as follows:

[0026]

[0027] Among them, Σ work The correlation coefficient matrix representing the daily load characteristic index error of the historical working days, Σ off The correlation coefficient matrix representing the error of daily load characteristic index during the historical rest year; ρ1, ρ2, ρ3, ρ4, ρ5, ρ6 all represent correlation coefficients, which are calculated as follows:

[0028]

[0029]

[0030]

[0031] The method of determining the daily load characteristic index for each day in the future year according to the marginal probability distribution of the daily load characteristic index for each day in the future year and the correlation coefficient matrix of the daily load characteristic index errors in the historical years includes:

[0032] Determine the total number of days in the future year, the type and arrangement order of each day in the future year, judge the type of the j-th day in the future year, if the j-th day in the future year is a working day, determine the daily load rate, daily peak-to-valley difference rate and daily maximum load of the working day in the future year; if the j-th day in the future year is a rest day, determine the daily load rate, daily peak-to-valley difference rate and daily maximum load of the rest day in the future year.

[0033] The daily load rate, daily peak-to-valley difference rate and daily maximum load of the working days in the future year include:

[0034] Based on Σ work Construct a multidimensional Gaussian distribution N(0,Σ work ), and randomly sample to generate random samples (y1, y2, y3);

[0035] Based on the random sample (y1, y2, y3), the daily load rate, daily peak-to-valley difference rate and daily maximum load of the working days in the future year are calculated by the following formula:

[0036]

[0037]

[0038]

[0039] Among them, α work (j) represents the daily load rate of the jth working day in the future year, β work (j) represents the daily peak-to-valley difference rate of the jth working day in the future year, γwork (j) represents the daily maximum load on the jth working day in the future year; represents the inverse function of the marginal probability distribution of daily load factor on future working days, It represents the inverse function of the marginal probability distribution of the daily peak-to-valley difference rate on future working days. Represents the inverse function of the marginal probability distribution of the daily maximum load on future working days.

[0040] The determination of the daily load rate, daily peak-to-valley difference rate and daily maximum load for the future annual rest days includes:

[0041] Based on Σ off Construct a multidimensional Gaussian distribution N(0,Σ off ), and randomly sample to generate random samples (y4,y5,y6);

[0042] Based on the random sample (y4, y5, y6), the daily load rate, daily peak-to-valley difference rate and daily maximum load of rest days in the future year are calculated by the following formula:

[0043]

[0044]

[0045]

[0046] Among them, α off (j) represents the daily load rate of the jth rest day in the future year, β off (j) represents the daily peak-to-valley difference rate of the jth rest day in the future year, γ off (j) represents the maximum daily load on the jth rest day in the future year; represents the inverse function of the marginal probability distribution of the daily load factor on future rest days, The inverse function of the marginal probability distribution of the daily peak-to-valley difference rate of rest days in the future year, Represents the inverse function of the marginal probability distribution of the daily maximum load on future working days.

[0047] The quadratic programming mathematical model includes an objective function and constraints;

[0048] The objective function aims to minimize the sum of squares of the daily load time series scenarios of future years and the daily load benchmark curve of historical years, and is constructed as follows:

[0049]

[0050] Among them, Δ1 represents the sum of square errors between the daily load time series scenario of a certain day in the future year and the daily load benchmark curve of the working day in the historical year, and Δ2 represents the sum of square errors between the daily load time series scenario of a certain day in the future year and the daily load benchmark curve of the rest day in the historical year; x t (j) represents the load at time t on the jth day of the future year; It represents the daily load reference value at time t on the working day of the historical year. It represents the daily load reference value at rest day t in the historical year. and It is obtained by performing k-means clustering on the load sets of working days and rest days in historical years.

[0051] The constraints include the daily load rate constraint on working days, the daily peak-to-valley difference rate constraint, and the daily maximum load constraint;

[0052] The daily load rate constraint of the working day is as follows:

[0053]

[0054] The daily peak-to-valley difference rate constraint on the working day is as follows:

[0055]

[0056] Among them, t min-work Indicates the time when the minimum load of the daily load benchmark curve of the historical working day occurs, x tmin-work (j) represents the jth working day t in the future year min-work The load at the time;

[0057] The daily maximum load constraint of the working day is as follows:

[0058]

[0059] Among them, t max-work Indicates the time when the maximum load of the daily load reference curve of the working days in the historical year occurs.

[0060] The constraints determined according to the daily load characteristic index of rest days in the future include the daily load rate constraint of rest days, the daily peak-to-valley difference rate constraint and the daily maximum load constraint;

[0061] The daily load rate constraint on the rest day is as follows:

[0062]

[0063] The daily peak-to-valley difference rate constraint on the rest day is as follows:

[0064]

[0065] Among them, t min-off Indicates the time when the minimum load of the daily load reference curve appears on rest days in historical years. Indicates the jth holiday t in the future year min-off The load at the time;

[0066] The maximum load constraint on the rest day is as follows:

[0067]

[0068] Among them, t max-off Indicates the time when the maximum load of the daily load reference curve occurs on rest days in historical years.

[0069] The method of determining the daily load time series scenario of a certain day in the future year according to the pre-built quadratic programming mathematical model includes:

[0070] The Cplex mathematical optimization solver is used to solve the quadratic programming mathematical model and obtain the daily load time series scenario d on the jth day of the future year. k (j) = {x t (j), t = 1, 2, ..., 24};

[0071] Assume that the total number of days in the future year is M, execute j=j+1, and determine whether j>M is satisfied. If j>M, connect the generated M daily load time series scenarios according to the order of arrangement of each day in the future year to obtain the load time series scenario {d k (1),d k (2),...,d k (M)}, and execute k=k+1, and when k is greater than the preset number of future year load time series scenarios N, N different future year daily load time series scenarios are obtained; if j≤M, continue to determine the daily load rate, daily peak-to-valley difference rate and daily maximum load of the remaining working days and rest days in the future year, and solve the secondary optimization model according to the daily load rate, daily peak-to-valley difference rate and daily maximum load of the remaining working days and rest days in the future year, until j>M.

[0072] In another aspect, the present invention further provides a device for determining a future annual load time series scenario, comprising:

[0073] The first determination module is used to determine the marginal probability distribution of the daily load characteristic indicators of the future years and the correlation coefficient matrix of the daily load characteristic indicator errors of the historical years according to the daily load characteristic indicators of the historical years, and determine the daily load characteristic indicators of the future years according to the marginal probability distribution of the daily load characteristic indicators of the future years and the correlation coefficient matrix of the daily load characteristic indicator errors of the historical years;

[0074] The second determination module is used to determine the daily load time series scenario of each day in the future year according to the pre-built quadratic programming mathematical model;

[0075] The third determination module is used to determine the future year load time series scenario based on the daily load time series scenario of each day in the future year;

[0076] The quadratic programming mathematical model is constructed based on the daily load characteristic indicators of each day in the future years.

[0077] The device also includes:

[0078] The error determination module is used to determine the daily load characteristic index error of each day in the historical year according to the following formula:

[0079]

[0080]

[0081]

[0082]

[0083]

[0084]

[0085] in, represents the daily load rate error of the mth working day in the historical year, It represents the daily peak-to-valley rate error of the mth working day in the historical year. Indicates the maximum daily load error of the mth working day in the historical year; M work Indicates the total number of working days in the historical year; γ work (m) represents the maximum daily load on the mth working day of the historical year, and γ work (m) = max{l t (m)},l t (m) represents the load time series at time t on the mth working day of the historical year; α work (m) represents the daily load rate of the mth working day in the historical year, and β work (m) represents the maximum daily load on the mth working day of the historical year, and β work (m) = (γ work (m)-min{l t (m)}) / γ work (m); Indicates the daily load rate error of the nth rest day in the historical year, It represents the daily peak-to-valley rate error of the nth rest day in the historical year. Indicates the daily maximum load error of the nth rest day in the historical year; N off Indicates the total number of rest days in a historical year; γ off (n) represents the maximum daily load of the nth rest day in the historical year, and γ off (n) = max{l t (n)},l t (n) represents the load time series at time t on the nth rest day of the historical year; α off (n) represents the daily load rate of the nth rest day in the historical year, and β off (n) represents the peak-to-valley difference rate of the nth rest day in the historical year, and β off (n) = (γ off (n)-min{l t (n)}) / γ off (n).

[0086] The first determining module comprises:

[0087] The marginal probability distribution determination unit is used to determine the marginal probability distribution of the daily load characteristic index of each day in the future year according to the following formula:

[0088]

[0089]

[0090] Among them, x represents the independent variable, represents the marginal probability distribution of daily load rate on future working days, Represents the marginal probability distribution of the daily peak-to-valley difference rate of future working days, Represents the marginal probability distribution of the daily maximum load on future working days;

[0091] express Pan The probability density function after It represents the base value of daily load factor for future working days. represents the probability density function of the daily load factor error on working days in the historical year, and h represents the kernel function bandwidth; express Pan The probability density function after Indicates the benchmark value of the daily peak-to-valley difference rate for future working days. represents the probability density function of the daily peak-to-valley rate error of the historical working days, and express Pan The probability density function after Indicates the daily maximum load reference value for future working days. represents the probability density function of the daily maximum load error on working days in the historical year, and express Pan The probability density function after Indicates the base value of daily load factor for rest days in the future. represents the probability density function of the daily load factor error on rest days in historical years, and express Pan The probability density function after The benchmark value of the daily peak-to-valley difference rate for rest days in the future years, represents the probability density function of the daily peak-to-valley rate error of the historical rest days, and express Pan The probability density function after Indicates the daily maximum load reference value for the rest days in the future. represents the probability density function of the daily maximum load error on rest days in historical years, and

[0092] The first determining module also includes:

[0093] The correlation coefficient matrix determination unit is used to determine the correlation coefficient matrix of the historical annual and daily load characteristic index errors according to the following formula:

[0094]

[0095] Among them, Σ work The correlation coefficient matrix representing the daily load characteristic index error of the historical working days, Σ off The correlation coefficient matrix representing the error of daily load characteristic index during the historical rest year; ρ1, ρ2, ρ3, ρ4, ρ5, ρ6 all represent correlation coefficients, which are calculated as follows:

[0096]

[0097]

[0098]

[0099]

[0100]

[0101]

[0102] The first determining module also includes:

[0103] The daily load characteristic index determination unit is used to determine the total number of days in the future year, the type and arrangement order of each day in the future year, and to judge the type of the j-th day in the future year. If the j-th day in the future year is a working day, the daily load rate, daily peak-to-valley difference rate and daily maximum load of the working day in the future year are determined; if the j-th day in the future year is a rest day, the daily load rate, daily peak-to-valley difference rate and daily maximum load of the rest day in the future year are determined.

[0104] The daily load characteristic index determination unit includes a daily load characteristic index determination unit for working days and a daily load characteristic index determination unit for rest days.

[0105] The working day daily load characteristic index determination unit includes:

[0106] The first random sample generation unit is used to generate a random sample based on Σ work Construct a multidimensional Gaussian distribution N(0,Σ work ), and randomly sample to generate random samples (y1, y2, y3);

[0107] The daily load characteristic index calculation unit of working days is used to calculate the daily load rate, daily peak-to-valley difference rate and daily maximum load of working days in the future year according to the random sample (y1, y2, y3) through the following formula:

[0108]

[0109]

[0110]

[0111] Among them, α work (j) represents the daily load rate of the jth working day in the future year, β work (j) represents the daily peak-to-valley difference rate of the jth working day in the future year, γ work (j) represents the daily maximum load on the jth working day in the future year; represents the inverse function of the marginal probability distribution of daily load factor on future working days, It represents the inverse function of the marginal probability distribution of the daily peak-to-valley difference rate on future working days. Represents the inverse function of the marginal probability distribution of the daily maximum load on future working days.

[0112] The rest day daily load characteristic index determination unit comprises:

[0113] The second random sample generation unit is used to generate a random sample based on Σ off Construct a multidimensional Gaussian distribution N(0,Σ off), and randomly sample to generate random samples (y4,y5,y6);

[0114] The daily load characteristic index calculation unit for rest days is used to calculate the daily load rate, daily peak-to-valley difference rate and daily maximum load of rest days in the future year according to the random sample (y4, y5, y6) through the following formula:

[0115]

[0116]

[0117]

[0118] Among them, α off (j) represents the daily load rate of the jth rest day in the future year, β off (j) represents the daily peak-to-valley difference rate of the jth rest day in the future year, γ off (j) represents the maximum daily load on the jth rest day in the future year; represents the inverse function of the marginal probability distribution of the daily load factor on future rest days, The inverse function of the marginal probability distribution of the daily peak-to-valley difference rate of rest days in the future year, Represents the inverse function of the marginal probability distribution of the daily maximum load on future working days.

[0119] The quadratic programming mathematical model includes an objective function and constraints;

[0120] The second determining module is specifically used for:

[0121] The goal is to minimize the sum of squares of the daily load time series scenarios in the future and the historical daily load benchmark curve, and construct the objective function as follows:

[0122]

[0123]

[0124] Among them, Δ1 represents the sum of square errors between the daily load time series scenario of a certain day in the future year and the daily load benchmark curve of the working day in the historical year, and Δ2 represents the sum of square errors between the daily load time series scenario of a certain day in the future year and the daily load benchmark curve of the rest day in the historical year; x t (j) represents the load at time t on the jth day of the future year; It represents the daily load reference value at time t on the working day of the historical year. It represents the daily load reference value at rest day t in the historical year. and It is obtained by performing k-means clustering on the load sets of working days and rest days in historical years.

[0125] The constraints include the daily load rate constraint on working days, the daily peak-to-valley difference rate constraint, and the daily maximum load constraint;

[0126] The daily load rate constraint of the working day is as follows:

[0127]

[0128] The daily peak-to-valley difference rate constraint on the working day is as follows:

[0129]

[0130] Among them, t min-work Indicates the time when the minimum load of the daily load benchmark curve of the historical working day occurs, x tmin-work (j) represents the jth working day t in the future year min-work The load at the time;

[0131] The daily maximum load constraint of the working day is as follows:

[0132]

[0133] Among them, t max-work Indicates the time when the maximum load of the daily load reference curve of the working days in the historical year occurs.

[0134] The constraints determined according to the daily load characteristic index of rest days in the future include the daily load rate constraint of rest days, the daily peak-to-valley difference rate constraint and the daily maximum load constraint;

[0135] The daily load rate constraint on the rest day is as follows:

[0136]

[0137] The daily peak-to-valley difference rate constraint on the rest day is as follows:

[0138]

[0139] Among them, t min-off Indicates the time when the minimum load of the daily load benchmark curve appears on the rest day of the historical year, x tmin-off (j) represents the jth holiday t in the future year min-off The load at the time;

[0140] The maximum load constraint on the rest day is as follows:

[0141]

[0142] Among them, t max-off Indicates the time when the maximum load of the daily load reference curve occurs on rest days in historical years.

[0143] The third determination module comprises:

[0144] The solving unit is used to solve the quadratic programming mathematical model using the Cplex mathematical optimization solver to obtain the daily load time series scenario d on the jth day of the future year. k (j) = {x t (j), t = 1, 2, ..., 24};

[0145] The judgment unit is used to execute j=j+1 and judge whether j>M is satisfied. If j>M, the generated M daily load time series scenarios are connected according to the order of arrangement of each day in the future year to obtain the load time series scenario {d k (1),d k (2),...,d k (M)}, and execute k=k+1, and when k is greater than the preset number of future year load time series scenarios N, N different future year daily load time series scenarios are obtained; if j≤M, continue to determine the daily load rate, daily peak-to-valley difference rate and daily maximum load of the remaining working days and rest days in the future year, and solve the secondary optimization model according to the daily load rate, daily peak-to-valley difference rate and daily maximum load of the remaining working days and rest days in the future year, until j>M, M represents the total number of days in the future year.

[0146] Compared with the closest prior art, the technical solution provided by the present invention has the following beneficial effects:

[0147] In the method for determining the future year load time series scenario provided by the present invention, the marginal probability distribution of the daily load characteristic index of each day in the future year and the correlation coefficient matrix of the daily load characteristic index error of the historical year are first determined, and the daily load characteristic index of each day in the future year is determined; then the daily load time series scenario of each day in the future year is determined, and finally the load time series scenario of the future year is determined, wherein the quadratic programming mathematical model is constructed according to the daily load characteristic index of each day in the future year. The present invention takes into account the randomness and correlation factors of the historical load characteristic index, generates a series of different future year load time series scenarios, and can meet the calculation needs of the new energy consumption capacity;

[0148] The device for determining the future year load time series scenario provided by the present invention comprises a first determination module, a second determination module and a third determination module. The first determination module is used to determine the marginal probability distribution of the daily load characteristic index of the future year and the correlation coefficient matrix of the daily load characteristic index errors of the historical years according to the daily load characteristic index of the historical years, and determine the daily load characteristic index of the future year according to the marginal probability distribution of the daily load characteristic index of the future year and the correlation coefficient matrix of the daily load characteristic index errors of the historical years; the second determination module is used to determine the daily load time series scenario of the future year according to the pre-constructed quadratic programming mathematical model; the third determination module is used to determine the future year load time series scenario according to the daily load time series scenario of the future year;

[0149] The present invention obtains the daily load characteristic indicators of each day in the future years based on the marginal probability distribution of the daily load characteristic indicators of each day in the future years and the correlation coefficient matrix of the errors of the daily load characteristic indicators of the historical years, and obtains the daily load time series scenarios of each day in the future years by solving the quadratic programming mathematical model based on the daily load characteristic indicators of each day in the future years. The obtained daily load time series scenarios of each day in the future years have similar daily load characteristic indicators, and their rest day curves and working day curves are consistent with the corresponding historical loads, thereby ensuring the rationality of the generated future year load time series scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0150] Figure 1 It is a flow chart of a method for determining a future annual load time series scenario in an embodiment of the present invention. DETAILED DESCRIPTION

[0151] The present invention will be further described in detail below in conjunction with the accompanying drawings.

[0152] Embodiment 1 of the present invention provides a method for determining a future annual load time series scenario, and the specific flow chart is as follows: Figure 1 As shown, the specific process is as follows:

[0153] S101: determining the marginal probability distribution of the daily load characteristic indicators of the future years and the correlation coefficient matrix of the daily load characteristic indicator errors of the historical years according to the daily load characteristic indicators of the historical years, and determining the daily load characteristic indicators of the future years according to the marginal probability distribution of the daily load characteristic indicators of the future years and the correlation coefficient matrix of the daily load characteristic indicator errors of the historical years;

[0154] S102: Determine the daily load time series scenario for each day in the future year according to the pre-built quadratic programming mathematical model;

[0155] S103: Determine the load time series scenario for the future year according to the daily load time series scenario for each day of the future year;

[0156] The quadratic programming mathematical model in S102 is constructed based on the daily load characteristic indicators of each day in the future year.

[0157] Before determining the marginal probability distribution and correlation coefficient matrix of the daily load characteristic index error of each day in the historical year according to the daily load characteristic index of each day in the historical year in S01, the daily load characteristic index error of each day in the historical year is determined according to the following formula:

[0158]

[0159]

[0160]

[0161]

[0162]

[0163]

[0164] in, represents the daily load rate error of the mth working day in the historical year, It represents the daily peak-to-valley rate error of the mth working day in the historical year. Indicates the maximum daily load error of the mth working day in the historical year; M work Indicates the total number of working days in the historical year; γ work (m) represents the maximum daily load on the mth working day of the historical year, and γ work (m) = max{l t (m)},l t (m) represents the load time series at time t on the mth working day of the historical year; α work (m) represents the daily load rate of the mth working day in the historical year, and β work (m) represents the maximum daily load on the mth working day of the historical year, and β work (m) = (γ work (m)-min{l t (m)}) / γ work (m); Indicates the daily load rate error of the nth rest day in the historical year, It represents the daily peak-to-valley rate error of the nth rest day in the historical year. Indicates the daily maximum load error of the nth rest day in the historical year; N off Indicates the total number of rest days in a historical year; γ off (n) represents the maximum daily load of the nth rest day in the historical year, and γ off (n) = max{l t(n)},l t (n) represents the load time series at time t on the nth rest day of the historical year; α off (n) represents the daily load rate of the nth rest day in the historical year, and β off (n) represents the peak-to-valley difference rate of the nth rest day in the historical year, and β off (n) = (γ off (n)-min{l t (n)}) / γ off (n).

[0165] In the above S101, the marginal probability distribution of the daily load characteristic index of each day in the future year is as follows:

[0166]

[0167]

[0168]

[0169]

[0170]

[0171]

[0172] Among them, x represents the independent variable, represents the marginal probability distribution of daily load rate on future working days, Represents the marginal probability distribution of the daily peak-to-valley difference rate of future working days, Represents the marginal probability distribution of the daily maximum load on future working days;

[0173] express Pan The probability density function after It represents the base value of daily load factor for future working days. represents the probability density function of the daily load factor error on working days in the historical year, and h represents the kernel function bandwidth;

[0174] express Pan The probability density function after Indicates the benchmark value of the daily peak-to-valley difference rate for future working days. represents the probability density function of the daily peak-to-valley rate error of the historical working days, and

[0175] express Pan The probability density function after Indicates the daily maximum load reference value for future working days. represents the probability density function of the daily maximum load error on working days in the historical year, and

[0176] express Pan The probability density function after Indicates the base value of daily load factor for rest days in the future. represents the probability density function of the daily load factor error on rest days in historical years, and

[0177] express Pan The probability density function after The benchmark value of the daily peak-to-valley difference rate for rest days in the future years, represents the probability density function of the daily peak-to-valley rate error of the historical rest days, and

[0178] express Pan The probability density function after Indicates the daily maximum load reference value for the rest days in the future. represents the probability density function of the daily maximum load error on rest days in historical years, and

[0179] In the above S101, the correlation coefficient matrix of the historical annual daily load characteristic index error is as follows:

[0180]

[0181]

[0182] Among them, Σ work The correlation coefficient matrix representing the daily load characteristic index error of the historical working days, Σ off The correlation coefficient matrix representing the error of daily load characteristic index during the historical rest year; ρ1, ρ2, ρ3, ρ4, ρ5, ρ6 all represent correlation coefficients, which are calculated as follows:

[0183]

[0184]

[0185]

[0186]

[0187]

[0188]

[0189] In the above S01, the daily load characteristic index of each day in the future year is determined according to the marginal probability distribution of the daily load characteristic index of each day in the future year and the correlation coefficient matrix of the daily load characteristic index error of the historical year. The specific process is as follows:

[0190] 1. Determine the total number of days in the future year, the type and arrangement order of each day in the future year;

[0191] 2. Determine the type of the j-th day in the future year. If the j-th day in the future year is a working day, determine the daily load rate, daily peak-to-valley difference rate and daily maximum load of the working day in the future year; if the j-th day in the future year is a rest day, determine the daily load rate, daily peak-to-valley difference rate and daily maximum load of the rest day in the future year.

[0192] For working days, the specific process of determining the daily load rate, daily peak-to-valley difference rate and daily maximum load of working days in the future year is as follows:

[0193] 1) Based on Σ work Construct a multidimensional Gaussian distribution N(0,Σ work ), and randomly sample to generate random samples (y1, y2, y3);

[0194] 2) Based on the random sample (y1, y2, y3), the daily load rate, daily peak-to-valley difference rate and daily maximum load of the working days in the future year are calculated by the following formula:

[0195]

[0196]

[0197]

[0198] Among them, α work (j) represents the daily load rate of the jth working day in the future year, β work (j) represents the daily peak-to-valley difference rate of the jth working day in the future year, γ work (j) represents the daily maximum load on the jth working day in the future year; represents the inverse function of the marginal probability distribution of daily load factor on future working days, It represents the inverse function of the marginal probability distribution of the daily peak-to-valley difference rate on future working days. Represents the inverse function of the marginal probability distribution of the daily maximum load on future working days.

[0199] For rest days, the specific process of determining the daily load rate, daily peak-to-valley difference rate and daily maximum load of rest days in the future years is as follows:

[0200] 1) Based on Σ off Construct a multidimensional Gaussian distribution N(0,Σ off ), and randomly sample to generate random samples (y4,y5,y6);

[0201] 2) Based on the random sample (y4, y5, y6), the daily load rate, daily peak-to-valley difference rate and daily maximum load of the rest days in the future year are calculated by the following formula:

[0202]

[0203]

[0204]

[0205] Among them, α off (j) represents the daily load rate of the jth rest day in the future year, β off (j) represents the daily peak-to-valley difference rate of the jth rest day in the future year, γ off (j) represents the maximum daily load on the jth rest day in the future year; represents the inverse function of the marginal probability distribution of the daily load factor on future rest days, The inverse function of the marginal probability distribution of the daily peak-to-valley difference rate of rest days in the future year, Represents the inverse function of the marginal probability distribution of the daily maximum load on future working days.

[0206] The quadratic programming mathematical model in S102 includes an objective function and constraint conditions;

[0207] The objective function aims to minimize the sum of squares of the daily load time series scenarios in the future year and the historical annual daily load benchmark curve, and is constructed as follows:

[0208]

[0209]

[0210] Among them, Δ1 represents the sum of square errors between the daily load time series scenario of a certain day in the future year and the daily load benchmark curve of the working day in the historical year, and Δ2 represents the sum of square errors between the daily load time series scenario of a certain day in the future year and the daily load benchmark curve of the rest day in the historical year; x t (j) represents the load at time t on the jth day of the future year; It represents the daily load reference value at time t on the working day of the historical year. It represents the daily load reference value at rest day t in the historical year. and It is obtained by performing k-means clustering on the load sets of working days and rest days in historical years.

[0211] The constraints include the daily load rate constraint on working days, the daily peak-to-valley difference rate constraint, and the daily maximum load constraint;

[0212] 1) The daily load rate constraint on working days is as follows:

[0213]

[0214] 2) The daily peak-to-valley difference rate constraint on working days is as follows:

[0215]

[0216] Among them, t min-work Indicates the time when the minimum load of the daily load benchmark curve of the historical working day occurs, x tmin-work (j) represents the jth working day t in the future year min-work The load at the time;

[0217] 3) The maximum daily load constraint on working days is as follows:

[0218]

[0219] Among them, t max-work Indicates the time when the maximum load of the daily load reference curve of the working days in the historical year occurs.

[0220] The constraints determined according to the daily load characteristic index of rest days in the future include the daily load rate constraint of rest days, the daily peak-to-valley difference rate constraint and the daily maximum load constraint;

[0221] 1) The daily load rate constraint on rest days is as follows:

[0222]

[0223] 2) The daily peak-to-valley difference rate constraint on rest days is as follows:

[0224]

[0225] Among them, t min-off Indicates the time when the minimum load of the daily load reference curve appears on rest days in historical years. Indicates the jth holiday t in the future year min-off The load at the time;

[0226] 3) The maximum load constraint on rest days is as follows:

[0227]

[0228] Among them, t max-off Indicates the time when the maximum load of the daily load reference curve occurs on rest days in historical years.

[0229] In the above S102, the daily load time series scenario of a certain day in the future year is determined according to the pre-built quadratic programming mathematical model, and the specific process is as follows:

[0230] 1) Use the Cplex mathematical optimization solver to solve the quadratic programming mathematical model and obtain the daily load time series scenario d on the jth day of the future year k (j) = {x t (j), t = 1, 2, ..., 24};

[0231] 2) Assume that the total number of days in the future year is M, execute j=j+1, and determine whether j>M. If j>M, connect the generated M daily load time series scenarios according to the order of arrangement of each day in the future year to obtain the load time series scenario {d k (1),d k (2),...,d k (M)}, and execute k=k+1, and when k is greater than the preset number of future year load time series scenarios N, N different future year daily load time series scenarios are obtained; if j≤M, continue to determine the daily load rate, daily peak-to-valley difference rate and daily maximum load of the remaining working days and rest days in the future year, and solve the secondary optimization model according to the daily load rate, daily peak-to-valley difference rate and daily maximum load of the remaining working days and rest days in the future year, until j>M.

[0232] Based on the same inventive concept, the device for determining the future annual load time series scenario provided by the embodiment of the present invention includes a first determination module, a second determination module and a third determination module. The functions of the above modules are described in detail below:

[0233] The first determination module is used to determine the marginal probability distribution of the daily load characteristic indicators of the future years and the correlation coefficient matrix of the daily load characteristic indicator errors of the historical years according to the daily load characteristic indicators of the historical years, and determine the daily load characteristic indicators of the future years according to the marginal probability distribution of the daily load characteristic indicators of the future years and the correlation coefficient matrix of the daily load characteristic indicator errors of the historical years;

[0234] The second determination module is used to determine the daily load time series scenario of each day in the future year according to the pre-built quadratic programming mathematical model;

[0235] The third determination module is used to determine the future year load time series scenario based on the daily load time series scenario of each day in the future year;

[0236] The above-mentioned quadratic programming mathematical model is constructed based on the daily load characteristic indicators of future years.

[0237] The device for determining the future annual load time series scenario provided by the present invention also includes:

[0238] The error determination module is used to determine the daily load characteristic index error of each day in the historical year according to the following formula:

[0239]

[0240]

[0241]

[0242]

[0243]

[0244]

[0245] in, represents the daily load rate error of the mth working day in the historical year, It represents the daily peak-to-valley rate error of the mth working day in the historical year. Indicates the maximum daily load error of the mth working day in the historical year; M work Indicates the total number of working days in the historical year; γ work (m) represents the maximum daily load on the mth working day of the historical year, and γ work (m) = max{l t (m)},l t (m) represents the load time series at time t on the mth working day of the historical year; α work (m) represents the daily load rate of the mth working day in the historical year, and β work (m) represents the maximum daily load on the mth working day of the historical year, and β work (m) = (γ work (m)-min{l t (m)}) / γ work (m); Indicates the daily load rate error of the nth rest day in the historical year, It represents the daily peak-to-valley rate error of the nth rest day in the historical year. Indicates the daily maximum load error of the nth rest day in the historical year; N off Indicates the total number of rest days in a historical year; γ off (n) represents the maximum daily load of the nth rest day in the historical year, and γ off (n) = max{l t(n)},l t (n) represents the load time series at time t on the nth rest day of the historical year; α off (n) represents the daily load rate of the nth rest day in the historical year, and β off (n) represents the peak-to-valley difference rate of the nth rest day in the historical year, and β off (n) = (γ off (n)-min{l t (n)}) / γ off (n).

[0246] The first determination module mentioned above includes:

[0247] 1) The marginal probability distribution determination unit is used to determine the marginal probability distribution of the daily load characteristic index of each day in the future year according to the following formula:

[0248]

[0249]

[0250]

[0251]

[0252]

[0253]

[0254] Among them, x represents the independent variable, represents the marginal probability distribution of daily load rate on future working days, Represents the marginal probability distribution of the daily peak-to-valley difference rate of future working days, Represents the marginal probability distribution of the daily maximum load on future working days;

[0255] express Pan The probability density function after It represents the base value of daily load factor for future working days. represents the probability density function of the daily load factor error on working days in the historical year, and h represents the kernel function bandwidth; express Pan The probability density function after Indicates the benchmark value of the daily peak-to-valley difference rate for future working days. represents the probability density function of the daily peak-to-valley rate error of the historical working days, and express Pan The probability density function after Indicates the daily maximum load reference value for future working days. represents the probability density function of the daily maximum load error on working days in the historical year, and express Pan The probability density function after Indicates the base value of daily load factor for rest days in the future. represents the probability density function of the daily load factor error on rest days in historical years, and express Pan The probability density function after The benchmark value of the daily peak-to-valley difference rate for rest days in the future years, represents the probability density function of the daily peak-to-valley rate error of the historical rest days, and express Pan The probability density function after Indicates the daily maximum load reference value for the rest days in the future. represents the probability density function of the daily maximum load error on rest days in historical years, and

[0256] 2) Correlation coefficient matrix determination unit, used to determine the correlation coefficient matrix of historical annual and daily load characteristic index errors according to the following formula:

[0257]

[0258] Among them, Σ work The correlation coefficient matrix representing the daily load characteristic index error of the historical working days, Σ off The correlation coefficient matrix representing the error of daily load characteristic index during the historical rest year; ρ1, ρ2, ρ3, ρ4, ρ5, ρ6 all represent correlation coefficients, which are calculated as follows:

[0259]

[0260]

[0261]

[0262] The first determining module also includes:

[0263] The daily load characteristic index determination unit is used to determine the total number of days in the future year, the type and arrangement order of each day in the future year, and to judge the type of the j-th day in the future year. If the j-th day in the future year is a working day, the daily load rate, daily peak-to-valley difference rate and daily maximum load of the working day in the future year are determined; if the j-th day in the future year is a rest day, the daily load rate, daily peak-to-valley difference rate and daily maximum load of the rest day in the future year are determined.

[0264] The daily load characteristic index determination unit includes a daily load characteristic index determination unit for working days and a daily load characteristic index determination unit for rest days.

[0265] 1. The daily load characteristic index determination unit for working days includes:

[0266] 1) The first random sample generation unit is used to generate random samples based on Σ work Construct a multidimensional Gaussian distribution N(0,Σ work ), and randomly sample to generate random samples (y1, y2, y3);

[0267] 2) The daily load characteristic index calculation unit of working days is used to calculate the daily load rate, daily peak-to-valley difference rate and daily maximum load of working days in the future year according to the random sample (y1, y2, y3) through the following formula:

[0268]

[0269]

[0270]

[0271] Among them, α work (j) represents the daily load rate of the jth working day in the future year, β work (j) represents the daily peak-to-valley difference rate of the jth working day in the future year, γ work (j) represents the daily maximum load on the jth working day in the future year; represents the inverse function of the marginal probability distribution of daily load factor on future working days, It represents the inverse function of the marginal probability distribution of the daily peak-to-valley difference rate on future working days. Represents the inverse function of the marginal probability distribution of the daily maximum load on future working days.

[0272] 2. The unit for determining the daily load characteristic index on rest days includes:

[0273] 1) The second random sample generation unit is used to generate random samples based on Σ off Construct a multidimensional Gaussian distribution N(0,Σ off ), and randomly sample to generate random samples (y4,y5,y6);

[0274] 2) The daily load characteristic index calculation unit for rest days is used to calculate the daily load rate, daily peak-to-valley difference rate and daily maximum load of rest days in the future year according to the random sample (y4, y5, y6) through the following formula:

[0275]

[0276]

[0277]

[0278] Among them, α off (j) represents the daily load rate of the jth rest day in the future year, β off (j) represents the daily peak-to-valley difference rate of the jth rest day in the future year, γ off (j) represents the maximum daily load on the jth rest day in the future year; represents the inverse function of the marginal probability distribution of the daily load factor on future rest days, The inverse function of the marginal probability distribution of the daily peak-to-valley difference rate of rest days in the future year, Represents the inverse function of the marginal probability distribution of the daily maximum load on future working days.

[0279] The quadratic programming mathematical model includes an objective function and constraints, and the second determination module is specifically used for:

[0280] The goal is to minimize the sum of squares of the daily load time series scenarios in the future and the daily load benchmark curve in the historical years, and the objective function is constructed as follows:

[0281]

[0282] Among them, Δ1 represents the sum of square errors between the daily load time series scenario of a certain day in the future year and the daily load benchmark curve of the working day in the historical year, and Δ2 represents the sum of square errors between the daily load time series scenario of a certain day in the future year and the daily load benchmark curve of the rest day in the historical year; x t (j) represents the load at time t on the jth day of the future year; It represents the daily load reference value at time t on the working day of the historical year. It represents the daily load reference value at rest day t in the historical year. and It is obtained by performing k-means clustering on the load sets of working days and rest days in historical years.

[0283] 1. Constraints include daily load rate constraints on working days, daily peak-to-valley difference rate constraints, and daily maximum load constraints;

[0284] 1) The daily load rate constraint on working days is as follows:

[0285]

[0286] 2) The daily peak-to-valley difference rate constraint on working days is as follows:

[0287]

[0288] Among them, t min-work Indicates the time when the minimum load of the daily load reference curve of the working day in the historical year occurs. represents the jth working day t in the future year min-work The load at the time;

[0289] 3) The maximum daily load constraint on working days is as follows:

[0290]

[0291] Among them, t max-work Indicates the time when the maximum load of the daily load reference curve of the working days in the historical year occurs.

[0292] 2. The constraints determined based on the daily load characteristic indicators of rest days in the future include the daily load rate constraint of rest days, the daily peak-to-valley difference rate constraint and the daily maximum load constraint;

[0293] 1) The daily load rate constraint on rest days is as follows:

[0294]

[0295] 2) The daily peak-to-valley difference rate constraint on rest days is as follows:

[0296]

[0297] Among them, t min-off Indicates the time when the minimum load of the daily load reference curve appears on rest days in historical years. Indicates the jth holiday t in the future year min-off Load at time;

[0298] 3) The maximum load constraint on rest days is as follows:

[0299]

[0300] Among them, t max-off Indicates the time when the maximum load of the daily load reference curve occurs on rest days in historical years.

[0301] The third determination module mentioned above includes:

[0302] 1) Solving unit, used to solve the quadratic programming mathematical model using Cplex mathematical optimization solver to obtain the daily load time series scenario d on the jth day of the future yeark (j) = {x t (j), t = 1, 2, ..., 24};

[0303] 2) A judgment unit is used to execute j=j+1 and judge whether j>M is satisfied. If j>M, the generated M daily load time series scenarios are connected according to the order of arrangement of each day in the future year to obtain the load time series scenario {d k (1),d k (2),...,d k (M)}, and execute k=k+1, and when k is greater than the preset number of future year load time series scenarios N, N different future year daily load time series scenarios are obtained; if j≤M, continue to determine the daily load rate, daily peak-to-valley difference rate and daily maximum load of the remaining working days and rest days in the future year, and solve the secondary optimization model according to the daily load rate, daily peak-to-valley difference rate and daily maximum load of the remaining working days and rest days in the future year, until j>M, M represents the total number of days in the future year.

[0304] For the convenience of description, the various parts of the above-mentioned device are divided into various modules or units according to their functions and described separately. Of course, when implementing the present application, the functions of each module or unit can be implemented in the same or multiple software or hardware.

[0305] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0306] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0307] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0308] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0309] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Ordinary technicians in the relevant field can still modify or make equivalent substitutions to the specific implementation methods of the present invention with reference to the above embodiments. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention are within the scope of protection of the claims of the present invention to be approved.

Claims

1. A method for determining future annual load time series scenarios, characterized in that: include: Determine the marginal probability distribution of the daily load characteristic indicators of the future years and the correlation coefficient matrix of the daily load characteristic indicator errors of the historical years according to the daily load characteristic indicators of the historical years, and determine the daily load characteristic indicators of the future years according to the marginal probability distribution of the daily load characteristic indicators of the future years and the correlation coefficient matrix of the daily load characteristic indicator errors of the historical years; Determine the daily load time series scenarios for each day in the future year based on the pre-built quadratic programming mathematical model; Determine the load time series scenario for the future year based on the daily load time series scenario for each day of the future year; The quadratic programming mathematical model is constructed based on the daily load characteristic index of each day in the future year; The quadratic programming mathematical model includes an objective function and constraints; The objective function aims to minimize the sum of squares of the daily load time series scenarios in the future and the daily load benchmark curve in the historical years, and is constructed as follows: Among them, Δ1 represents the sum of square errors between the daily load time series scenario of a certain day in the future year and the daily load benchmark curve of the working day in the historical year, and Δ2 represents the sum of square errors between the daily load time series scenario of a certain day in the future year and the daily load benchmark curve of the rest day in the historical year; x t (j) represents the load at time t on the jth day of the future year; It represents the daily load reference value at time t on the working day of the historical year. It represents the daily load reference value at rest day t in the historical year. and The load sets of working days and rest days in the historical year are obtained by performing k-means clustering respectively; The constraints include daily load rate constraints on working days, daily peak-to-valley difference rate constraints, and daily maximum load constraints; The daily load rate constraint of the working day is as follows: The daily peak-to-valley difference rate constraint on the working day is as follows: Among them, t min-work Indicates the time when the minimum load of the daily load reference curve of the working day in the historical year occurs. represents the jth working day t in the future year min-work Load at time; The daily maximum load constraint of the working day is as follows: Among them, t max-work Indicates the time when the maximum load of the daily load reference curve of the working day of the historical year occurs; The constraints determined according to the daily load characteristic index of rest days in the future include the daily load rate constraint of rest days, the daily peak-to-valley difference rate constraint and the daily maximum load constraint; The daily load rate constraint on the rest day is as follows: The daily peak-to-valley difference rate constraint on the rest day is as follows: Among them, t min-off Indicates the time when the minimum load of the daily load reference curve appears on rest days in historical years. Indicates the jth holiday t in the future year min-off Load at time; The maximum load constraint on the rest day is as follows: Among them, t max-off Indicates the time when the maximum load of the daily load reference curve appears on rest days in historical years; The method of determining the daily load time series scenario of a certain day in the future year according to the pre-built quadratic programming mathematical model includes: The Cplex mathematical optimization solver is used to solve the quadratic programming mathematical model and obtain the daily load time series scenario d on the jth day of the future year. k (j) = {x t (j), t = 1, 2, ..., 24}; Assume that the total number of days in the future year is M, execute j=j+1, and determine whether j>M is satisfied. If j>M, connect the generated M daily load time series scenarios according to the order of arrangement of each day in the future year to obtain the load time series scenario {d k (1),d k (2),...,d k (M)}, and execute k=k+1, and when k is greater than the preset number of future year load time series scenarios N, N different future year daily load time series scenarios are obtained; if j≤M, continue to determine the daily load rate, daily peak-to-valley difference rate and daily maximum load of the remaining working days and rest days in the future year, and solve the secondary optimization model according to the daily load rate, daily peak-to-valley difference rate and daily maximum load of the remaining working days and rest days in the future year, until j>M.

2. The method for determining the future annual load time series scenario according to claim 1, characterized in that: Before determining the marginal probability distribution of the daily load characteristic index errors of future days and the correlation coefficient matrix of the daily load characteristic index errors of historical years based on the daily load characteristic indexes of historical years, the following steps are included: Determine the daily load characteristic index error of each day in the historical year according to the following formula: in, represents the daily load rate error of the mth working day in the historical year, It represents the daily peak-to-valley rate error of the mth working day in the historical year. Indicates the maximum daily load error of the mth working day in the historical year; M work Indicates the total number of working days in the historical year; γ work (m) represents the maximum daily load on the mth working day of the historical year, and γ work (m) = max{l t (m)},l t (m) represents the load time series at time t on the mth working day of the historical year; α work (m) represents the daily load rate of the mth working day in the historical year, and β work (m) represents the daily peak-to-valley difference rate of the mth working day in the historical year, and β work (m) = (γ work (m)-min{l t (m)})γ work (m); Indicates the daily load rate error of the nth rest day in the historical year, It represents the daily peak-to-valley rate error of the nth rest day in the historical year. Indicates the daily maximum load error of the nth rest day in the historical year; N off Indicates the total number of rest days in a historical year; γ off (n) represents the maximum daily load of the nth rest day in the historical year, and γ off (n) = max{l t (n)},l t (n) represents the load time series at time t on the nth rest day of the historical year; α off (n) represents the daily load rate of the nth rest day in the historical year, and β off (n) represents the peak-to-valley difference rate of the nth rest day in the historical year, and β off (n) = (γ off (n)-min{l t (n)})γ off (n).

3. The method for determining the future annual load time series scenario according to claim 2, characterized in that: The marginal probability distribution of the daily load characteristic index of the future year is as follows: Among them, x represents the independent variable, represents the marginal probability distribution of daily load rate on future working days, Represents the marginal probability distribution of the daily peak-to-valley difference rate of future working days, Represents the marginal probability distribution of the daily maximum load on future working days; express Pan The probability density function after It represents the base value of daily load factor for future working days. represents the probability density function of the daily load factor error on working days in the historical year, and h represents the kernel function bandwidth; express Pan The probability density function after Indicates the benchmark value of the daily peak-to-valley difference rate for future working days. represents the probability density function of the daily peak-to-valley rate error of the historical working days, and express Pan The probability density function after Indicates the daily maximum load reference value for future working days. represents the probability density function of the daily maximum load error on working days in the historical year, and express Pan The probability density function after Indicates the base value of daily load factor for rest days in the future. represents the probability density function of the daily load factor error on rest days in historical years, and express Pan The probability density function after The benchmark value of the daily peak-to-valley difference rate for rest days in the future years, represents the probability density function of the daily peak-to-valley rate error of the historical rest days, and express Pan The probability density function after Indicates the daily maximum load reference value for the rest days in the future. represents the probability density function of the daily maximum load error on rest days in historical years, and 4. The method for determining the future annual load time series scenario according to claim 3 is characterized in that: The correlation coefficient matrix of the historical annual daily load characteristic index error is as follows: Among them, Σ work The correlation coefficient matrix representing the daily load characteristic index error of the historical working days, Σ off The correlation coefficient matrix representing the daily load characteristic index errors of historical rest days; ρ1, ρ2, ρ3, ρ4, ρ5, ρ6 all represent correlation coefficients, which are calculated as follows:

5. The method for determining the future annual load time series scenario according to claim 4, characterized in that: The method of determining the daily load characteristic index for each day in the future year according to the marginal probability distribution of the daily load characteristic index for each day in the future year and the correlation coefficient matrix of the daily load characteristic index errors in the historical years includes: Determine the total number of days in the future year, the type and arrangement order of each day in the future year, judge the type of the j-th day in the future year, if the j-th day in the future year is a working day, determine the daily load rate, daily peak-to-valley difference rate and daily maximum load of the working day in the future year; if the j-th day in the future year is a rest day, determine the daily load rate, daily peak-to-valley difference rate and daily maximum load of the rest day in the future year.

6. The method for determining the future annual load time series scenario according to claim 5, characterized in that: The daily load rate, daily peak-to-valley difference rate and daily maximum load of the working days in the future year include: Based on Σ work Construct a multidimensional Gaussian distribution N(0,Σ work ), and randomly sample to generate random samples (y1, y2, y3); Based on the random sample (y1, y2, y3), the daily load rate, daily peak-to-valley difference rate and daily maximum load of the working days in the future year are calculated by the following formula: Among them, α work (j) represents the daily load rate of the jth working day in the future year, β work (j) represents the daily peak-to-valley difference rate of the jth working day in the future year, γ work (j) represents the daily maximum load on the jth working day in the future year; represents the inverse function of the marginal probability distribution of daily load factor on future working days, It represents the inverse function of the marginal probability distribution of the daily peak-to-valley difference rate on future working days. Represents the inverse function of the marginal probability distribution of the daily maximum load on future working days.

7. The method for determining the future annual load time series scenario according to claim 6, characterized in that: The determination of the daily load rate, daily peak-to-valley difference rate and daily maximum load for the future annual rest days includes: Based on Σ off Construct a multidimensional Gaussian distribution N(0,Σ off ), and randomly sample to generate random samples (y4,y5,y6); Based on the random sample (y4, y5, y6), the daily load rate, daily peak-to-valley difference rate and daily maximum load of rest days in the future year are calculated by the following formula: Among them, α off (j) represents the daily load rate of the jth rest day in the future year, β off (j) represents the daily peak-to-valley difference rate of the jth rest day in the future year, γ off (j) represents the maximum daily load on the jth rest day in the future year; represents the inverse function of the marginal probability distribution of the daily load factor on future rest days, The inverse function of the marginal probability distribution of the daily peak-to-valley difference rate of rest days in the future year, Represents the inverse function of the marginal probability distribution of the daily maximum load on future working days.

8. A device for determining future annual load time series scenarios, characterized in that: include: The first determination module is used to determine the marginal probability distribution of the daily load characteristic indicators of the future years and the correlation coefficient matrix of the daily load characteristic indicator errors of the historical years according to the daily load characteristic indicators of the historical years, and determine the daily load characteristic indicators of the future years according to the marginal probability distribution of the daily load characteristic indicators of the future years and the correlation coefficient matrix of the daily load characteristic indicator errors of the historical years; The second determination module is used to determine the daily load time series scenario of each day in the future year according to the pre-built quadratic programming mathematical model; The third determination module is used to determine the future year load time series scenario based on the daily load time series scenario of each day in the future year; The quadratic programming mathematical model is constructed based on the daily load characteristic index of each day in the future year; The quadratic programming mathematical model includes an objective function and constraints; The second determining module is specifically used for: The goal is to minimize the sum of squares of the daily load time series scenarios in the future and the daily load benchmark curve in the historical years, and construct the objective function as follows: Among them, Δ1 represents the sum of square errors between the daily load time series scenario of a certain day in the future year and the daily load benchmark curve of the working day in the historical year, and Δ2 represents the sum of square errors between the daily load time series scenario of a certain day in the future year and the daily load benchmark curve of the rest day in the historical year; x t (j) represents the load at time t on the jth day of the future year; It represents the daily load reference value at time t on the working day of the historical year. It represents the daily load reference value at rest day t in the historical year. and The load sets of working days and rest days in the historical year are obtained by performing k-means clustering respectively; The constraints include daily load rate constraints on working days, daily peak-to-valley difference rate constraints, and daily maximum load constraints; The daily load rate constraint of the working day is as follows: The daily peak-to-valley difference rate constraint on the working day is as follows: Among them, t min-work Indicates the time when the minimum load of the daily load reference curve of the working day in the historical year occurs. represents the jth working day t in the future year min-work Load at time; The daily maximum load constraint of the working day is as follows: Among them, t max-work Indicates the time when the maximum load of the daily load reference curve of the working day of the historical year occurs; The constraints determined according to the daily load characteristic index of rest days in the future include the daily load rate constraint of rest days, the daily peak-to-valley difference rate constraint and the daily maximum load constraint; The daily load rate constraint on the rest day is as follows: The daily peak-to-valley difference rate constraint on the rest day is as follows: Among them, t min-off Indicates the time when the minimum load of the daily load reference curve appears on rest days in historical years. Indicates the jth holiday t in the future year min-off Load at time; The maximum load constraint on the rest day is as follows: Among them, t max-off Indicates the time when the maximum load of the daily load reference curve appears on rest days in historical years; The third determination module comprises: The solving unit is used to solve the quadratic programming mathematical model using the Cplex mathematical optimization solver to obtain the daily load time series scenario d on the jth day of the future year. k (j) = {x t (j), t = 1, 2, ..., 24}; The judgment unit is used to execute j=j+1 and judge whether j>M is satisfied. If j>M, the generated M daily load time series scenarios are connected according to the order of arrangement of each day in the future year to obtain the load time series scenario {d k (1),d k (2),...,d k (M)}, and execute k=k+1, and when k is greater than the preset number of future year load time series scenarios N, N different future year daily load time series scenarios are obtained; if j≤M, continue to determine the daily load rate, daily peak-to-valley difference rate and daily maximum load of the remaining working days and rest days in the future year, and solve the secondary optimization model according to the daily load rate, daily peak-to-valley difference rate and daily maximum load of the remaining working days and rest days in the future year, until j>M, M represents the total number of days in the future year.

9. The device for determining the future annual load time series scenario according to claim 8, characterized in that: The device also includes: The error determination module is used to determine the daily load characteristic index error of each day in the historical year according to the following formula: in, represents the daily load rate error of the mth working day in the historical year, It represents the daily peak-to-valley rate error of the mth working day in the historical year. Indicates the maximum daily load error of the mth working day in the historical year; M work Indicates the total number of working days in the historical year; γ work (m) represents the maximum daily load on the mth working day of the historical year, and γ work (m) = max{l t (m)},l t (m) represents the load time series at time t on the mth working day of the historical year; α work (m) represents the daily load rate of the mth working day in the historical year, and β work (m) represents the daily peak-to-valley difference rate of the mth working day in the historical year, and β work (m) = (γ work (m)-min{l t (m)})γ work (m); Indicates the daily load rate error of the nth rest day in the historical year, It represents the daily peak-to-valley rate error of the nth rest day in the historical year. Indicates the daily maximum load error of the nth rest day in the historical year; N off Indicates the total number of rest days in a historical year; γ off (n) represents the maximum daily load of the nth rest day in the historical year, and γ off (n) = max{l t (n)},l t (n) represents the load time series at time t on the nth rest day of the historical year; α off (n) represents the daily load rate of the nth rest day in the historical year, and β off (n) represents the peak-to-valley difference rate of the nth rest day in the historical year, and β off (n) = (γ off (n)-min{l t (n)})γ off (n).

10. The device for determining the future annual load time series scenario according to claim 8, characterized in that: The first determining module comprises: The marginal probability distribution determination unit is used to determine the marginal probability distribution of the daily load characteristic index of each day in the future year according to the following formula: Among them, x represents the independent variable, represents the marginal probability distribution of daily load rate on future working days, Represents the marginal probability distribution of the daily peak-to-valley difference rate of future working days, Represents the marginal probability distribution of the daily maximum load on future working days; express Pan The probability density function after It represents the base value of daily load factor for future working days. represents the probability density function of the daily load factor error on working days in the historical year, and h represents the kernel function bandwidth; express Pan The probability density function after Indicates the benchmark value of the daily peak-to-valley difference rate for future working days. represents the probability density function of the daily peak-to-valley rate error of the historical working days, and express Pan The probability density function after Indicates the daily maximum load reference value for future working days. represents the probability density function of the daily maximum load error on working days in the historical year, and express Pan The probability density function after Indicates the base value of daily load factor for rest days in the future. represents the probability density function of the daily load factor error on rest days in historical years, and express Pan The probability density function after The benchmark value of the daily peak-to-valley difference rate for rest days in the future years, represents the probability density function of the daily peak-to-valley rate error of the historical rest days, and express Pan The probability density function after Indicates the daily maximum load reference value for the rest days in the future. represents the probability density function of the daily maximum load error on rest days in historical years, and 11. The device for determining the future annual load time series scenario according to claim 10, characterized in that: The first determining module also includes: The correlation coefficient matrix determination unit is used to determine the correlation coefficient matrix of the historical annual and daily load characteristic index errors according to the following formula: Among them, Σ work The correlation coefficient matrix representing the daily load characteristic index error of the historical working days, Σ off The correlation coefficient matrix representing the daily load characteristic index errors of historical rest days; ρ1, ρ2, ρ3, ρ4, ρ5, ρ6 all represent correlation coefficients, which are calculated as follows:

12. The device for determining future annual load time series scenarios according to claim 11, characterized in that: The first determining module also includes: The daily load characteristic index determination unit is used to determine the total number of days in the future year, the type and arrangement order of each day in the future year, and to judge the type of the j-th day in the future year. If the j-th day in the future year is a working day, the daily load rate, daily peak-to-valley difference rate and daily maximum load of the working day in the future year are determined; if the j-th day in the future year is a rest day, the daily load rate, daily peak-to-valley difference rate and daily maximum load of the rest day in the future year are determined.

13. The device for determining the future annual load time series scenario according to claim 12, characterized in that: The daily load characteristic index determination unit includes a daily load characteristic index determination unit for working days and a daily load characteristic index determination unit for rest days.

14. The device for determining the future annual load time series scenario according to claim 13, characterized in that: The working day daily load characteristic index determination unit includes: The first random sample generation unit is used to generate a random sample based on Σ work Construct a multidimensional Gaussian distribution N(0,Σ work ), and randomly sample to generate random samples (y1, y2, y3); The daily load characteristic index calculation unit of working days is used to calculate the daily load rate, daily peak-to-valley difference rate and daily maximum load of working days in the future year according to the random sample (y1, y2, y3) through the following formula: Among them, α work (j) represents the daily load rate of the jth working day in the future year, β work (j) represents the daily peak-to-valley difference rate of the jth working day in the future year, γ work (j) represents the daily maximum load on the jth working day in the future year; represents the inverse function of the marginal probability distribution of daily load factor on future working days, It represents the inverse function of the marginal probability distribution of the daily peak-to-valley difference rate on future working days. Represents the inverse function of the marginal probability distribution of the daily maximum load on future working days.

15. The device for determining the future annual load time series scenario according to claim 13, characterized in that: The rest day daily load characteristic index determination unit comprises: The second random sample generation unit is used to generate a random sample based on Σ off Construct a multidimensional Gaussian distribution N(0,Σ off ), and randomly sample to generate random samples (y4,y5,y6); The daily load characteristic index calculation unit for rest days is used to calculate the daily load rate, daily peak-to-valley difference rate and daily maximum load of rest days in the future year according to the random sample (y4, y5, y6) through the following formula: Among them, α off (j) represents the daily load rate of the jth rest day in the future year, β off (j) represents the daily peak-to-valley difference rate of the jth rest day in the future year, γ off (j) represents the maximum daily load on the jth rest day in the future year; represents the inverse function of the marginal probability distribution of the daily load factor on future rest days, The inverse function of the marginal probability distribution of the daily peak-to-valley difference rate of rest days in the future year, Represents the inverse function of the marginal probability distribution of the daily maximum load on future working days.