A method for generating period-by-period power load curve integrating current and future annual load characteristics

Abstract

The present invention provides a method for generating a time-period electric load curve that integrates the current and future annual load characteristics, including: obtaining the monthly load characteristic sequence and quarterly load characteristic sequence of the future design year of the study area, and performing correlation analysis with the monthly load characteristic sequence and quarterly load characteristic sequence of the working condition year to generate a reference load factor sequence of the reference working condition year; based on the characteristics of the reference load factor sequence of the reference working condition year, obtaining a time-period load factor curve that integrates the current and future design annual load characteristics. The present invention uses limited current annual power grid operation data and future design year power load operation characteristic data to predict the time-period electric load curve in the study area. This method can use less data to quickly predict the electric load curve process of the study area, and the prediction result is accurate and reliable, providing data support for power system planning and optimal scheduling.

Description

Technical Field

[0001] The present invention belongs to the technical field of power load curve generation, and in particular relates to a method for generating a time period power load curve integrating current and future annual load characteristics. Background Art

[0002] The power load curve refers to the curve of various types of power loads in the power system changing over time. The power load curve is an important basis for dispatching power in the power system and planning the power system. The generation of the power load curve is the basis for arranging daily dispatch plans, determining start-up and shutdown plans, economically distributing loads, and arranging rotating reserve capacity.

[0003] The current power load curve generation technology can be summarized into the following four categories:

[0004] (1) Statistical methods based on historical data: This type of method usually uses historical load data to predict future power load curves through statistical methods such as regression analysis and time series analysis. Since it assumes that load changes follow a certain predictable pattern, the prediction accuracy is affected when dealing with complex and changing power grid environments.

[0005] (2) Machine learning algorithms: Machine learning algorithms such as support vector machines (SVM), random forests, and neural networks are widely used in load forecasting. These methods can handle nonlinear relationships and adapt to changes in load data to a certain extent. By training the model, the inherent laws and characteristics of the load data can be learned, thereby making predictions.

[0006] (3) Deep learning methods: With the rise of deep learning technology, load forecasting methods based on deep learning have gradually attracted attention. For example, deep learning models such as recurrent neural networks (RNNs), long short-term memory networks (LSTMs), and convolutional neural networks (CNNs) are able to handle long-term dependencies in time series data.

[0007] (4) Ensemble learning method: Ensemble learning method combines multiple prediction models to improve the overall prediction accuracy by combining the prediction results. For example, you can use an ensemble learning framework (such as Bagging, Boosting, etc.) to integrate multiple machine learning or deep learning models to obtain more stable prediction results.

[0008] The above methods have the following problems: Method (1) mainly uses the regularity of statistical analysis of historical data, and does not fully consider the load characteristics of the study area in future years. The power load forecasting results are a mechanical repetition of historical data based on a certain regularity, and the reliability of the power load forecasting results is poor; Methods (2), (3), and (4) are mainly based on machine learning, deep learning, ensemble learning and other methods, which require a large amount of long series, multi-scale, multi-type, and multi-source data. The power load forecasting results are limited by the scale of training data, and the reliability and stability of the results are difficult to verify. They are not very applicable in some study areas where conditions and data are lacking. Summary of the invention

[0009] In view of the defects of the prior art, the present invention provides a method for generating a time period power load curve that integrates the current and future annual load characteristics, which can effectively solve the above problems.

[0010] The technical solution adopted by the present invention is as follows:

[0011] The present invention provides a method for generating a time-period power load curve integrating current and future annual load characteristics, comprising the following steps:

[0012] Step S1, obtain the power load planning parameters of the future design year x in the study area, including: annual utilization hours t x , monthly load characteristic sequence of design year x {M x,1 ,M x,2 ,...,M x,12} and seasonal load characteristic series {S x,1 ,S x,2 ,S x,3 ,S x,4};

[0013] Among them, the monthly load characteristic M x,i , i=1,2,...,12, meaning: the maximum value of the load factor p(t) in each time period t in the i-th month of the design year x; among which, the time period in which the maximum value of the load factor p(t) occurs is an unknown quantity;

[0014] Seasonal load characteristics S x,j , j = 1, 2, 3, 4, meaning: load factor fluctuation sequence of unknown typical day k in the jth quarter of design year x, that is: S x,j ={p x,j,k (t)}={p x,j,k (1),p x,j,k (2),...,p x,j,k (24)}, where the unknown typical day k is the day to be determined, which is the day when the maximum value of the load factor p(t) in the jth quarter of the design year x occurs;

[0015] Step S2, select Y working conditions in the study area, and obtain the fluctuation sequence of the load factor in each working condition year y in each time period t, y = 1, 2, ..., Y, expressed as {p y (1),p y (2),...,p y (N)}, N represents the number of time periods t in the operating year y;

[0016] Step S3: for each load factor fluctuation sequence {p y (1),p y (2),...,p y (N)} to obtain the monthly load characteristics M of each working condition year y in the i-th month y,i , and the quarterly load characteristics S of each operating year y in the jth quarter y,j ;

[0017] Step S4, compare the monthly load characteristics M of various working conditions in year y y,i and the monthly load characteristic M of the design year x x,i The correlation between the seasonal load characteristics S of various working conditions in year y is compared. y,j and the quarterly load characteristics S of the design year x x,j The correlation is used to generate the reference load factor sequence {p r (1),p r (2),...,p r (N)}; wherein the reference operating year r has the same number of time periods t as the operating year y;

[0018] Step S5, taking the month as the analysis unit, the benchmark load factor sequence {p r (1),p r (2),...,p r (N)} to determine the week in which the load factor p(t) is the maximum in each i-th month of the reference operating year r, which is expressed as week W. Since the reference operating year r has 12 months, it is located at 12 weeks;

[0019] According to the weekly positioning results of the reference operating condition year r, the monthly load characteristic sequence {M x,1 ,M x,2 ,...,M x,12} is converted into a fine-grained weekly load characteristic sequence by:

[0020] The monthly load characteristic M of the i-th month of design year x is x,i, assigned to the corresponding week W located as the weekly load characteristics of the corresponding week W, thereby obtaining 12 weekly load characteristics in the design year x;

[0021] The 12 weekly load characteristics are interpolated to obtain the weekly load characteristic sequence of all weeks in the design year x, which is expressed in order as F x,1 ,F x,2 ,...,F x,U ; Where U represents the number of weeks in a year;

[0022] Step S6, taking quarter as the analysis unit, the benchmark load factor sequence {p r (1),p r (2),...,p r (N)} to determine the week with the maximum load factor p(t) in each jth quarter of the reference operating year r. Since the reference operating year r has 4 quarters, it is located at 4 weeks;

[0023] According to the weekly positioning results of the reference operating condition year r, the quarterly load characteristics S of the design year x are x,j ={p x,j,k (t)}={p x,j,k (1),p x,j,k (2),...,p x,j,k (24)} is converted into a fine-grained load factor fluctuation sequence of unknown typical days in a week as the weekly load feature, and the method is:

[0024] The quarterly load characteristic S of the jth quarter of design year x is x,j ={p x,j,k (t)}={p x,j,k (1),p x,j,k (2),...,p x,j,k (24)}, assigned to the corresponding week located as the weekly load characteristics of the corresponding week, thereby obtaining four weekly load characteristics in the design year x;

[0025] The four weekly load characteristics are interpolated to obtain the weekly load characteristics of the design year x in all weeks, which are expressed in order as FC x,1 ,FC x,2 ,...,FC x,U ;

[0026] Step S7, further analyzing the weekly load characteristics F of all weeks in the design year x x,1 ,F x,2 ,...,F x,U and weekly load characteristics FC x,1 ,FC x,2 ,...,FC x,U, and the correlation between the weekly load characteristics and weekly load characteristics of the benchmark operating year r in all weeks, and obtain the power load factor p(t) curve for each period t that integrates the current and future design annual load characteristics.

[0027] Preferably, in step S1, the annual utilization hours t x The method to obtain is:

[0028] t x =Q x / P x *10 5

[0029] in:

[0030] Q x is the annual power demand in design year x;

[0031] P x It is the maximum load of the whole society in design year x.

[0032] Preferably, step S4 is specifically:

[0033] Step S4.1, calculate the monthly load characteristic sequence {M x,1 ,M x,2 ,...,M x,12} and the monthly load characteristic sequence of working year y {M y,1 ,M y,2 ,...,M y,12 The correlation coefficient R between x,y , relative error RE x,y and efficiency coefficient NASH x,y ;

[0034] Calculate the quarterly load characteristics S in the jth quarter of the design year x x,j The quarterly load characteristic S of the operating year y in the jth quarter y,j The correlation coefficient between Relative error and efficiency coefficient Where, j = 1, 2, 3, 4;

[0035] Therefore, the correlation coefficient matrix RM between the design year x and the operating year y is obtained x,y :

[0036]

[0037] Step S4.2: Since there are Y types of working condition years, a total of Y correlation coefficient matrices are obtained, which are: Correlation coefficient matrix RM x,1 ,RM x,2 ,...,RMx,Y ;

[0038] For the correlation coefficient matrix RM x,1 ,RM x,2 ,...,RM x,Y The indicators in are ranked by entropy weight method to obtain the weight factor α for each working condition year y. y ;

[0039] Step S4.3: The fluctuation sequence {p y (1),p y (2),...,p y (N)}, according to the weight factor α y By superimposing each time period t, the benchmark load factor sequence {p r (1),p r (2),...,p r (N)}, the specific formula is:

[0040]

[0041] The base load factor sequence {p r (1),p r (2),...,p r (N)}.

[0042] Preferably, step S7 specifically includes:

[0043] Step S7.1, integrating the weekly load characteristics to the benchmark load factor sequence {p r (1),p r (2),...,p r (N)} to make corrections:

[0044] Step S7.1.1, in units of weeks, calculate the base load factor sequence {p r (1),p r (2),...,p r (N)}, the weekly load characteristic sequence of the benchmark operating condition year r is obtained, which is expressed as: F r,1 ,F r,2 ,...,F r,U ;

[0045] Step S7.1.2: The weekly load characteristic sequence F of design year x is x,1 ,F x,2 ,...,F x,U The weekly load characteristic sequence F of the reference operating condition year r r,1 ,Fr,2 ,...,F r,U By comparison, the weekly load characteristic expansion coefficient β of each week u in the year is obtained according to the following formula u , where u=1,2,...,U:

[0046] β u =F x,u / F r,u

[0047] Step S7.1.3, for the base load factor sequence {p r (1),p r (2),...,p r Each benchmark load factor in (N)} is multiplied by the weekly load characteristic expansion coefficient of the week in which it is located, thereby obtaining the revised benchmark load factor sequence of the benchmark operating condition year r, which is expressed as:

[0048] Step S7.2, integrating the load factor fluctuation sequence of typical days in the week, and correcting the benchmark load factor sequence of the benchmark working condition year r To make corrections:

[0049] Step S7.2.1, in units of weeks, the reference load factor series of the revised reference operating condition year r After analysis, the weekly load characteristic sequence is obtained, which is expressed as: FC r,1 ,FC r,2 ,...,FC r,U ;

[0050] Step S7.2.2: The weekly load characteristic sequence FC of design year x is x,1 ,FC x,2 ,...,FC x,U The weekly load characteristic sequence FC of the benchmark operating condition year r r,1 ,FC r,2 ,...,FC r,U By comparison, the weekly load characteristic expansion coefficient γ of each week u in the year is obtained as follows: u , where u=1,2,...,U:

[0051] γ u =FC x,u / FC r,u

[0052] Step S7.2.3: The reference load factor sequence of the revised reference operating condition year r Each benchmark load factor in is multiplied by the weekly load characteristic expansion coefficient of the week in which it is located, thereby obtaining the revised benchmark load factor sequence of the benchmark operating condition year r, which is expressed as:

[0053] Step S7.3: The modified base load factor sequence of base operating condition year r Further correction is made to obtain the corrected benchmark load factor sequence of benchmark operating year r

[0054] Step S7.3.1: The base load factor sequence of the revised base operating year r Add up and get the corrected annual utilization hours t r , according to the following formula, we can get the annual expansion coefficient δ:

[0055] δ=t x / t r

[0056] Step S7.3.2, determine whether the deviation between the annual expansion coefficient δ and 1 is less than the set threshold, if yes, execute step S7.3.5; if no, execute step S7.3.3;

[0057] Step S7.3.3, the base load factor sequence for the revised base operating year r In the example, the weekly unit is divided into U sequences, which are expressed as: {C(1), C(2), ..., C(U)}; for the sequence C(u), u = 1, 2, ..., U, which represents the benchmark load factor sequence corresponding to the uth week;

[0058] Since there are Z time periods per week, the sequence C(u) has Z base load factors, expressed as:

[0059] For the sequence C(u), the day is taken as the unit. Since a week has 7 days, the sequence C(u) is further divided into 7 subsequences, which are represented as: {D(1), D(2), ..., D(7)}; in the subsequences D(1), D(2), ..., D(7), the day where the maximum value of the modified benchmark load factor is located is identified, and is set as day d, d = 1, 2, ..., 7, then day d is the typical day of the uth week, and the modified benchmark load factor sequence in the subsequence D(d) corresponding to the typical day of the uth week is not modified and remains unchanged;

[0060] The subsequences corresponding to the other six atypical days in week u are expressed as: D(z), z = 1, 2, ..., 7, and z ≠ d; since there are 24 time periods per day, there are 24 modified baseline load factor sequences in the atypical day sequence D(z), which are re-expressed as: The maximum load factor and the minimum load factor in are expressed as: and

[0061] For each modified base load factor in the atypical day sequence D(z) Where I = 1, 2, ..., 24, the following formula is used for correction to obtain the corrected reference load factor

[0062]

[0063] Step S7.3.4, therefore, adopt the method of step S7.3.3 to modify the benchmark load factor sequence of the revised benchmark operating year r After each item in is corrected, the corrected benchmark load factor sequence of benchmark operating year r is obtained.

[0064] use replace Return to step S7.3.1 and perform the cyclic correction process;

[0065] Step S7.3.5, output the final corrected benchmark load factor sequence for benchmark operating year r Thus, a time-period power load factor curve is obtained that integrates the current and future design annual load characteristics.

[0066] The method for generating a time-period power load curve integrating current and future annual load characteristics provided by the present invention has the following advantages:

[0067] In the case that it is difficult to obtain sufficient long series, multi-scale, multi-type and multi-source data in the study area, the present invention proposes a method for generating a time-period power load curve that integrates the current and future annual load characteristics. The method uses limited current annual power grid operation data and future design year power load operation characteristic data to predict the time-period power load curve in the study area. This method can use less data to quickly predict the power load curve process of the study area. The prediction result is accurate and reliable, providing data support for power system planning and optimal scheduling. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 A schematic flow chart of a method for generating a time-period power load curve that integrates current and future annual load characteristics provided by the present invention. DETAILED DESCRIPTION

[0069] In order to make the technical problems, technical solutions and beneficial effects solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0070] In view of the problem that the reliability of power load forecast results in the future design year is poor due to insufficient consideration of power grid operation characteristics or insufficient data, the present invention provides a method for generating a time period power load curve that integrates the current and future annual load characteristics, which has important practical significance for dispatching power system electricity and conducting power system planning.

[0071] See also Figure 1 The present invention provides a method for generating a time-period power load curve integrating current and future annual load characteristics, comprising the following steps:

[0072] Step S1, obtain the power load planning parameters of the future design year x in the study area, including: annual utilization hours t x , monthly load characteristic sequence of design year x {M x,1 ,M x,2 ,...,M x,12} and seasonal load characteristic series {S x,1 ,S x,2 ,S x,3 ,S x,4};

[0073] Among them, the monthly load characteristic M x,i , i=1,2,...,12, which means: the maximum value of the load factor p(t) in each time period t in the i-th month of the design year x; wherein, the time period in which the maximum value of the load factor p(t) occurs is an unknown quantity; in the present invention, the time period t means hours, and each hour forms a time period.

[0074] Seasonal load characteristics S x,j , j = 1, 2, 3, 4, meaning: load factor fluctuation sequence of unknown typical day k in the jth quarter of design year x, that is: S x,j ={p x,j,k (t)}={p x,j,k (1),p x,j,k (2),...,p x,j,k (24)}, where the unknown typical day k is the day to be determined, which is the day when the maximum value of the load factor p(t) in the jth quarter of the design year x occurs;

[0075] Annual utilization hours t x The method to obtain is: x =Q x / P x *10 5 ; Among them: Q x P is the annual power demand in design year x, in 100 million kWh; x It is the maximum load of the whole society in design year x, in MW.

[0076] For example, as an example, when planning the power load for a certain study area in 2030, the maximum load P of the whole society is planned. x The total annual electricity demand of the whole society is Q x 473 billion kWh, then the annual utilization hours are t x =Q x / P x *10 5 =473 billion kWh / 65,000MW*10 5 =7280h.

[0077] The monthly load characteristic sequence of design year x {M x,1 ,M x,2 ,...,M x,12}, the maximum value of the load factor of each month forms the monthly load characteristics, and the specific values ​​are shown in Table 1:

[0078] Table 1 Monthly load characteristics of a region in 2030 (i.e. the maximum monthly load factor)

[0079] month January February March April May June coefficient 0.880 0.840 0.860 0.900 0.900 0.890 month July August September October November December coefficient 0.960 0.970 0.960 0.990 0.990 1.000

[0080] Therefore, in the present invention, the load factor takes a value of 0 to 1, which is a normalized value of the maximum load of each month.

[0081] The quarterly load characteristic sequence of design year x {S x,1 ,S x,2 ,S x,3 ,S x,4}, the quarterly load characteristics of each quarter are the load factor fluctuation sequence of the unknown typical day in each quarter. Since the unknown typical day has 24 time periods, its load factor fluctuation sequence has 24 load factors. The specific values ​​are shown in Table 2:

[0082] Table 2 Seasonal load characteristics of unknown typical days in each season in a certain region in 2030

[0083]

[0084]

[0085] Step S2, select Y working conditions in the study area, and obtain the fluctuation sequence of the load factor in each working condition year y in each time period t, y = 1, 2, ..., Y, expressed as {p y (1),p y (2),...,p y (N)}, N represents the number of time periods t in the operating year y;

[0086] In the present invention, the operating year is a historical year. For example, there are 8 operating years, namely 2015, 2016, 2017, 2018, 2019, 2020, 2021, and 2022. Therefore, each operating year can obtain the load factor of each period in the year. If each year is calculated as 365 days, then each operating year can obtain 365*24=8760 load factors, that is, Y is equal to 8760, forming a fluctuation sequence of the load factor in each period t.

[0087] Step S3: for each load factor fluctuation sequence {p y (1),p y (2),...,p y (N)} to obtain the monthly load characteristics M of each working condition year y in the i-th month y,i , and the quarterly load characteristics S of each operating year y in the jth quarter y,j ;

[0088] Since each operating year y has detailed load factors for each period t, its monthly load characteristics and quarterly load characteristics can be analyzed.

[0089] As an example, the monthly load characteristics from 2015 to 2022 are shown in Table 3:

[0090] Table 3 Monthly load characteristics of a region from 2015 to 2022

[0091] 2015 2016 2017 2018 2019 2020 2021 2022 January 0.939 0.952 0.839 0.876 0.871 0.917 1.000 0.859 February 0.898 0.920 0.811 0.855 0.846 0.818 0.944 0.878 March 0.856 0.910 0.828 0.855 0.847 0.848 0.898 0.842 April 0.891 0.926 0.844 0.835 0.841 0.853 0.899 0.807 May 0.894 0.983 0.837 0.845 0.831 0.868 0.911 0.861 June 0.906 0.969 0.874 0.851 0.878 0.908 0.911 0.928 July 0.937 0.992 0.948 0.890 0.887 0.910 0.926 0.888 August 0.959 0.941 0.950 0.923 0.902 0.900 0.907 0.917 September 0.900 0.939 0.914 0.910 0.882 0.895 0.863 0.898 October 1.000 0.965 0.952 0.956 0.910 0.924 0.992 0.936 November 0.931 0.996 0.979 0.974 0.944 0.957 0.914 0.961 December 0.934 1.000 1.000 1.000 1.000 1.000 0.932 1.000

[0092] There are many data items in the quarterly load characteristics from 2015 to 2022, so we will not give detailed examples here.

[0093] Step S4, compare the monthly load characteristics M of various working conditions in year y y,i and the monthly load characteristic M of the design year x x,i The correlation between the seasonal load characteristics S of various working conditions in year y is compared. y,j and the quarterly load characteristics S of the design year x x,j The correlation is used to generate the reference load factor sequence {p r (1),p r (2),...,p r (N)}; wherein the reference operating year r has the same number of time periods t as the operating year y;

[0094] Step S4 is specifically as follows:

[0095] Step S4.1, calculate the monthly load characteristic sequence {Mx,1 ,M x,2 ,...,M x,12} and the monthly load characteristic sequence of working year y {M y,1 ,M y,2 ,...,M y,12 The correlation coefficient R between x,y , relative error RE x,y and efficiency coefficient NASH x,y ;

[0096] Calculate the quarterly load characteristics S in the jth quarter of the design year x x,j The quarterly load characteristic S of the operating year y in the jth quarter y,j The correlation coefficient between Relative error and efficiency coefficient Where, j = 1, 2, 3, 4;

[0097] Therefore, the correlation coefficient matrix RM between the design year x and the operating year y is obtained x,y :

[0098]

[0099] Step S4.2: Since there are Y types of working condition years, a total of Y correlation coefficient matrices are obtained, which are: Correlation coefficient matrix RM x,1 ,RM x,2 ,...,RM x,Y ;

[0100] For the correlation coefficient matrix RM x,1 ,RM x,2 ,...,RM x,Y The indicators in are ranked by entropy weight method to obtain the weight factor α for each working condition year y. y ;

[0101] Step S4.3: The fluctuation sequence {p y (1),p y (2),...,p y (N)}, according to the weight factor α y By superimposing each time period t, the benchmark load factor sequence {p r (1),p r (2),...,p r (N)}, the specific formula is:

[0102]

[0103] The base load factor sequence {p r (1),p r (2),...,p r (N)}.

[0104] Step S5, taking the month as the analysis unit, the benchmark load factor sequence {p r (1),p r (2),...,p r (N)} to determine the week in which the load factor p(t) is the maximum in each i-th month of the reference operating year r, which is expressed as week W. Since the reference operating year r has 12 months, it is located at 12 weeks;

[0105] According to the weekly positioning results of the reference operating condition year r, the monthly load characteristic sequence {M x,1 ,M x,2 ,...,M x,12} is converted into a fine-grained weekly load characteristic sequence by:

[0106] The monthly load characteristic M of the i-th month of design year x is x,i , assigned to the corresponding week W located as the weekly load characteristics of the corresponding week W, thereby obtaining 12 weekly load characteristics in the design year x;

[0107] The 12 weekly load characteristics are interpolated to obtain the weekly load characteristic sequence of all weeks in the design year x, which is expressed in order as F x,1 ,F x,2 ,...,F x,U ; Where U represents the number of weeks in a year;

[0108] Step S6, taking quarter as the analysis unit, the benchmark load factor sequence {p r (1),p r (2),...,p r (N)} to determine the week with the maximum load factor p(t) in each jth quarter of the reference operating year r. Since the reference operating year r has 4 quarters, it is located at 4 weeks;

[0109] According to the weekly positioning results of the reference operating condition year r, the quarterly load characteristics S of the design year x are x,j ={p x,j,k (t)}={p x,j,k (1),p x,j,k (2),...,p x,j,k (24)} is converted into a fine-grained load factor fluctuation sequence of unknown typical days in a week as the weekly load feature, and the method is:

[0110] The quarterly load characteristic S of the jth quarter of design year x is x,j ={p x,j,k (t)}={p x,j,k (1),p x,j,k (2),...,p x,j,k (24)}, assigned to the corresponding week located as the weekly load characteristics of the corresponding week, thereby obtaining four weekly load characteristics in the design year x;

[0111] The four weekly load characteristics are interpolated to obtain the weekly load characteristics of the design year x in all weeks, which are expressed in order as FC x,1 ,FC x,2 ,...,FC x,U ;

[0112] Step S7, further analyzing the weekly load characteristics F of all weeks in the design year x x,1 ,F x,2 ,...,F x,U and weekly load characteristics FC x,1 ,FC x,2 ,...,FC x,U , and the correlation between the weekly load characteristics and weekly load characteristics of the benchmark operating year r in all weeks, and obtain the power load factor p(t) curve for each period t that integrates the current and future design annual load characteristics.

[0113] Step S7 is specifically as follows:

[0114] Step S7.1, integrating the weekly load characteristics to the benchmark load factor sequence {p r (1),p r (2),...,p r (N)} to make corrections:

[0115] Step S7.1.1, in units of weeks, calculate the base load factor sequence {p r (1),p r (2),...,p r (N)}, the weekly load characteristic sequence of the benchmark operating condition year r is obtained, which is expressed as: F r,1 ,F r,2 ,...,F r,U ;

[0116] Step S7.1.2: The weekly load characteristic sequence F of design year x is x,1 ,F x,2 ,...,F x,U The weekly load characteristic sequence F of the reference operating condition year r r,1 ,F r,2,...,F r,U By comparison, the weekly load characteristic expansion coefficient β of each week u in the year is obtained according to the following formula u , where u=1,2,...,U:

[0117] β u =F x,u / F r,u

[0118] Step S7.1.3, for the base load factor sequence {p r (1),p r (2),...,p r Each benchmark load factor in (N)} is multiplied by the weekly load characteristic expansion coefficient of the week in which it is located, thereby obtaining the revised benchmark load factor sequence of the benchmark operating condition year r, which is expressed as:

[0119] Step S7.2, integrating the load factor fluctuation sequence of typical days in the week, and correcting the benchmark load factor sequence of the benchmark working condition year r To make corrections:

[0120] Step S7.2.1, in units of weeks, the reference load factor series of the revised reference operating condition year r After analysis, the weekly load characteristic sequence is obtained, which is expressed as: FC r,1 ,FC r,2 ,...,FC r,U ;

[0121] Step S7.2.2: The weekly load characteristic sequence FC of design year x is x,1 ,FC x,2 ,...,FC x,U The weekly load characteristic sequence FC of the benchmark operating condition year r r,1 ,FC r,2 ,...,FC r,U By comparison, the weekly load characteristic expansion coefficient γ of each week u in the year is obtained as follows: u , where u=1,2,...,U:

[0122] γ u =FC x,u / FC r,u

[0123] Step S7.2.3: The reference load factor sequence of the revised reference operating condition year r Each benchmark load factor in is multiplied by the weekly load characteristic expansion coefficient of the week in which it is located, thereby obtaining the revised benchmark load factor sequence of the benchmark operating condition year r, which is expressed as:

[0124] Step S7.3: The modified base load factor sequence of base operating condition year r Further correction is made to obtain the corrected benchmark load factor sequence of benchmark operating year r

[0125] Step S7.3.1: The base load factor sequence of the revised base operating year r Add up and get the corrected annual utilization hours t r , according to the following formula, we can get the annual expansion coefficient δ:

[0126] δ=t x / t r

[0127] Step S7.3.2, determine whether the deviation between the annual expansion coefficient δ and 1 is less than the set threshold, if yes, execute step S7.3.5; if no, execute step S7.3.3;

[0128] Step S7.3.3, the base load factor sequence for the revised base operating year r In the example, the weekly unit is divided into U sequences, which are expressed as: {C(1), C(2), ..., C(U)}; for the sequence C(u), u = 1, 2, ..., U, which represents the benchmark load factor sequence corresponding to the uth week;

[0129] Since there are Z time periods per week, the sequence C(u) has Z base load factors, expressed as:

[0130] For the sequence C(u), the day is taken as the unit. Since a week has 7 days, the sequence C(u) is further divided into 7 subsequences, which are represented as: {D(1), D(2), ..., D(7)}; in the subsequences D(1), D(2), ..., D(7), the day where the maximum value of the modified benchmark load factor is located is identified, and is set as day d, d = 1, 2, ..., 7, then day d is the typical day of the uth week, and the modified benchmark load factor sequence in the subsequence D(d) corresponding to the typical day of the uth week is not modified and remains unchanged;

[0131] The subsequences corresponding to the other six atypical days in week u are expressed as: D(z), z = 1, 2, ..., 7, and z ≠ d; since there are 24 time periods per day, there are 24 modified baseline load factor sequences in the atypical day sequence D(z), which are re-expressed as: The maximum load factor and the minimum load factor in are expressed as: and

[0132] For each modified base load factor in the atypical day sequence D(z) Where I = 1, 2, ..., 24, the following formula is used for correction to obtain the corrected reference load factor

[0133]

[0134] Step S7.3.4, therefore, adopt the method of step S7.3.3 to modify the benchmark load factor sequence of the revised benchmark operating year r After each item in is corrected, the corrected benchmark load factor sequence of benchmark operating year r is obtained.

[0135] use replace Return to step S7.3.1 and perform the cyclic correction process;

[0136] Step S7.3.5, output the final corrected benchmark load factor sequence for benchmark operating year r Thus, a time-period power load factor curve is obtained that integrates the current and future design annual load characteristics.

[0137] The present invention provides a method for generating a time-period power load curve integrating current and future annual load characteristics, which has the following characteristics:

[0138] Since only a small number of parameters are planned when planning the power load for the future design year, including: annual utilization hours, monthly load characteristic sequence and quarterly load characteristic sequence of the design year; therefore, it is necessary to determine the power load of each time period of the design year that meets the power load planning needs and form a power load curve, thereby providing data support for the optimal scheduling and planning of the power system.

[0139] In view of this situation, the present invention first obtains the load factor fluctuation sequence of various working conditions in the study area in each period, and generates a benchmark load factor sequence reflecting the design annual load characteristics by comparing the correlation between the working condition annual load and the design annual load;

[0140] Then, through the characteristics of the benchmark load factor sequence, the planned monthly load characteristic sequence and quarterly load characteristic sequence of the design year are located to the fine-grained week, and then interpolated to obtain the weekly load characteristics and weekly load characteristics of the design year;

[0141] Then, by continuously revising the benchmark load factor sequence, a revised benchmark load factor sequence that meets the weekly load characteristics and weekly load features of the design year is obtained. At this time, the benchmark load factor sequence has integrated the load characteristics of the current and future years.

[0142] Then, the annual utilization hours of the revised benchmark load factor sequence are analyzed to determine whether they meet the annual utilization hours planned for the design year. If not, the method of the present invention is used for further correction. After multiple iterations of correction, the final time-period power load process that meets all planning requirements for the design year is obtained, that is, the time-period power load curve that integrates the current and future annual load characteristics.

[0143] The method for generating a time-period power load curve integrating current and future annual load characteristics provided by the present invention has the following advantages:

[0144] In the case that it is difficult to obtain sufficient long series, multi-scale, multi-type and multi-source data in the study area, the present invention proposes a method for generating a time-period power load curve that integrates the current and future annual load characteristics. The method uses limited current annual power grid operation data and future design year power load operation characteristic data to predict the time-period power load curve in the study area. This method can use less data to quickly predict the power load curve process of the study area. The prediction result is accurate and reliable, providing data support for power system planning and optimal scheduling.

[0145] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be considered as the scope of protection of the present invention.

Claims

1. A method for generating a time-period power load curve integrating current and future annual load characteristics, characterized in that: The following steps are involved: Step S1, obtain the power load planning parameters of the future design year x in the study area, including: annual utilization hours t x , monthly load characteristic sequence of design year x {M x,1 ,M x,2 ,...,M x,12 } and seasonal load characteristic series {S x,1 ,S x,2 ,S x,3 ,S x,4 }; Among them, the monthly load characteristic M x,i , i=1,2,...,12, meaning: the maximum value of the load factor p(t) in each time period t in the i-th month of the design year x; among which, the time period in which the maximum value of the load factor p(t) occurs is an unknown quantity; Seasonal load characteristics S x,j , j = 1, 2, 3, 4, meaning: load factor fluctuation sequence of unknown typical day k in the jth quarter of design year x, that is: S x,j ={p x,j,k (t)}={p x,j,k (1),p x,j,k (2),...,p x,j,k (24)}, where the unknown typical day k is the day to be determined, which is the day when the maximum value of the load factor p(t) in the jth quarter of the design year x occurs; Step S2, select Y working conditions in the study area, and obtain the fluctuation sequence of the load factor in each working condition year y in each time period t, y = 1, 2, ..., Y, expressed as {p y (1),p y (2),...,p y (N)}, N represents the number of time periods t in the operating year y; Step S3: for each load factor fluctuation sequence {p y (1),p y (2),...,p y (N)} to obtain the monthly load characteristics M of each working condition year y in the i-th month y,i , and the quarterly load characteristics S of each operating year y in the jth quarter y,j ; Step S4, compare the monthly load characteristics M of various working conditions in year y y,i and the monthly load characteristic M of the design year x x,i The correlation between the seasonal load characteristics S of various working conditions in year y is compared. y,j and the quarterly load characteristics S of the design year x x,j The correlation is used to generate the reference load factor sequence {p r (1),p r (2),...,p r (N)}; wherein the reference operating year r has the same number of time periods t as the operating year y; Step S5, taking the month as the analysis unit, the benchmark load factor sequence {p r (1),p r (2),...,p r (N)} to determine the week in which the load factor p(t) is the maximum in each i-th month of the reference operating year r, which is expressed as week W. Since the reference operating year r has 12 months, it is located at 12 weeks; According to the weekly positioning results of the reference operating condition year r, the monthly load characteristic sequence {M x,1 ,M x,2 ,...,M x,12 } is converted into a fine-grained weekly load characteristic sequence by: The monthly load characteristic M of the i-th month of design year x is x,i , assigned to the corresponding week W located as the weekly load characteristics of the corresponding week W, thereby obtaining 12 weekly load characteristics in the design year x; The 12 weekly load characteristics are interpolated to obtain the weekly load characteristic sequence of all weeks in the design year x, which is expressed in order as F x,1 ,F x,2 ,...,F x,U ; Where U represents the number of weeks in a year; Step S6, taking quarter as the analysis unit, the benchmark load factor sequence {p r (1),p r (2),...,p r (N)} to determine the week with the maximum load factor p(t) in each jth quarter of the reference operating year r. Since the reference operating year r has 4 quarters, it is located at 4 weeks; According to the weekly positioning results of the reference operating condition year r, the quarterly load characteristics S of the design year x are x,j ={p x,j,k (t)}={p x,j,k (1),p x,j,k (2),...,p x,j,k (24)} is converted into a fine-grained load factor fluctuation sequence of unknown typical days in a week as the weekly load feature, and the method is: The quarterly load characteristic S of the jth quarter of design year x is x,j ={p x,j,k (t)}={p x,j,k (1),p x,j,k (2),...,p x,j,k (24)}, assigned to the corresponding week located as the weekly load characteristics of the corresponding week, thereby obtaining four weekly load characteristics in the design year x; The four weekly load characteristics are interpolated to obtain the weekly load characteristics of the design year x in all weeks, which are expressed in order as FC x,1 ,FC x,2 ,...,FC x,U ; Step S7, further analyzing the weekly load characteristics F of all weeks in the design year x x,1 ,F x,2 ,...,F x,U and weekly load characteristics FC x,1 ,FC x,2 ,...,FC x,U , and the correlation between the weekly load characteristics and weekly load characteristics of the benchmark operating year r in all weeks, and obtain the power load factor p(t) curve for each period t that integrates the current and future design annual load characteristics.

2. The method for generating a time-period power load curve integrating current and future annual load characteristics according to claim 1 is characterized in that: In step S1, the annual utilization hours t x The method to obtain is: t x =Q x / P x *10 5 in: Q x is the annual power demand in design year x; P x It is the maximum load of the whole society in design year x.

3. The method for generating a time-period power load curve integrating current and future annual load characteristics according to claim 1 is characterized in that: Step S4 is specifically as follows: Step S4.1, calculate the monthly load characteristic sequence {M x,1 ,M x,2 ,...,M x,12 } and the monthly load characteristic sequence of working year y {M y,1 ,M y,2 ,...,M y,12 The correlation coefficient R between x,y , relative error RE x,y and efficiency coefficient NASH x,y ; Calculate the quarterly load characteristics S in the jth quarter of the design year x x,j The quarterly load characteristic S of the operating year y in the jth quarter y,j The correlation coefficient between Relative error and efficiency coefficient Where, j = 1, 2, 3, 4; Therefore, the correlation coefficient matrix RM between the design year x and the operating year y is obtained x,y : Step S4.2: Since there are Y types of working condition years, a total of Y correlation coefficient matrices are obtained, which are: Correlation coefficient matrix RM x,1 ,RM x,2 ,...,RM x,Y ; For the correlation coefficient matrix RM x,1 ,RM x,2 ,...,RM x,Y The indicators in are ranked by entropy weight method to obtain the weight factor α for each working condition year y. y ; Step S4.3: The fluctuation sequence {p y (1),p y (2),...,p y (N)}, according to the weight factor α y By superimposing each time period t, the benchmark load factor sequence {p r (1),p r (2),...,p r (N)}, the specific formula is: The base load factor sequence {p r (1),p r (2),...,p r (N)}.

4. The method for generating a time-period power load curve integrating current and future annual load characteristics according to claim 1 is characterized in that: Step S7 is specifically as follows: Step S7.1, integrating the weekly load characteristics to the benchmark load factor sequence {p r (1),p r (2),...,p r (N)} to make corrections: Step S7.1.1, in units of weeks, calculate the base load factor sequence {p r (1),p r (2),...,p r (N)}, the weekly load characteristic sequence of the benchmark operating condition year r is obtained, which is expressed as: F r,1 ,F r,2 ,...,F r,U ; Step S7.1.2: The weekly load characteristic sequence F of design year x is x,1 ,F x,2 ,...,F x,U The weekly load characteristic sequence F of the reference operating condition year r r,1 ,F r,2 ,...,F r,U By comparison, the weekly load characteristic expansion coefficient β of each week u in the year is obtained according to the following formula u , where u=1,2,...,U: b u =F x,u / F r,u Step S7.1.3, for the base load factor sequence {p r (1),p r (2),...,p r Each benchmark load factor in (N)} is multiplied by the weekly load characteristic expansion coefficient of the week in which it is located, thereby obtaining the revised benchmark load factor sequence of the benchmark operating condition year r, which is expressed as: Step S7.2, integrating the load factor fluctuation sequence of typical days in the week, and correcting the benchmark load factor sequence of the benchmark working condition year r To make corrections: Step S7.2.1, in units of weeks, the reference load factor series of the revised reference operating condition year r After analysis, the weekly load characteristic sequence is obtained, which is expressed as: FC r,1 ,FC r,2 ,...,FC r,U ; Step S7.2.2: The weekly load characteristic sequence FC of design year x is x,1 ,FC x,2 ,...,FC x,U The weekly load characteristic sequence FC of the benchmark operating condition year r r,1 ,FC r,2 ,...,FC r,U By comparison, the weekly load characteristic expansion coefficient γ of each week u in the year is obtained as follows: u , where u=1,2,...,U: γ u =FC x,u / FC r,u Step S7.2.3: The reference load factor sequence of the revised reference operating year r Each benchmark load factor in is multiplied by the weekly load characteristic expansion coefficient of the week in which it is located, thereby obtaining the revised benchmark load factor sequence of the benchmark operating condition year r, which is expressed as: Step S7.3: The modified reference load factor sequence of reference operating year r Further correction is made to obtain the corrected benchmark load factor sequence of benchmark operating year r Step S7.3.1: The base load factor sequence of the revised base operating year r Add up and get the corrected annual utilization hours t r , according to the following formula, we can get the annual expansion coefficient δ: δ=t x / t r Step S7.3.2, determine whether the deviation between the annual expansion coefficient δ and 1 is less than the set threshold, if yes, execute step S7.3.5; if no, execute step S7.3.3; Step S7.3.3, the base load factor sequence for the revised base operating year r In the example, the weekly unit is divided into U sequences, which are expressed as: {C(1), C(2), ..., C(U)}; for the sequence C(u), u = 1, 2, ..., U, which represents the benchmark load factor sequence corresponding to the uth week; Since there are Z time periods per week, the sequence C(u) has Z base load factors, expressed as: For the sequence C(u), the day is taken as the unit. Since a week has 7 days, the sequence C(u) is further divided into 7 subsequences, which are represented as: {D(1), D(2), ..., D(7)}; in the subsequences D(1), D(2), ..., D(7), the day where the maximum value of the modified benchmark load factor is located is identified, and is set as day d, d = 1, 2, ..., 7, then day d is the typical day of the uth week, and the modified benchmark load factor sequence in the subsequence D(d) corresponding to the typical day of the uth week is not modified and remains unchanged; The subsequences corresponding to the other six atypical days in week u are expressed as: D(z), z = 1, 2, ..., 7, and z ≠ d; since there are 24 time periods per day, there are 24 modified baseline load factor sequences in the atypical day sequence D(z), which are re-expressed as: The maximum load factor and the minimum load factor in are expressed as: and For each modified base load factor in the atypical day sequence D(z) Where I = 1, 2, ..., 24, the following formula is used for correction to obtain the corrected reference load factor Step S7.3.4, therefore, adopt the method of step S7.3.3 to modify the benchmark load factor sequence of the revised benchmark operating year r After each item in is corrected, the corrected benchmark load factor sequence of benchmark operating year r is obtained. use replace Return to step S7.3.1 and perform the cyclic correction process; Step S7.3.5, output the final corrected benchmark load factor sequence for benchmark operating year r Thus, a time-period power load factor curve is obtained that integrates the current and future design annual load characteristics.

Images