Capacity-regulating transformer configuration method fusing load clustering and evolution analysis

Abstract

The invention provides a capacity-regulating transformer configuration method fusing load clustering and evolutionary analysis, which comprises the following steps of: extracting a corresponding annual maximum net load # imgabs1 # to obtain initial typical daily load data Sc by acquiring original net load data # imgabs0 # of an ith data point in a year of a low-voltage outlet side of a transformer in a power distribution area and corresponding monitoring time acquired at an interval of T hours; t and the number of days Dc of each cluster every year; setting typical daily net load evolution data Sy, c, t in n years when photovoltaic is not added; solving typical daily net load evolution data SPVy, c, t of sunny days in n years when photovoltaic devices are additionally arranged; calculating the clustering loss correction percentage PM of the capacity regulating transformer; constructing a configuration evaluation model F (Sy, c, t, Dc) of the capacity-regulating transformer in the n-year period when the photovoltaic is not additionally arranged; constructing an improved configuration evaluation model FPV (Sy, c, t, SPVy, c, t, Dc, DSc) of the capacity-regulating transformer in the n years when photovoltaic devices are additionally arranged; obtaining a capacity regulating transformer configuration scheme; the method can improve the accuracy and flexibility of the long-term configuration scheme of the capacity-regulating transformer, reduces resource waste caused by improper configuration of the transformer, and helps a power distribution area to achieve the purposes of energy conservation and loss reduction.

Description

Technical Field

[0001] The present invention relates to a method for configuring a capacity - regulating transformer by integrating load clustering and evolution analysis, and belongs to the technical field of distribution transformer configuration. Background Technique

[0002] With the extensive and large - scale access of distributed photovoltaics to distribution substations, there are even reverse overload problems when the load factor is low. Coupled with the obvious seasonal fluctuations of the electricity load, the peak - valley difference of the electricity load has increased significantly, resulting in the transformer working under light load or heavy load for a long time, reducing the operation safety and reliability of the transformer. At the same time, it will also reduce the economy of the distribution substation. In many existing distribution substations, due to the early planning of the distribution substation not considering the disordered access of distributed photovoltaics and the load growth rate, the applicability of the existing transformer configuration scheme is extremely poor, and this problem needs to be solved urgently.

[0003] Analysis shows that ordinary distribution transformers with a fixed - capacity mode often cannot balance the stability and economy of the operation of the distribution substation, while the capacity - regulating transformer can switch the capacity state in real time according to the load change, thus improving the operation safety of the transformer and reducing losses. Therefore, reasonably designing the configuration scheme of the capacity - regulating transformer is one of the key means to ensure the efficient and stable operation of the distribution network.

[0004] At present, the fixed - capacity distribution transformers in existing distribution substations are upgraded to capacity - regulating transformers, but there are few existing configuration and evaluation schemes for capacity - regulating transformers. In previous studies, historical load data was usually used to calculate the annual losses and costs, which were used as the main basis for configuring capacity - regulating transformers, while ignoring the load changes and photovoltaic increments in the development plan of the distribution substation. For example, in the literature: Su Yu, Wang Qianggang, Lei Chao, etc. Planning Method of On - Load Tap - Changing Distribution Transformers in Urban Distribution Networks under Electric Energy Substitution [J]. Transactions of China Electrotechnical Society, 2019, 34(07): 1496 - 1504, a capacity - regulating transformer with a single large - to - small capacity ratio of 3:1 was used, and the substation losses were calculated only based on the annual historical data, ignoring the periodic change characteristics of the load, and at the same time not considering the impact of load growth on the transformer capacity selection and the impact of photovoltaic access on the substation, thus reducing the accuracy of the configuration results. Summary of the Invention

[0005] To fully consider the impact of the load evolution result and photovoltaic access increment in the distribution substation on the distribution substation, the purpose of the present invention is to provide a method for configuring a capacity - regulating transformer by integrating load clustering and evolution analysis, so as to solve the problem that the accuracy and flexibility of the long - term planning and configuration scheme of the capacity - regulating transformer in the existing technology need to be improved.

[0006] The technical solution of the present invention is:

[0007] A method for configuring a capacity - regulating transformer by integrating load clustering and evolution analysis includes the following steps:

[0008] S1. Collect the original net load data of the i - th data point at the low - voltage outlet side of the transformer in the distribution sub - station area within a years, and the monitoring time collected at intervals of T hours, and extract the corresponding annual maximum net load.

[0009] S2. Combining the linear relationship between loss calculation and the square value of the net load, after performing square operation on the original net load data in step S1, cluster according to the number of clustering clusters k using a clustering algorithm to obtain the initial typical - day load data S c,t and the number of days D of each cluster in each year, c where c is the category number of the initial typical - day load data, c = 1, 2, 3,..., k, and t is the number of data points within a day.

[0010] S3. Based on the annual average load growth rate r of the sub - station area, set the typical - day net load evolution data S y,c,t for n years without adding photovoltaic power, where y is the year number within n years.

[0011] S4. Combining the expected access capacity of photovoltaic power in the sub - station area and the economic demand with the minimum line loss P, loss optimize and solve the typical - day net load evolution data SPV y,c,t for sunny days within n years when adding photovoltaic power. y,c,t ;

[0012] S5. Calculate the clustering loss correction percentage P M ;

[0013] S6. Based on the number of days D of each cluster in each year in step S2 c and the typical - day net load evolution data S y,c,t in step S3, introduce the clustering loss correction percentage P M in step S5 to construct an evaluation model F(S y,c,t , D c ) for configuring the capacity - regulating transformer without adding photovoltaic power within n years;

[0014] S7. Introduce the number of sunny days DS c for each cluster, the typical - day net load evolution data SPV y,c,t for sunny days within n years when adding photovoltaic power in step S4, to construct an improved evaluation model FPV(S y,c,t , SPV y,c,t , D c , DS c ) for configuring the capacity - regulating transformer when adding photovoltaic power within n years;

[0015] S8. Select an applicable evaluation model from Steps S6 and S7 according to the substation area planning requirements to obtain a configuration plan for capacity-adjustable transformers.

[0016] Further, in Step S2, the clustering algorithm is selected as follows: After selecting the number of clustering clusters k according to the elbow method using the K-Means clustering algorithm and the Fuzzy C-Means clustering algorithm, perform clustering separately; by comparing the performance of the clustering results on three indicators, namely the silhouette coefficient SC, the Davies-Bouldin index DBI, and the Calinski-Harabasz index CH, select the clustering algorithm with the largest number of dominant indicators for clustering.

[0017] Further, in Step S3, set the typical daily net load evolution data S for an n-year period without adding photovoltaic y,c,t : S y,c,t = S c,t (1 + r) y , where S c,t is the initial typical daily load data, and r is the average annual load growth rate of the substation area.

[0018] Further, in Step S4, based on the typical daily net load evolution data S y,c,t optimally solve the typical daily net load evolution data SPV for sunny days in an n-year period when adding photovoltaic y,c,t , specifically,

[0019] S41. Use the substation area topology, line impedance, expected photovoltaic access capacity, and typical daily net load evolution data S y,c,t as the basic setting parameters;

[0020] S42. Set conditional constraints for the output of static var generators and photovoltaic power generation in the substation area, and set power flow constraints for node voltage, current, and power balance, with the minimum substation area line loss P loss as the optimization goal:

[0021]

[0022] where I ij,t is the current of branch ij corresponding to data point t, r ij is the resistance of branch ij, and E is the set of substation area branches;

[0023] S43. Use a planning algorithm to solve the typical daily net load evolution data SPV for sunny days in an n-year period when adding photovoltaic to the low-voltage outlet side of the transformer for the basic setting parameters in Step S41, the constraint settings in Step S42, and the optimization goal. y,c,t .

[0024] Further, in step S5, calculate the clustering loss correction percentage P of the tap-changing transformer M :

[0025]

[0026] where k is the number of clustering clusters, ΔP1 is the loss of the tap-changing transformer, is the original net load data, S c,t is the initial typical daily load data, D c is the number of days in each cluster per year.

[0027] Further, in step S6, construct the evaluation model F(S y,c,t , D c ) for the tap-changing transformer configuration over an n-year period without adding PV, specifically,

[0028] S61. Introduce the clustering loss correction percentage P M of the tap-changing transformer, and establish the loss electrical energy ΔW0(S y,c,t , D c ) of the ordinary transformer and the loss electrical energy ΔW1(S y,c,t , D c ) of the tap-changing transformer in the y-th year without adding PV:

[0029]

[0030] where ΔP0 is the loss of the ordinary transformer, S y,c,t is the typical daily net load evolution data, D c is the number of days in each cluster per year;

[0031] S62. Construct the evaluation model F(S y,c,t , D c ) for the tap-changing transformer configuration over an n-year period without adding PV:

[0032]

[0033] where e is the electricity price, C Z0 , C Z1 are the cost expenses of the ordinary transformer and the tap-changing transformer respectively.

[0034] Further, in step S7, construct the improved evaluation model FPV(S y,c,t , SPV y,c,t , D c , DS c ) for the tap-changing transformer configuration over an n-year period with added PV, specifically,

[0035] S71. Establish the loss electrical energy ΔW0(S y,c,t, SPV y,c,t , D c , DS c ), and the power loss ΔW1 of the tap-changing transformer (S y,c,t , SPV y,c,t , D c , DS c ):

[0036]

[0037] Among them, S y,c,t is the daily net load evolution data of a typical day, P M is the clustering loss correction percentage, SPV y,c,t is the daily net load evolution data of a typical sunny day in the n-year period when photovoltaic is added, D c is the number of days in each cluster per year, DS c is the number of sunny days in each cluster, ΔP0 is the loss of the ordinary transformer, and ΔP1 is the loss of the tap-changing transformer;

[0038] S72. Construct an evaluation model FPV(S of the improved configuration of the tap-changing transformer in the n-year period when photovoltaic is added y,c,t , SPV y,c,t , D c , DS c ):

[0039]

[0040] Among them, e is the electricity price, C Z0 , C Z1 are the cost expenses of the ordinary transformer and the tap-changing transformer respectively.

[0041] Furthermore, step S8 is specifically

[0042] S81. When photovoltaic is not added in the substation area planning, select the tap-changing transformer configuration evaluation model F(S y,c,t , D c ) in step S6; when photovoltaic is added in the substation area planning, select the improved tap-changing transformer configuration evaluation model FPV(S y,c,t , SPV y,c,t , D c , DS c );

[0043] S82. Obtain the tap-changing transformer configuration plan: In the tap-changing transformer configuration evaluation model F(S y,c,t , D c ) ≥ 0 or the improved tap-changing transformer configuration evaluation model FPV(S y,c,t , SPV y,c,t , D c , DS c) ≥ 0, the substation area is suitable for configuring a capacity - regulating transformer; in the capacity - regulating transformer configuration evaluation model F(S y,c,t , D c ) < 0 or the improved configuration evaluation model FPV(S y,c,t , SPV y,c,t , D c , DS c ) < 0, the substation area is not suitable for configuring a capacity - regulating transformer and is suitable for configuring an ordinary transformer.

[0044] The beneficial effects of the present invention are as follows:

[0045] First, this method for configuring a capacity - regulating transformer by integrating load clustering and evolution analysis, by introducing the clustering loss correction percentage, fully considers the net load growth in the substation area and the load evolution scenarios when photovoltaic is added or not, and also considers the continuous incremental photovoltaic access in the distribution substation area. Two capacity - regulating transformer configuration evaluation models are respectively constructed, which can improve the accuracy and flexibility of the long - term configuration plan of the capacity - regulating transformer, perfect the theoretical system of capacity - regulating transformer configuration evaluation, reduce the waste of resources caused by improper transformer configuration, and help the distribution substation area achieve the purpose of energy conservation and loss reduction.

[0046] Second, this method for configuring a capacity - regulating transformer by integrating load clustering and evolution analysis, different from the previous method of directly calculating losses from historical load data without in - depth analysis, uses the K - Means clustering algorithm and the Fuzzy C - Means clustering algorithm to cluster respectively and then comprehensively compares and analyzes multiple evaluation indicators. It can obtain relatively accurate typical - day classification results while ensuring the effectiveness of the clustering results, thus improving the accuracy of loss calculation and evaluation of the capacity - regulating transformer.

[0047] Third, this method for configuring a capacity - regulating transformer by integrating load clustering and evolution analysis, combined with the linear relationship between loss calculation and the square value of the net load, first performs a square operation on the collected load data and then clusters. It can convert the two - way uncertainty of the loss calculation errors of ordinary transformers and capacity - regulating transformers obtained by directly clustering the load data into the one - way uncertainty of the loss calculation error of the capacity - regulating transformer. At this time, since the correctness of the loss calculation result of the ordinary transformer is greatly improved, it can be used as a comparison reference value for capacity - regulating transformer configuration evaluation; at the same time, the clustering loss correction percentage is introduced, thereby improving the rationality and effectiveness of capacity - regulating transformer configuration. Description of the Drawings

[0048] Figure 1 is a flow schematic diagram of the method for configuring a capacity - regulating transformer by integrating load clustering and evolution analysis in an embodiment of the present invention;

[0049] Figure 2 is an explanatory schematic diagram for selecting the number of clustering clusters k according to the elbow method in the embodiment;

[0050] Figure 3 It is the evolution data SPV of the typical sunny day in the nth year when photovoltaic is added in the embodiment y,c,t Schematic explanatory diagram;

[0051] Figure 4 It is the schematic explanatory diagram for obtaining the clustering loss correction percentage P of the on-load tap-changing transformer in the embodiment M Schematic explanatory diagram;

[0052] Figure 5 It is the schematic explanatory diagram of the typical daily data S of the substation area in the embodiment, where c = 1, 2, 3, and t = 1, 2, 3,..., 96. c,t Specific implementation manners

[0053] The preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0054] The embodiment provides a method for configuring an on-load tap-changing transformer by integrating load clustering and evolution analysis, as Figure 1 , including the following steps,

[0055] S1. Collect the original net load data of the i-th data point in the a years at the low-voltage outlet side of the transformer in the distribution substation area and the monitoring time collected at intervals of T hours, and extract the corresponding annual maximum net load

[0056] In step S1, in one example, a ∈ {1, 2, 3}, i = 1, 2, 3,..., 8760·a / T, and T ∈ {0.25, 1}.

[0057] S2. Combining the linear relationship between loss calculation and the square value of the net load, after performing square operation processing on the original net load data in step S1 cluster according to the number of clustering clusters k by using the clustering algorithm to obtain the initial typical daily load data S c,t and the number of days D of each cluster every year c , where c is the category number of the initial typical daily load data, c = 1, 2, 3,..., k, and t is the data point number within the day; as the basis for the net load evolution data. In the embodiment, t = 1, 2, 3,..., 24 / T, and k ≤ 5.

[0058] In step S2, the clustering algorithm is selected as follows: After the number of clustering clusters k is selected by the elbow method using the K-Means clustering algorithm and the Fuzzy C-Means clustering algorithm, clustering is performed respectively; By comparing the performance of the clustering results on three indicators, namely the Silhouette Coefficient (SC), the Davies-Bouldin Index (DBI), and the Calinski-Harabasz Index (CH), the algorithm with the largest number of dominant indicators is selected for clustering. Among them, the number of clustering clusters k is selected according to the elbow method, as Figure 2 Specifically: Draw an elbow method graph with the number of clustering clusters as the abscissa and the sum of squared errors within the cluster as the ordinate, and select the number of clustering clusters k corresponding to the elbow inflection point.

[0059] S3. Based on the annual average load growth rate r of the substation area, set the typical daily net load evolution data S for an n-year period without adding photovoltaic power, y,c,t where y is the annual number within the n-year period, y = 1, 2, 3,..., n. In the embodiment, n ∈ {5, 10, 15, 20}.

[0060] In step S3, set the typical daily net load evolution data S for an n-year period without adding photovoltaic power, y,c,t : S y,c,t = S c,t (1 + r) y where S c,t is the initial typical daily load data, and r is the annual average load growth rate of the substation area.

[0061] S4. Combining the expected access capacity of photovoltaic power in the substation area and the economic demand with the minimum line loss P loss Based on the typical daily net load evolution data S y,c,t Optimize and solve the typical daily net load evolution data SPV for sunny days in an n-year period when adding photovoltaic power y,c,t such as Figure 3 .

[0062] In step S4, based on the typical daily net load evolution data S y,c,t Optimize and solve the typical daily net load evolution data SPV for sunny days in an n-year period when adding photovoltaic power y,c,t Specifically,

[0063] S41. Take the substation area topology, line impedance, expected access capacity of photovoltaic power, and typical daily net load evolution data S y,c,t as the basic setting parameters;

[0064] S42. Set conditional constraints for the output of the static var generator and the output of the photovoltaic power in the transformer area, and set power flow constraints for the node voltage, current, and power balance, with the minimum of the transformer area line loss P loss as the optimization goal:

[0065]

[0066] Among them, I ij,t is the current of branch ij corresponding to data point t, and r ij is the resistance of branch ij, and E is the set of transformer area branches;

[0067] S43. Solve the n-year evolution data SP V of the net load on a typical sunny day when a photovoltaic is added to the low-voltage outlet side of the transformer by using a programming algorithm for the basic setting parameters in step S41, the constraint settings in step S42, and the optimization goal y,c,t .

[0068] In step S43, considering the problem of avoiding falling into local optimality, a second-order cone programming algorithm is used for solving.

[0069] S5. Calculate the clustering loss correction percentage P M . Such as Figure 4 :

[0070] S51. Calculate the annual maximum net load constraint condition in the nth year Based on the annual maximum net load constraint condition in the nth year Determine the rated capacity S N of the candidate ordinary transformer and the large rated capacity S ND of the on-load tap-changing transformer: δ is the capacity margin percentage;

[0071] S52. Obtain the loss parameters including the total reactive power loss P Z00 and the total load loss P ZK0 of the ordinary transformer, the total reactive power loss P Z01 and the total load loss P ZK1 under the large rated capacity of the on-load tap-changing transformer, and the total reactive power loss P Z02 and the total load loss P ZK2 under the small rated capacity of the on-load tap-changing transformer;

[0072] S53. Calculate the critical tap-changing point S C of the on-load tap-changing transformer:

[0073]

[0074] Among them, S ND , S NX are the large rated capacity and the small rated capacity of the on-load tap-changing transformer respectively;

[0075] S54. Establish the loss model ΔP1(S) of the tap-changing transformer:

[0076]

[0077] Where ΔP1 is the loss of the tap-changing transformer and S is the load substituted.

[0078] S55. Calculate the clustering loss correction percentage P of the tap-changing transformer M :

[0079]

[0080] Where k is the number of clustering clusters, ΔP1 is the loss of the tap-changing transformer, is the original net load data, S c,t is the initial typical day load data, D c is the number of days in each cluster per year.

[0081] S6. Based on the number of days D in each cluster per year in step S2 c and the typical day net load evolution data S in step S3 y,c,t , introduce the clustering loss correction percentage P in step S5 M , and construct the n-year configuration evaluation model F(S y,c,t , D c ) of the tap-changing transformer without adding PV.

[0082] In step S6, construct the n-year configuration evaluation model F(S y,c,t , D c ) of the tap-changing transformer without adding PV. Specifically,

[0083] S61. Introduce the clustering loss correction percentage P of the tap-changing transformer M , and establish the loss electrical energy ΔW0(S y,c,t , D c ) of the ordinary transformer and the loss electrical energy ΔW1(S y,c,t , D c ) of the tap-changing transformer in the y-th year without adding PV:

[0084]

[0085] Where ΔP0 is the loss of the ordinary transformer, S y,c,t is the typical day net load evolution data, D c is the number of days in each cluster per year;

[0086] S62. Construct the n-year configuration evaluation model F(S y,c,t , D c):

[0087]

[0088] Among them, e is the electricity price, and C Z0 and C Z1 are the cost expenses of the ordinary transformer and the capacity - regulating transformer respectively.

[0089] S7. Introduce the number of sunny days DS in each cluster c , and the evolution data SPV of the net load on typical sunny days in the n - year period when photovoltaic is added in step S4 y,c,t , and construct an improved configuration evaluation model FPV(S y,c,t , SPV y,c,t , D c , DS c ) for the capacity - regulating transformer in the n - year period when photovoltaic is added.

[0090] In step S7, construct an improved configuration evaluation model FPV(S y,c,t , SPV y,c,t , D c , DS c ), specifically:

[0091] S71. Establish the loss electric energy ΔW0(S y,c,t , SPV y,c,t , D c , DS c ) of the ordinary transformer and the loss electric energy ΔW1(S y,c,t , SPV y,c,t , D c , DS c ) of the capacity - regulating transformer:

[0092]

[0093]

[0094] Among them, S y,c,t is the evolution data of the net load on typical days, P M is the clustering loss correction percentage, SPV y,c,t is the evolution data of the net load on typical sunny days in the n - year period when photovoltaic is added, D c is the number of days in each cluster per year, DS c is the number of sunny days in each cluster, ΔP0 is the loss of the ordinary transformer, and ΔP1 is the loss of the capacity - regulating transformer;

[0095] S72. Construct an improved configuration evaluation model FPV(S y,c,t , SPV y,c,t , Dc , DS c ):

[0096]

[0097] Among them, e is the electricity price, and C Z0 , C Z1 are the cost expenses of the ordinary transformer and the capacity-adjustable transformer respectively.

[0098] S8. According to the requirements of the substation area planning, select the applicable evaluation model from steps S6 and S7 to obtain the capacity-adjustable transformer configuration plan.

[0099] S81. When no photovoltaic power is added to the substation area planning, select the capacity-adjustable transformer configuration evaluation model F(S y,c,t , D c ) of step S6; when photovoltaic power is added to the substation area planning, select the improved capacity-adjustable transformer configuration evaluation model FPV(S y,c,t , SPV y,c,t , D c , DS c );

[0100] S82. Obtain the capacity-adjustable transformer configuration plan: when the capacity-adjustable transformer configuration evaluation model F(S y,c,t , D c ) ≥ 0 or the improved capacity-adjustable transformer configuration evaluation model FPV(S y,c,t , SPV y,c,t , D c , DS c ) ≥ 0, the substation area is suitable for configuring a capacity-adjustable transformer; when the capacity-adjustable transformer configuration evaluation model F(S y,c,t , D c ) < 0 or the improved capacity-adjustable transformer configuration evaluation model FPV(S y,c,t , SPV y,c,t , D c , DS c ) < 0, the substation area is not suitable for configuring a capacity-adjustable transformer and is suitable for configuring an ordinary transformer.

[0101] The method for configuring a capacity - regulating transformer by integrating load clustering and evolution analysis includes collecting the original data at the outlet side of the transformer in the sub - station area; performing square - operation processing on the original net - load data and then clustering to obtain the initial typical - day load data and the number of days in each cluster per year, which serve as the basis for the evolution data of the net load; setting the typical - day load evolution data for an n - year period without adding photovoltaic power; optimizing and solving the typical - day load evolution data for sunny days in an n - year period with added photovoltaic power; calculating the clustering - loss correction percentage of the capacity - regulating transformer; constructing an evaluation model for the configuration of the capacity - regulating transformer for an n - year period without adding photovoltaic power; constructing an improved evaluation model for the configuration of the capacity - regulating transformer for an n - year period with added photovoltaic power; and selecting an evaluation model according to the sub - station area planning requirements to obtain the configuration scheme of the capacity - regulating transformer. By introducing the clustering - loss correction percentage and considering the continuous incremental photovoltaic access in the distribution sub - station area, this method improves the accuracy and flexibility of the long - term configuration scheme of the capacity - regulating transformer and perfects the theoretical system for the evaluation of the capacity - regulating transformer configuration.

[0102] The method for configuring a capacity - regulating transformer by integrating load clustering and evolution analysis fully considers the growth of the net load in the sub - station area and the load evolution scenarios with or without adding photovoltaic power. Two evaluation models for the configuration of the capacity - regulating transformer are constructed respectively, which perfects the theoretical system for the evaluation of the capacity - regulating transformer configuration, reduces the waste of resources caused by improper transformer configuration, and helps the distribution sub - station area achieve the goal of energy conservation and loss reduction.

[0103] Different from the previous method of directly calculating losses from historical load data without in - depth analysis, the method for configuring a capacity - regulating transformer by integrating load clustering and evolution analysis uses the K - Means clustering algorithm and the Fuzzy C - Means clustering algorithm to cluster respectively and then comprehensively compares multiple evaluation indicators. It can obtain a relatively accurate typical - day classification result while ensuring the effectiveness of the clustering result, thus improving the accuracy of the loss calculation and evaluation of the capacity - regulating transformer.

[0104] The evaluation model and method for the configuration of the capacity - regulating transformer by integrating load clustering and evolution analysis, combined with the linear relationship between loss calculation and the square value of the net load, perform square - operation on the collected load data first and then cluster. It can convert the two - way uncertainty of the loss calculation errors of ordinary transformers and capacity - regulating transformers obtained by directly clustering the load data into the one - way uncertainty of the loss calculation error of the capacity - regulating transformer. At this time, since the correctness of the loss calculation result of the ordinary transformer is greatly improved, it can be used as a comparison reference value for the evaluation of the capacity - regulating transformer configuration; at the same time, the clustering - loss correction percentage is introduced, thus improving the rationality and effectiveness of the configuration model and method.

[0105] The following is an exemplary description of the method for configuring a capacity - regulating transformer by integrating load clustering and evolution analysis of the embodiment with specific experimental data as follows:

[0106] This experiment selects the typical day S of the sub - station areac,t , where c = 1, 2, 3, t = 1, 2, 3,..., 96, as Figure 5 shown, the corresponding number of days D1 = 183, D2 = 92, D3 = 90. Considering the 10-year development plan of the distribution area, according to the annual load growth rate r = 2% and the capacity margin percentage δ = 10% of the distribution area, select a 315 kVA ordinary transformer, a 315(100) kVA adjustable capacity transformer with a large-to-small capacity ratio of 3:1, and a 315(80) kVA adjustable capacity transformer with a large-to-small capacity ratio of 4:1 as the transformers to be selected, and evaluate the configuration of the adjustable capacity transformer when considering two scenarios: no additional photovoltaic installation and 60 kW photovoltaic installation in the distribution area. The configuration evaluation results without additional photovoltaic installation are shown in Table 1:

[0107] Table 1 Configuration evaluation results without additional photovoltaic installation

[0108] Transformer capacity (kVA) <![CDATA[F(S 0,y,c,t ,D c )]]> Configuration plan 315、315(100) -10135.32 315 kVA ordinary transformer 315、315(80) -455.3025 315 kVA ordinary transformer

[0109] It can be seen from Table 1 that under the 10-year plan of this distribution area, without additional photovoltaic installation, it is more suitable to configure a 315 kVA ordinary transformer, which saves 10,135.3 yuan compared with the 315(100) kVA adjustable capacity transformer and 455.3 yuan compared with the 315(80) kVA transformer.

[0110] When 60 kW of photovoltaic is added to the distribution area, DS1 = 82, DS2 = 46, DS3 = 40, and the configuration evaluation results are shown in Table 2:

[0111] Table 2 Configuration evaluation results when 60 kW of photovoltaic is added to the distribution area

[0112] Transformer capacity (kVA) <![CDATA[FPV(S y,c,t ,SPV y,c,t ,D c ,DS c )]]> Configuration plan 315、315(100) -8235.06 315 kVA ordinary transformer 315、315(80) 1418.95 315(80) kVA capacity-adjustable transformer

[0113] It can be seen from Table 2 that when 60 kW of photovoltaic is added to this distribution area, it is more suitable to configure a 315(80) kVA 4:1 adjustable capacity transformer. By comparing the configuration results of the two scenarios, it can be obtained that the adjustable capacity transformer configuration method of this kind of integrated load clustering and evolution analysis in the embodiment considers the access scenario of distributed photovoltaic. The selected 315(80) kVA 4:1 adjustable capacity transformer can effectively save 1,418.95 yuan compared with the 315 kVA ordinary transformer configured in the single scenario without additional photovoltaic installation, and reduce the waste of resources caused by improper configuration.

[0114] The above is only the implementation manner of the present invention and is not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the scope of the claims of the present invention.

Claims

1. A method for configuring a capacity-adjustable transformer by integrating load clustering and evolutionary analysis, characterized in that: including the following steps, S1. Collect the original net load data of the i-th data point at the low-voltage outlet side of the distribution transformer in the distribution substation area within a years and the monitoring time collected at intervals of T hours, and extract the corresponding annual maximum net load S2. Combine the linear relationship between loss calculation and the square value of the net load, and perform a square operation on the original net load data in step S1 After that, use the clustering algorithm to cluster according to the number of clustering clusters k to obtain the initial typical daily load data S c,t and the number of days D in each cluster every year c , where c is the category number of the initial typical daily load data, c = 1, 2, 3,..., k, and t is the intra-day data point number; S3. Based on the annual average load growth rate r of the substation area, set the typical daily net load evolution data S for n years without adding photovoltaic y,c,t , where y is the annual number within n years; S4. Combine the expected access capacity of rooftop PV and line loss P loss with the minimum economic demand, and based on the net load evolution data S of a typical day y,c,t optimally solve for the net load evolution data SPV of a sunny typical day over an n-year period when adding rooftop PV y,c,t ; S5. Calculate the clustering loss correction percentage P of the on-load tap-changing transformer M ; S6. Based on the number of days D in each cluster per year in step S2 c and the typical daily net load evolution data S in step S3 y,c,t , introduce the clustering loss correction percentage P in step S5 M , and construct an evaluation model F(S y,c,t , D c ) for the configuration of on-load tap-changing transformers over an n-year period without adding photovoltaic power S7. Introduce the number of sunny days DS per cluster c In step S4, add the evolution data SPV of the net load on typical sunny days in the n-year period when photovoltaic is installed y,c,t , and construct an improved configuration evaluation model FPV(S y,c,t , SPV y,c,t , D c , DS c ) for the capacity-adjustable transformer when photovoltaic is installed S8. According to the distribution area planning requirements, select an applicable evaluation model from steps S6 and S7 to obtain a capacity-adjustable transformer configuration plan.

2. The method for configuring a capacity-adjustable transformer by fusing load clustering and evolutionary analysis according to claim 1, wherein: In step S2, the clustering algorithm is selected by the following steps: After using the K-Means clustering algorithm and the Fuzzy C-Means clustering algorithm to select the number of clustering clusters k according to the elbow method, perform clustering respectively; by comparing the performance of the clustering results on three indicators of the silhouette coefficient SC, the Davies-Bouldin index DBI, and the Calinski-Harabasz index CH, select the clustering algorithm with the largest number of dominant indicators for clustering.

3. The method for configuring a capacity-adjustable transformer by fusing load clustering and evolutionary analysis according to claim 1, characterized in that: In step S3, the typical daily net load evolution data S for an n-year period without adding PV is set y,c,t : S y,c,t = S c,t (1 + r) y , where S c,t is the initial typical daily load data, and r is the annual average load growth rate of the substation area 4. The method for configuring a capacity-adjustable transformer by integrating load clustering and evolutionary analysis according to any one of claims 1-3, characterized in that: In step S4, based on the typical daily net load evolution data S y,c,t Optimize and solve the typical daily net load evolution data SPV of sunny days in the n-year period when photovoltaic is added y,c,t , specifically, S41. Use the substation area topology, line impedance, expected PV access capacity, and typical daily net load evolution data S y,c,t as the basic set parameters; S42. Set conditional constraints for the output of the static var generator and the output of the photovoltaic power in the substation area, and set power flow constraints for the node voltage, current, and power balance, with the minimum of the substation area line loss P loss as the optimization goal: Among them, I ij,t is the current of branch ij corresponding to data point t, and r ij is the resistance of branch ij, and E is the set of branch circuits in the substation area; S43. Solve the basic setting parameters in step S41, the constraint setting in step S42, and the optimization objective by using a programming algorithm to obtain the n-year typical sunny day net load evolution data SPV when a photovoltaic system is added to the low-voltage outlet side of the transformer y,c,t .

5. The method for configuring a tap-changing transformer by integrating load clustering and evolutionary analysis according to any one of claims 1-3, characterized in that: In step S5, calculate the clustering loss correction percentage P of the on-load tap-changing transformer M : Among them, k is the number of clustering clusters, and ΔP1 is the loss of the tap-changing transformer. is the original net load data, S c,t is the initial typical daily load data, D c is the number of days in each cluster per year.

6. The method for configuring a capacity-adjustable transformer by integrating load clustering and evolutionary analysis according to claim 5, wherein: In step S6, an evaluation model F(S y,c,t , D c ) for the configuration of on-load tap-changer over an n-year period without adding PV is constructed. Specifically, S61. Introduce the clustering loss correction percentage P of the tap-changing transformer M , and establish the power loss ΔW0(S y,c,t , D c ) of the ordinary transformer in the y-th year without adding PV, and the power loss ΔW1(S y,c,t , D c ) of the tap-changing transformer: Among them, ΔP0 is the loss of the ordinary transformer, S y,c,t is the daily net load evolution data of the typical day, D c is the number of days in each cluster per year; S62. Build an evaluation model for the configuration of on-load tap-changing transformers with a term of n years without adding photovoltaic power generation, F(S y,c,t , D c ): Among them, e is the electricity price, C Z0 and C Z1 are the cost expenses of the ordinary transformer and the capacity - regulating transformer respectively.

7. The method for configuring a capacity-adjustable transformer by fusing load clustering and evolution analysis according to claim 5, wherein: In step S7, an evaluation model FPV(S y,c,t , SPV y,c,t , D c , DS c ) for the improved configuration of the tap-changing transformer with the addition of photovoltaic power generation over an n-year period is constructed. Specifically, S71. Establish the loss electrical energy ΔW0(S y,c,t , SPV y,c,t , D c , DS c ) of the ordinary transformer in the yth year when adding photovoltaic power, and the loss electrical energy ΔW1(S y,c,t , SPV y,c,t , D c , DS c ): Among them, S y,c,t is the typical daily net load evolution data, P M is the clustering loss correction percentage, SPV y,c,t is the typical daily net load evolution data of sunny days in the n-year period when photovoltaic is added, D c is the number of days in each cluster per year, DS c is the number of sunny days in each cluster, ΔP0 is the loss of the ordinary transformer, and ΔP1 is the loss of the capacity-adjustable transformer; S72. Construct an improved configuration evaluation model FPV(S y,c,t , SPV y,c,t , D c , DS c ) for the capacity-adjustable transformer in the n-year period when adding photovoltaic Among them, e is the electricity price, and C Z0 , C Z1 are the cost expenses of the ordinary transformer and the capacity - regulating transformer respectively.

8. The method for configuring a capacity-adjustable transformer by fusing load clustering and evolution analysis according to any one of claims 1-3, characterized in that: Step S8 is specifically as follows S81. When no additional PV is added in the substation area planning, the evaluation model F(S y,c,t , D c ) of the tap-changing transformer configuration in step S6 is selected; when additional PV is added in the substation area planning, the improved evaluation model FPV(S y,c,t , SPV y,c,t , D c , DS c ) of the tap-changing transformer configuration in step S7 is selected; S82. Obtain the configuration scheme of the on-load tap-changing transformer: In the on-load tap-changing transformer configuration evaluation model F(S y,c,t , D c ) ≥ 0 or the improved on-load tap-changing transformer configuration evaluation model FPV(S y,c,t , SPV y,c,t , D c , DS c ) ≥ 0, the distribution transformer substation is suitable for configuring an on-load tap-changing transformer; in the on-load tap-changing transformer configuration evaluation model F(S y,c,t , D c ) < 0 or the improved on-load tap-changing transformer configuration evaluation model FPV(S y,c,t , SPV y,c,t , D c , DS c ) < 0, the distribution transformer substation is not suitable for configuring an on-load tap-changing transformer and is suitable for configuring an ordinary transformer.

Images