A Power System Load Composition Identification Method Based on Sparse Dictionary Learning

Abstract

This invention discloses a power system load component identification method based on sparse dictionary learning. It includes: acquiring standardized historical load data and standardized historical meteorological data for various types of loads, establishing a comprehensive historical load data dictionary and a comprehensive historical meteorological data dictionary; acquiring load data for the day to be identified and meteorological data for various types of loads on the day to be identified; and, based on the comprehensive historical load data dictionary and the comprehensive historical meteorological data dictionary, using an orthogonal matching pursuit algorithm incorporating meteorological factors to perform load component identification on the load data for the day to be identified, obtaining the load component identification result. This load component identification result is used for power system load forecasting or scheduling. This method can decompose the load curve of the day to be identified into load curves of various components and calculate the proportion of each component, which has significant value and meaning for identifying the proportion of generalized load components.

Description

Technical Field

[0001] This invention belongs to the field of power system technology, specifically relating to a power system load component identification method based on sparse dictionary learning. Background Technology

[0002] With the increasing penetration of new energy equipment such as distributed photovoltaics, electric vehicles, air conditioning, and small hydropower into the power grid, the accurate extraction and proportion identification of typical load components in the power system have become more complex, posing a severe challenge to the study of the observability of generalized load terminals. Therefore, it is necessary to study the problem of load component identification.

[0003] The generalized load components are numerous and complex. Coupled with unavoidable measurement errors and the randomness of the load itself, the compositional decomposition of these components is diverse, making it impossible to accurately characterize various typical loads and difficult to directly select typical load curves for proportion identification. Therefore, it is necessary to explore a method that can accurately characterize the proportion of load components. Summary of the Invention

[0004] To address the problems existing in the background technology, this invention provides a power system load component identification method based on sparse dictionary learning. First, this invention establishes a historical data dictionary based on historical load data and historical meteorological data. Then, based on information such as the types of load components, load data, and meteorological data for the day to be identified, the component decomposition problem is modeled as a sparse dictionary model. Finally, an orthogonal matching pursuit algorithm incorporating meteorological factors is used to decompose the load curve of the day to be identified. This method can decompose the load curve of the day to be identified into load curves of each component and calculate the proportion of each component, which has certain value and significance for identifying the proportion of generalized load components.

[0005] The technical solution adopted in this invention is:

[0006] The method of the present invention includes the following steps:

[0007] S1) Collect historical load data and historical meteorological data of various types of loads, and obtain standardized historical load data and standardized historical meteorological data of various types of loads through preprocessing.

[0008] The various types of loads mentioned are selected from photovoltaic, wind power, air conditioning loads, small hydropower, charging piles, and base loads.

[0009] S2) Establish a comprehensive historical load data dictionary and a comprehensive historical meteorological data dictionary based on standardized historical load data and standardized historical meteorological data of various types of loads.

[0010] Step S2 includes the following steps:

[0011] S2.1) Establish historical load matrices for each type of load based on standardized historical load data, and combine the historical load matrices of all loads to obtain a comprehensive historical load data dictionary; in the historical load matrix of each type of load, the column vector is the load data collected from all sampling points on the same sampling day, and the row vector is the load data collected from the same sampling point on all sampling days, where the sampling point is the sampling time point.

[0012] S2.2) For each type of load, a historical meteorological matrix is ​​obtained by performing correlation analysis on standardized historical load data and standardized historical meteorological data; by combining the historical meteorological matrices of all loads, a comprehensive historical meteorological data dictionary is obtained.

[0013] Step S2.2 specifically includes:

[0014] S2.2.1) For each type of load, standardize the resolution of the historical load data and the historical meteorological data;

[0015] S2.2.2) After unifying the resolution, for each type of load, the Spearman correlation coefficient between the load data and each type of meteorological data is obtained by performing correlation analysis on the load data and multiple meteorological data for each sampling day.

[0016] S2.2.3) For each type of load, the weighted historical meteorological data for each sampling point on the sampling day is obtained according to the following formula, based on the multiple meteorological data for each sampling day and the Spearman correlation coefficient of each type of meteorological data for the load data:

[0017]

[0018] In the formula, The weighted historical meteorological data of the nth type of load at the i-th sampling point on the d-th day; For the nth type of load, the meteorological data of the kth type is collected at the i-th sampling point on the d-th day, where k = 1 to K. n K n f represents the total number of meteorological types for the nth type of load. n The weighted calculation function for the nth type of load is... Let ρ be the average correlation coefficient between the load data of type n and the meteorological data of type k. n,d,k Let be the correlation coefficient between the load data of type n on day d and the meteorological data of type k;

[0019] S2.2.4) For each type of load, combine the weighted historical meteorological data of all sampling points on each sampling day to obtain a unified historical meteorological vector for each day. Use the unified historical meteorological vector for each day as a column vector, arrange the column vectors in order of the collection date, and construct the historical meteorological matrix of the load.

[0020] The historical meteorological matrix for each type of load constructed in step S2.2.4 is as follows:

[0021]

[0022] In the formula, A wn The historical meteorological matrix for the nth type of load, This represents the weighted historical meteorological data of the nth type of load at the i-th sampling point on the d-th day, where d is the serial number of the sampling day, d = 1 to D. n D n Let i be the total number of sampling days for the nth type of load, and i be the index of the sampling point, i = 1 to M. t M t This represents the total number of sampling points for each sampling day after unifying the resolution.

[0023] S3) Obtain the load data of the day to be identified and the meteorological data of various types of loads on the day to be identified. Based on the comprehensive historical load data dictionary and comprehensive historical meteorological data dictionary obtained in step S2, use the orthogonal matching pursuit algorithm combined with meteorological factors to identify the load components of the load data of the day to be identified, and obtain the load component identification results. The load component identification results are used for load forecasting or scheduling of the power system.

[0024] Step S3 specifically involves:

[0025] S3.1) Using the daily unified historical load vector (i.e., column vector) in the comprehensive historical load data dictionary as an element, and taking the minimum number of load data dictionary elements and accurate representation of the total load as the solution objective, a load component identification model is established based on the comprehensive historical load data dictionary.

[0026] In step S3.1, the load component identification model is specifically as follows:

[0027] X = arg min||E||2,

[0028] stY = AX + E

[0029] ||X||0≤L

[0030] In the formula, Y is the daily load vector to be identified, X is the component proportion vector, A is the comprehensive historical load data dictionary, E is the error vector, ||X||0 is the number of non-zero elements in the component proportion vector X, L is the preset upper limit value for selecting elements in the comprehensive historical load data dictionary, ||E||2 is the 2-norm of the error vector, that is, the square root of the sum of squares of errors of all sampling points, st is the constraint condition, arg min is the value of the variable that makes the formula reach the minimum value, and X=arg min||E||2 is the value of the component proportion vector X that makes the 2-norm of the error vector E reach the minimum value;

[0031] Specifically, the component proportion vector X is:

[0032]

[0033] In the formula, X n To form a matrix representing the proportion of elements of the nth type of load in the comprehensive historical load data dictionary, n = 1 to N, where N is the total number of load types, and D n The total number of sampling days for the nth type of load;

[0034] Specifically, the error vector E is:

[0035] E = [ε 1 ... ε i ... ε M ] T ∈R M×1

[0036] In the formula, ε i Let be the error of the i-th sampling point, i = 1 to M, where M is the total number of sampling points on the day to be identified, which is the same as the total number of sampling points on each sampling day.

[0037] S3.2) Obtain the load data of the day to be identified and the meteorological data of various types of loads on the day to be identified, and construct the load vector and meteorological vector of the day to be identified respectively; use the orthogonal matching pursuit algorithm combined with meteorological factors to perform load component identification on the load vector of the day to be identified according to the load component identification model and the meteorological vector of the day to be identified, and obtain the load component identification result.

[0038] Step S3.2 specifically includes:

[0039] S3.2.1) Collect load data from M sampling points on the day to be identified, construct a load vector for the day to be identified, and collect the location of each type of load on the day to be identified. w Meteorological data from sampling points are used to obtain meteorological data for various types of loads to be identified on a given day. The meteorological data for each type of load is then used to construct a meteorological vector for the day to be identified using a weighted calculation function for each type of load.

[0040] In step S3.2.1, the weighted calculation function f for each type of load is used according to the following formula. n The meteorological data of the day to be identified for the load is used to construct a meteorological vector for the day to be identified:

[0041]

[0042] In the formula, The weighted meteorological data for the nth type of load and the i-th sampling point. For the raw meteorological data of the i-th sampling point of the k-th meteorological data type to be identified, Y wn Let be the daily meteorological vector to be identified for the nth type of load.

[0043] S3.2.2) Take the inner product of the comprehensive historical meteorological data dictionary and the meteorological vector of the day to be identified to obtain the meteorological inner product matrix, which is used as the meteorological inner product matrix; at the same time, initialize the index list;

[0044] In step S3.2.2, the inner product matrix of the load inner product matrix and the meteorological inner product matrix are taken according to the following formula to obtain the comprehensive inner product matrix:

[0045]

[0046] In the formula, G j Let be the comprehensive inner product matrix obtained in the j-th iteration. Let G be the load inner product matrix obtained in the j-th iteration. w Let be the meteorological inner product matrix, and ⊙ denote the Hadamard product of the matrix.

[0047] S3.2.3) Take the inner product of the comprehensive historical load data dictionary and the daily load vector to be identified to obtain the load inner product matrix. Take the inner product of the load inner product matrix and the meteorological inner product matrix obtained in step S3.2.2 to obtain the comprehensive inner product matrix. From the comprehensive inner product matrix, select the index corresponding to the maximum value of other elements except the elements corresponding to the indexes in the index list and add them to the index list.

[0048] S3.2.4) According to the index list, select the elements corresponding to the index from the comprehensive historical load data dictionary to construct an index dictionary matrix. Input the index dictionary matrix as the comprehensive historical load data dictionary into the load component identification model. Process the load component identification model after inputting the index dictionary matrix by the least squares method to obtain the component proportion vector and residual vector.

[0049] S3.2.5) Using the residual vector as the daily load vector to be identified, new component proportion vectors and residual vectors are obtained according to steps S3.2.3 and S3.2.4;

[0050] S3.2.6) Repeat step S3.2.5 until the preset number of iterations or the residual converges, to obtain the final index list and component proportion vector. Superimpose the elements of the same type in the index list according to the component proportion vector to obtain the load component identification result.

[0051] The beneficial effects of this invention are:

[0052] 1. This invention takes into account historical load and meteorological big data, and incorporates the load and meteorological data of the day to be identified into the identification system, thereby improving the model decomposition accuracy;

[0053] 2. This invention considers different types of historical meteorological data, uses correlation analysis and weighted processing of meteorological data, and comprehensively examines the impact of different types of historical meteorological data;

[0054] 3. This invention uses a sparse dictionary learning method for load component identification, which can perform dimensionality reduction representation on massive datasets, has good noise reduction capabilities, and strong adaptability. Attached Figure Description

[0055] Figure 1 This is a flowchart of the load component identification method in this invention;

[0056] Figure 2 The daily load curves to be identified before and after the application of noise disturbance;

[0057] Figure 3 A 96-point load curve diagram of the daily load to be identified and its four components;

[0058] Figure 4 This is a diagram showing the identification results of the load component identification method in this invention. Detailed Implementation

[0059] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that these examples are only for illustrating the present invention and are not intended to limit the scope of the present invention. After reading the present invention, any modifications of the present invention in various equivalent forms by those skilled in the art fall within the scope defined by the appended claims.

[0060] This invention provides a load component identification method based on sparse dictionary learning, such as... Figure 1 As shown, it includes the following steps:

[0061] The method of the present invention includes the following steps:

[0062] S1) Collect historical load data and historical meteorological data of various types of loads, and obtain standardized historical load data and standardized historical meteorological data of various types of loads through preprocessing;

[0063] The various load types are selected from photovoltaic, wind power, air conditioning loads, small hydropower, charging piles, and base loads. In practice, to ensure identification accuracy, the number and composition of load types in historical load data and the daily load data to be identified can be kept consistent.

[0064] Historical meteorological data for all types of loads include multiple meteorological types, and the number and composition of meteorological types may differ for different loads. Meteorological types include weather conditions at the load location, temperature, temperature at 2m above ground, relative humidity at 2m above ground, pressure, wind speed at 10m above ground, wind direction at 10m above ground, wind speed at 100m above ground, wind direction at 100m above ground, and radiation, etc.

[0065] S2) Establish a comprehensive historical load data dictionary and a comprehensive historical meteorological data dictionary based on standardized historical load data and standardized historical meteorological data of various types of loads;

[0066] Step S2 includes the following steps:

[0067] S2.1) Based on the standardized historical load data of various types of loads, establish historical load matrices for each type of load, combine the historical load matrices of all loads to obtain a comprehensive historical load data dictionary; in the historical load matrix of each type of load, the column vector represents the load data collected from all sampling points on the same sampling day, the row vector represents the load data collected from the same sampling point on all sampling days, and the sampling point represents the sampling time point;

[0068] Specifically, in step S2.1, the historical load matrix for each type of load is as follows:

[0069]

[0070] In the formula, This is the historical load matrix for the nth type of load; n is the index of the load type, n = 1 to N, and N is the total number of load types; This refers to the load data of the nth type of load at the i-th sampling point on the d-th day, where d is the sequence number of the sampling day, and d = 1 to D. n D n Let represent the total number of sampling days for the nth type of load, where i is the serial number of the sampling point, i = 1 to M, and M is the total number of sampling points for each sampling day.

[0071] S2.2) For each type of load, a historical meteorological matrix is ​​obtained by performing correlation analysis on standardized historical load data and standardized historical meteorological data; by combining the historical meteorological matrices of all loads, a comprehensive historical meteorological data dictionary is obtained.

[0072] Step S2.2 specifically includes:

[0073] S2.2.1) For each type of load, the standardized historical load data and standardized historical meteorological data are unified in resolution; wherein, the unified resolution specifically means: unifying the number of sampling points and the time interval between two adjacent sampling points; after unifying the resolution, the historical load data and historical meteorological data have the same number of sampling points and the same time interval between sampling points;

[0074] S2.2.2) For each type of load, the Spearman correlation coefficient between the load data and each type of meteorological data is obtained by performing correlation analysis on the load data and multiple meteorological data for each sampling day.

[0075] S2.2.3) For each type of load, the weighted historical meteorological data for each sampling point on the sampling day is obtained according to the following formula, based on the multiple meteorological data for each sampling day and the Spearman correlation coefficient of each type of meteorological data for the load data:

[0076]

[0077] In the formula, The weighted historical meteorological data of the nth type of load at the i-th sampling point on the d-th day; For the nth type of load, the meteorological data of the kth type is collected at the i-th sampling point on the d-th day, where k = 1 to K. n K n f represents the total number of meteorological types for the nth type of load. n The weighted calculation function for the nth type of load is... Let ρ be the average correlation coefficient between the load data of type n and the meteorological data of type k. n,d,k Let be the correlation coefficient between the load data of type n on day d and the meteorological data of type k;

[0078] S2.2.4) For each type of load, combine the weighted historical meteorological data of all sampling points on each sampling day to obtain a unified historical meteorological vector for each day. Use the unified historical meteorological vector for each day as a column vector, arrange the column vectors in order of the collection date, and construct the historical meteorological matrix of the load.

[0079] The historical meteorological matrix for each type of load constructed in step S2.2.4 is as follows:

[0080]

[0081] In the formula, A wn The historical meteorological matrix for the nth type of load, This represents the weighted historical meteorological data of the nth type of load at the i-th sampling point on the d-th day, where d is the serial number of the sampling day, d = 1 to D. n D n Let i be the total number of sampling days for the nth type of load, and i be the index of the sampling point, i = 1 to M. t M t This represents the total number of sampling points for each sampling day after unifying the resolution.

[0082] S3) Obtain the load data of the day to be identified and the meteorological data of various types of loads on the day to be identified. Based on the comprehensive historical load data dictionary and comprehensive historical meteorological data dictionary obtained in step S2, use the orthogonal matching pursuit algorithm combined with meteorological factors to identify the load components of the load data of the day to be identified, and obtain the load component identification results. The load component identification results are used for load forecasting or scheduling of the power system.

[0083] Step S3 is as follows:

[0084] S3.1) Using the daily unified historical load vector (i.e., column vector) in the comprehensive historical load data dictionary as an element, and taking the minimum number of load data dictionary elements and accurate representation of the total load as the solution objective, a load component identification model is established based on the comprehensive historical load data dictionary.

[0085] In step S3.1, the load component identification model is specifically as follows:

[0086] X = arg min||E||2,

[0087] stY = AX + E

[0088] ||X||0≤L

[0089] In the formula, Y is the daily load vector to be identified, X is the component proportion vector, A is the comprehensive historical load data dictionary, E is the error vector, ||X||0 is the number of non-zero elements in the component proportion vector X, L is the preset upper limit value for selecting elements in the comprehensive historical load data dictionary, ||E||2 is the 2-norm of the error vector, that is, the square root of the sum of squares of errors of all sampling points, st is the constraint condition, arg min is the value of the variable that makes the formula reach the minimum value, and X=arg min||E||2 is the value of the component proportion vector X that makes the 2-norm of the error vector E reach the minimum value;

[0090] Specifically, the component proportion vector X is:

[0091]

[0092] In the formula, X n To form a matrix representing the proportion of elements of the nth type of load in the comprehensive historical load data dictionary, n = 1 to N, where N is the total number of load types, and D n The total number of sampling days for the nth type of load;

[0093] Specifically, the error vector E is:

[0094] E = [ε 1 ... ε i ... ε M ] T ∈RM×1

[0095] In the formula, ε i Let be the error of the i-th sampling point, i = 1 to M, where M is the total number of sampling points on the day to be identified, which is the same as the total number of sampling points on each sampling day.

[0096] Wherein, the load vector Y for the day to be identified is a vector composed of load data collected from M sampling points on the day to be identified, expressed as:

[0097] Y = [y 1 ... y i ... y M ] T ∈R M×1

[0098] In the formula, y i This refers to the load data collected at the i-th sampling point on the day to be identified.

[0099] S3.2) Obtain the load data of the day to be identified and the meteorological data of various types of loads on the day to be identified, and construct the load vector and meteorological vector of the day to be identified respectively; use the orthogonal matching pursuit algorithm combined with meteorological factors to perform load component identification on the load vector of the day to be identified according to the load component identification model and the meteorological vector of the day to be identified, and obtain the load component identification result.

[0100] Step S3.2 specifically includes:

[0101] S3.2.1) Collect load data from M sampling points on the day to be identified, construct a load vector for the day to be identified, and collect the location of each type of load on the day to be identified. w Meteorological data from sampling points are used to obtain meteorological data for various types of loads to be identified on a given day. The meteorological data for each type of load is then used to construct a meteorological vector for the day to be identified using a weighted calculation function for each type of load.

[0102] In step S3.2.1, the weighted calculation function f for each type of load is used according to the following formula. n The meteorological data of the day to be identified for the load is used to construct a meteorological vector for the day to be identified:

[0103]

[0104] In the formula, The weighted meteorological data for the nth type of load and the i-th sampling point. For the raw meteorological data of the i-th sampling point of the k-th meteorological data type to be identified, Y wn Let be the daily meteorological vector to be identified for the nth type of load.

[0105] S3.2.2) Take the inner product of the comprehensive historical meteorological data dictionary and the meteorological vector of the day to be identified to obtain the meteorological inner product matrix, which is used as the meteorological inner product matrix; at the same time, initialize the index list;

[0106] Specifically: For meteorological data of each type of load, a load weighting calculation function is used to perform weighted summation on all types of meteorological data at each sampling point to obtain unified meteorological data for the sampling point. The load is then calculated on the day M to be identified. w The unified meteorological data combination of each sampling point constitutes the daily meteorological vector to be identified for the load;

[0107] In step S3.2.2, the inner product matrix of the load inner product matrix and the meteorological inner product matrix are taken according to the following formula to obtain the comprehensive inner product matrix:

[0108]

[0109] In the formula, G j Let be the comprehensive inner product matrix obtained in the j-th iteration. Let G be the load inner product matrix obtained in the j-th iteration. w Let be the meteorological inner product matrix, and ⊙ denote the Hadamard product of the matrix.

[0110] S3.2.3) Take the inner product of the comprehensive historical load data dictionary and the daily load vector to be identified to obtain the load inner product matrix. Take the inner product of the load inner product matrix and the meteorological inner product matrix obtained in step S3.2.2 to obtain the comprehensive inner product matrix. From the comprehensive inner product matrix, select the index corresponding to the maximum value of other elements except the elements corresponding to the indexes in the index list and add them to the index list.

[0111] S3.2.4) According to the index list, select the elements corresponding to the index from the comprehensive historical load data dictionary to construct an index dictionary matrix. Input the index dictionary matrix as the comprehensive historical load data dictionary into the load component identification model. Process the load component identification model after inputting the index dictionary matrix by the least squares method to obtain the component proportion vector and residual vector.

[0112] S3.2.5) Using the residual vector as the daily load vector to be identified, new component proportion vectors and residual vectors are obtained according to steps S3.2.3 and S3.2.4;

[0113] S3.2.6) Repeat step S3.2.5 until the preset number of iterations or the residual converges, to obtain the final index list and component proportion vector. Superimpose the elements of the same type in the index list according to the component proportion vector to obtain the load component identification result.

[0114] Specific embodiments of the present invention are as follows:

[0115] This example uses 10kV load data from a province in China from May to August 2023, along with matching local meteorological data, for a case study analysis. The load data includes 10kV medium-voltage outgoing line data for four different load types: air conditioning, charging piles, photovoltaic, and base load. Data is sampled every 15 minutes, for a total of 96 points per day (M=96). The meteorological data includes 10 different meteorological types of 10kV medium-voltage outgoing line data, including: weather, temperature, temperature at 2m above ground, relative humidity at 2m above ground, pressure, wind speed at 10m above ground, wind direction at 10m above ground, wind speed at 100m above ground, wind direction at 100m above ground, and radiation. Data is sampled every 60 minutes, for a total of 24 points per day (M=96). w =24).

[0116] In this embodiment, four different types of load elements for the same day in the dictionary are superimposed, and white noise with a mean of 0 and a variance of 0.005 is applied to simulate the load of the day to be identified.

[0117] This embodiment provides a load component identification method based on sparse dictionary learning, such as... Figure 1 As shown, it includes the following steps:

[0118] S1) Collect historical load data and historical meteorological data for various load types, and perform data preprocessing to reduce the impact of abnormal sampling points, thereby obtaining standardized historical load data and standardized historical meteorological data for various load types; Step S1 specifically includes:

[0119] S1.1) In this embodiment, the total number of load types is N (N=4). Historical load data of N types of loads are collected, outliers and missing values ​​are processed on the historical load data, and then normalization is performed to obtain standardized historical load data of four types of loads.

[0120] Taking 96 load data points from a single day as an example (i.e., M=96), the data is normalized according to the following formula:

[0121]

[0122] in, This represents the normalized load data for the i-th sampling point on day d. This represents the raw load data for the i-th sampling point on day d; d is the sampling day number, and i is the sampling point number, ranging from 1 to 96.

[0123] S1.2) After processing outliers and missing values ​​in the historical meteorological data, for historical meteorological data with different measurement resolutions, where the lowest and highest resolutions are 1 hour and 15 minutes respectively, the historical meteorological data are sampled at intervals to unify their resolution to the lowest resolution of 1 hour (i.e., the total number of sampling points M for each sampling day of the historical meteorological data). w =24), and standardized historical meteorological data for four types of loads are obtained.

[0124] S2) Based on the data obtained in step S1, establish a comprehensive historical load data dictionary, and conduct correlation analysis between the daily historical load data of various types of loads and the corresponding daily meteorological data to comprehensively analyze the impact of various meteorological factors on daily loads, and finally establish a comprehensive historical meteorological data dictionary.

[0125] Step S2 specifically includes the following steps:

[0126] S2.1) Establish the historical load matrix for N=4 load categories. The historical load matrix for each load category is shown below:

[0127]

[0128] In the formula, This is the historical load matrix for the nth type of load; n is the index of the load type, n = 1 to N, and N is the total number of load types; This refers to the load data of the nth type of load at the i-th sampling point on the d-th day, where d is the sequence number of the sampling day, and d = 1 to D. n D n Let i be the total number of sampling days for the nth type of load, and i be the serial number of the sampling point, i = 1 to 96.

[0129] S2.2) Perform correlation analysis between the daily historical load data of type N loads and the corresponding daily meteorological data. Taking photovoltaics as an example, let its load number be n, and its historical meteorological data include K n Categories, such as the weather of the day, temperature, temperature at 2m above the ground, relative humidity at 2m above the ground, pressure, wind speed at 10m above the ground, wind direction at 10m above the ground, wind speed at 100m above the ground, wind direction at 100m above the ground, radiation, etc.

[0130] After processing in step S1, the resolutions of the historical load data and historical meteorological data for photovoltaics are 15 minutes and 1 hour, respectively. To standardize the resolution, the daily historical load data for photovoltaics is first... Interval sampling is Then calculate its relationship with the corresponding daily K. n Historical meteorological data (where k is the sequence number of the weather type, k = 1 to K) n The Spearman correlation coefficient of ) is calculated using the following formula:

[0131]

[0132] Where, ρ n,d,k The correlation coefficient between the photovoltaic load data on day d and the meteorological data of type k; This refers to the load data of the photovoltaic system at the i-th sampling point on the d-th day. This represents the average load data from all sampling points on day d of the photovoltaic system. For photovoltaic data of type k at the i-th sampling point on day d, This represents the average value of the k-th type of meteorological data from all sampling points on day d of the photovoltaic system.

[0133] Correlation analysis was used to obtain the correlation coefficient ρ between the daily historical photovoltaic load data and the corresponding daily meteorological data. n,d,k (d=1~D n k = 1 to K n After that, the average correlation coefficient of various photovoltaic meteorological data is calculated according to the following formula:

[0134]

[0135] In the formula, Let ρ be the average correlation coefficient between photovoltaic load data and type k meteorological data. n,d,k Let be the correlation coefficient between the photovoltaic load data on day d and the meteorological data of type k.

[0136] To comprehensively examine the relationship between various meteorological data for photovoltaic (PV) systems and daily load data, a unified historical meteorological vector for daily PV systems is calculated by weighting the various meteorological data according to their average correlation coefficients, as shown in the following formula:

[0137]

[0138] in, The weighted historical meteorological data for the photovoltaic system at the i-th sampling point on the d-th day; f n This is the weighted calculation function for photovoltaics.

[0139] After the above weighted calculation process, a photovoltaic historical meteorological data dictionary A is established based on the daily unified historical meteorological vector. wn The details are as follows:

[0140]

[0141] Similarly, a historical load and meteorological data dictionary for N types of loads is established, and finally, a historical load data dictionary A and a historical meteorological data dictionary A for all types of loads are established. w The details are as follows:

[0142]

[0143] S3) Based on the comprehensive historical load data dictionary and comprehensive historical meteorological data dictionary obtained in step S2, the load component identification is performed on the load data of the day to be identified by using the orthogonal matching pursuit algorithm combined with meteorological factors, and the load component identification results are used for load forecasting or dispatching of the power system.

[0144] Step S3 is as follows:

[0145] S3.1) Treat the daily historical load data column vector in the historical load data dictionary A as an element of the dictionary, and then determine the solution objective as: select the fewest load data dictionary elements to represent the total load as accurately as possible, and then establish a solution model, as follows:

[0146] X = arg min||E|| 2 ,

[0147] stY = AX + E

[0148] ||X||0≤L

[0149] Where Y is the daily load vector of 96 points to be identified, Y = [y 1 ... y i ...y 96 ] T ∈R 96×1 y i This refers to the load data collected at the i-th sampling point on the day to be identified;

[0150] X is the component proportion vector. X n Let be the matrix representing the proportion of elements of the nth type of load in the dictionary.

[0151] E is the 96-point error vector, E = [ε] 1 ... ε i ... ε 96 ] T ∈R 96×1 , ε i The error of the i-th sampling point on the day to be identified;

[0152] ||X||0≤L means that the number of non-zero elements in the component proportion vector must be less than the set upper limit value L for selecting dictionary elements.

[0153] S3.2) For the above planning problem, an orthogonal matching pursuit algorithm incorporating meteorological factors is used to solve it. The specific steps are as follows:

[0154] Collect daily load data and meteorological data for the day to be identified, and preprocess the meteorological data to construct the meteorological vector for the day to be identified.

[0155] First, collect load data for the day to be identified [y]. 1 ... y i ... y 96 ] T With K n Meteorological data of various meteorological types

[0156] Then, in order to compare the meteorological data of the day to be identified with historical meteorological data, it is necessary to calculate the meteorological data of the day to be identified according to the weighted calculation function f corresponding to the load type. n Preprocessing:

[0157]

[0158] in, The meteorological data of the i-th sampling point after weighted calculation of the n-th type of load; Y represents the raw meteorological data of the i-th sampling point for the k-th meteorological data type to be identified; wn ∈R 24×1 The weighted calculation of the daily meteorological vector to be identified for the nth type of load.

[0159] 1) Initialization:

[0160] ① Calculate the initial comprehensive difference between the load and meteorological data of each element in the historical dictionary and the load and meteorological data of the day to be identified:

[0161] First, calculate the initial comprehensive difference between the load data of each element in the comprehensive historical load data dictionary and the load data of the day to be identified, that is, calculate the inner product of the load vector Y of the day to be identified and the comprehensive historical load data dictionary A to obtain the initial load inner product matrix:

[0162]

[0163] in, This is the initial load inner product matrix that integrates each element in the historical load data dictionary with the load data of the day to be identified; To synthesize the initial load inner product of the load data of the nth type of load and the element of the dth day in the historical load data dictionary with the load data of the day to be identified, the above formula can be written as the following detailed equation:

[0164]

[0165] In the formula, y i For the load data collected at the i-th sampling point on the day to be identified, This represents the load data for the nth type of load at the i-th sampling point on day d.

[0166] Then, the comprehensive difference between the meteorological data of each element in the comprehensive historical meteorological data dictionary and the meteorological data of the day to be identified is calculated, that is, the meteorological vector Y of the day to be identified is calculated. wn With Comprehensive Historical Meteorological Data Dictionary A w The inner product of these elements yields the initial meteorological inner product matrix:

[0167] G w =Y wn T A w

[0168]

[0169] Among them, G w This is a meteorological inner product matrix that integrates all elements in the historical meteorological data dictionary with the meteorological data of the day to be identified; g w,n,d To synthesize the meteorological inner product of the meteorological data of the nth type of load and the element of the dth day in the historical meteorological data dictionary with the meteorological data of the day to be identified, the above formula can be written as the following detailed equation:

[0170]

[0171] In the formula, The meteorological data of the i-th sampling point after weighted calculation of the n-th type of load. This represents the weighted historical meteorological data of the nth type of load at the i-th sampling point on the d-th day.

[0172] Finally, the initial integrated difference between each element in the historical dictionary and the daily load and meteorological data to be identified is calculated, i.e., the initial integrated inner product matrix G is calculated. 0 :

[0173]

[0174]

[0175] Among them, G 0 Let be the initial synthesized inner product matrix. The initial integrated inner product of the nth type of load, the element of the dth day, the load of the day to be identified, and the meteorological data in the historical load data dictionary is represented by ⊙, which denotes the Hadamard product of the matrix.

[0176] The above equation can be written as the following detailed equation:

[0177]

[0178] In the formula, To integrate the initial load inner product of the load data of the nth type of load and the element of the dth day in the historical load data dictionary with the load data of the day to be identified, g w,n,d The meteorological inner product is the sum of the meteorological data of the nth type of load and the element of the dth day in the historical meteorological data dictionary and the meteorological data of the day to be identified.

[0179] ② In order to attempt to solve for the component proportion vector, the element with the smallest difference from the curve to be identified is selected, that is, the initial inner product matrix G is selected. 0 The elements with the largest inner product form the initial index list Γ 0 :

[0180] Γ 0 ={G 0 Maximum index}

[0181] ③According to the index list Γ 0 Elements from the comprehensive historical load data dictionary A are selected to form the initial index dictionary matrix. Using the least squares method, based on the aforementioned load component identification model and initial index dictionary matrix... Solve for the initial component proportion vector And calculate the initial residual r 0 :

[0182]

[0183] in, For a list containing only indices Γ 0 The initial index dictionary matrix of elements, where Y is the daily load vector to be identified. For the index list Γ 0 The calculated initial component proportion vector, r 0 This is the initial residual vector.

[0184] 2) Start the loop, with loop number j = 1.

[0185] ①The residual vector r j-1 Treating the daily load vector to be identified, the elements in the comprehensive historical load data dictionary A and the residual vector r are obtained according to the following formula. j-1 The comprehensive inner product matrix G j :

[0186]

[0187] In the formula, G j Let be the comprehensive inner product matrix obtained in the j-th iteration. Let G be the load inner product matrix obtained in the j-th iteration. w Let r be the meteorological inner product matrix. j-1 Let be the residual vector obtained in the (j-1)th iteration, and ⊙ denote the Hadamard product of the matrix.

[0188] ② Select the index list Γ j-1 In addition to the marked elements, the comprehensive inner product matrix G j Add the element with the largest inner product to the index list to update the index list Γ. j :

[0189] Γ j ={G j-1 G j Maximum index}

[0190] In the formula, Γ j Let Γ be the list of indices obtained in the j-th iteration. j-1 This is the list of indices obtained in the (j-1)th iteration.

[0191] ③According to the index list Γ j Elements from the comprehensive historical load data dictionary A are selected to form an index dictionary matrix. The component proportion vector is solved using the least squares method based on the above load component identification model. And calculate the residual vector r j :

[0192]

[0193] In the formula, Let Y be the component proportion vector obtained from the j-th iteration, and let Y be the daily load vector to be identified. Let r be the index dictionary matrix obtained in the j-th iteration. j Let be the residual vector obtained in the j-th iteration.

[0194] Let the loop index j = j + 1, and repeat the above steps until the loop reaches its upper limit or the residual converges.

[0195] j≥j limit or||r j -r j-1 ||2≤r limit

[0196] In the formula, j limit To preset the number of iterations, ||r j -r j-1 ||2 is the 2-norm of the residual vector obtained from the j-th iteration and the (j-1)-th iteration, r limit This is the preset threshold for the residual.

[0197] 3) After the iteration stops, the final index list Γ is obtained. j With the final component proportion vector The final index list Γ j Elements of the same load type are arranged according to the final component proportion vector. By superimposing the results, we can obtain the final decomposition result:

[0198]

[0199] In the formula, This is the final component proportion vector when the iteration converges after j iterations; Y is the element proportion matrix of the nth type of load when iteration reaches convergence after j iterations; n-hat X is a 96-point summary curve of the identification results for the nth type of load; n-hat The identification result is the proportion of the nth type of load.

[0200] This embodiment describes the identification results of the daily load to be identified as follows: Figures 2-4 As shown.

[0201] Figure 2 This demonstrates the changes in the simulated load to be identified before and after applying noise disturbance. Figure 2 It can be seen that after applying the disturbance, the load to be identified fluctuates slightly while the overall trend remains unchanged, which simulates the actual situation well.

[0202] Figure 3 The diagram shows the daily load vector to be identified and its 96-point load curves for the four load types it contains.

[0203] Figure 4 The identification results using the method of this invention are presented. Table 1 shows the theoretical proportion, calculated proportion, and relative error of each load component. As can be seen from the table, the overall identification effect of the load component proportion is good.

[0204] Table 1. Theoretical proportion, calculated proportion and relative error of four load types

[0205] Typical ingredients air conditioner charging pile Photovoltaics base load Theoretical percentage (%) 19.96 9.95 19.95 50.13 Calculate the percentage (%) 17.31 10.40 20.53 51.75 Relative error (%) 13.27 4.56 2.90 3.22

[0206] The above description is only a preferred embodiment of the present invention. Any equivalent changes and modifications made within the scope of the claims of the present invention shall fall within the scope of the present invention.

Claims

1. A power system load component identification method based on sparse dictionary learning, characterized in that, Includes the following steps: S1) Collect historical load data and historical meteorological data of various types of loads, and obtain standardized historical load data and standardized historical meteorological data of various types of loads through preprocessing; S2) Establish a comprehensive historical load data dictionary and a comprehensive historical meteorological data dictionary based on standardized historical load data and standardized historical meteorological data of various types of loads; Step S2 includes the following steps: S2.1) Establish historical load matrices based on the standardized historical load data of various types of loads, and combine them to obtain a comprehensive historical load data dictionary; in the historical load matrix of each type of load, the column vectors are the load data collected by all sampling points on the same sampling day, and the sampling points are the sampling time points; S2.2) For each type of load, a historical meteorological matrix is ​​obtained by performing correlation analysis on standardized historical load data and standardized historical meteorological data; the historical meteorological matrices of all loads are combined to obtain a comprehensive historical meteorological data dictionary; Step S2.2 specifically includes: S2.2.1) For each type of load, standardize the resolution of the historical load data and the historical meteorological data; S2.2.2) For each type of load, the Spearman correlation coefficient between the load data and each type of meteorological data is obtained by performing correlation analysis on the load data and multiple meteorological data for each sampling day. S2.2.3) For each type of load, the weighted historical meteorological data for each sampling point on the sampling day is obtained according to the following formula, based on the multiple meteorological data for each sampling day and the Spearman correlation coefficient between the load data and each type of meteorological data: ； ； In the formula, The weighted historical meteorological data of the nth type of load at the i-th sampling point on the d-th day; For the nth type of load, the meteorological data of the kth type is collected at the i-th sampling point on the d-th day, where k=1~ , The total number of meteorological types for the nth type of load. The weighted calculation function for the nth type of load is... Let be the average correlation coefficient between the load data of type n and the meteorological data of type k. Let be the correlation coefficient between the load data of type n on day d and the meteorological data of type k; The total number of sampling days for the nth type of load; S2.2.4) For each type of load, combine the weighted historical meteorological data of all sampling points on each sampling day to obtain a unified historical meteorological vector for each day. Use the unified historical meteorological vector for each day as a column vector to construct the historical meteorological matrix of the load. S3) Obtain the load data and meteorological data of the day to be identified. Based on the comprehensive historical load data dictionary and comprehensive historical meteorological data dictionary obtained in step S2, use the orthogonal matching pursuit algorithm combined with meteorological factors to identify the load components of the load data of the day to be identified, and obtain the load component identification results. The load component identification results are used for load forecasting or scheduling of the power system.

2. The power system load component identification method based on sparse dictionary learning according to claim 1, characterized in that: The various types of loads mentioned are selected from photovoltaic, wind power, air conditioning loads, small hydropower, charging piles, and base loads.

3. The power system load component identification method based on sparse dictionary learning according to claim 1, characterized in that: The historical meteorological matrix for each type of load constructed in step S2.2.4 is as follows: ； In the formula, The historical meteorological matrix of the nth type of load, This represents the weighted historical meteorological data of the nth type of load at the i-th sampling point on day d, where d is the sampling day number, d=1~ , Let i be the total number of sampling days for the nth type of load, and i be the index of the sampling point, i=1~M. t M t This represents the total number of sampling points for each sampling day after unifying the resolution.

4. The power system load component identification method based on sparse dictionary learning according to claim 1, characterized in that: Step S3 specifically involves: S3.1) Using the daily unified historical load vector in the comprehensive historical load data dictionary as an element, and taking the minimum number of load data dictionary elements and accurate representation of the total load as the solution objective, a load component identification model is established based on the comprehensive historical load data dictionary. S3.2) Obtain the load data of the day to be identified and the meteorological data of various types of loads on the day to be identified, and construct the load vector and meteorological vector of the day to be identified respectively; use the orthogonal matching pursuit algorithm combined with meteorological factors to perform load component identification on the load vector of the day to be identified according to the load component identification model and the meteorological vector of the day to be identified, and obtain the load component identification result.

5. The power system load component identification method based on sparse dictionary learning according to claim 4, characterized in that: In step S3.1, the load component identification model is specifically as follows: ； In the formula, For the daily load vector to be identified, Let A be the component proportion vector, and A be the comprehensive historical load data dictionary. For the error vector, component proportion vector The number of non-zero elements, where L is the preset upper limit for selecting elements from the comprehensive historical load data dictionary. Let be the 2-norm of the error vector. As constraints, The value of the component proportion vector X is determined so that the 2-norm of the error vector E reaches its minimum. Among them, the component proportion vector Specifically: ； In the formula, To form a matrix representing the proportion of elements of the nth type of load in the comprehensive historical load data dictionary, n = 1 to N, where N is the total number of load types. The total number of sampling days for the nth type of load; Wherein, error vector Specifically: ； In the formula, Let be the error of the i-th sampling point, i = 1 to M, where M is the total number of sampling points for the day to be identified.

6. The power system load component identification method based on sparse dictionary learning according to claim 4, characterized in that: Step S3.2 specifically includes: S3.2.1) Collect load data for the day to be identified and construct a load vector for the day to be identified. Collect meteorological data of the location of each type of load on the day to be identified to obtain meteorological data of each type of load on the day to be identified. Construct a meteorological vector of the day to be identified from the meteorological data of each type of load using a weighted calculation function. S3.2.2) Take the inner product of the comprehensive historical meteorological data dictionary and the meteorological vector of the day to be identified to obtain the meteorological inner product matrix; at the same time, initialize the index list; S3.2.3) Take the inner product of the comprehensive historical load data dictionary and the daily load vector to be identified to obtain the load inner product matrix. Take the inner product of the load inner product matrix and the meteorological inner product matrix obtained in step S3.2.2 to obtain the comprehensive inner product matrix. From the comprehensive inner product matrix, select the index corresponding to the maximum value of other elements except the elements corresponding to the indexes in the index list and add them to the index list. S3.2.4) Select elements from the comprehensive historical load data dictionary according to the index list to construct an index dictionary matrix. Input the index dictionary matrix as the comprehensive historical load data dictionary into the load component identification model and obtain the component proportion vector and residual vector by the least squares method. S3.2.5) Using the residual vector as the daily load vector to be identified, new component proportion vectors and residual vectors are obtained according to steps S3.2.3 and S3.2.4; S3.2.6) Repeat step S3.2.5 until the preset number of iterations or the residual converges, to obtain the final index list and component proportion vector. Superimpose the elements of the same type in the index list according to the component proportion vector to obtain the load component identification result.

7. The power system load component identification method based on sparse dictionary learning according to claim 6, characterized in that: In step S3.2.3, the inner product matrix of the load inner product matrix and the meteorological inner product matrix are taken according to the following formula to obtain the comprehensive inner product matrix: ； In the formula, Let be the comprehensive inner product matrix obtained in the j-th iteration. Let be the load inner product matrix obtained in the j-th iteration. For meteorological inner product matrix, It represents the Hadamardi (or Hadama) stack.

8. The power system load component identification method based on sparse dictionary learning according to claim 6, characterized in that: In step S3.2.1, the weighted calculation function for each type of load is used according to the following formula. The meteorological data of the day to be identified for the load is used to construct a meteorological vector for the day to be identified: ； ； In the formula, The weighted meteorological data for the i-th sampling point of the n-th load type are as follows: For the raw meteorological data of the i-th sampling point of the k-th meteorological data type to be identified, Let be the daily meteorological vector to be identified for the nth type of load.

Images