Low-voltage transformer area topology identification application method based on energy gateway

Abstract

The invention belongs to the technical field of power distribution area topology identification, and relates to a low-voltage area topology identification application method based on an energy gateway. The method comprises the following steps: classifying multi-zone-area coupling users and single-zone-area coupling users, and setting an energy gateway for data acquisition; screening the voltage data based on the phase voltage unbalance rate; calculating a Pearson's correlation coefficient and a Spearman correlation coefficient of the three-phase voltage; clustering all coupling users by using a K-means clustering algorithm; the clustering result is checked and screened according to the electric energy balance principle; and determining a final topological structure of the low-voltage transformer area. According to the method, more high-precision voltage data are obtained through the energy gateway, the voltage unbalance rate at the critical moment is screened, and the accuracy of topology identification of the low-voltage transformer area is improved; the reliability of an identification result is enhanced by comprehensively evaluating the similarity of the voltage sequences; and the calculation efficiency and the user classification accuracy are improved by optimizing the topology recognition result through the electric energy balance test.

Description

Technical Field

[0001] The present invention belongs to the technical field of distribution station area topology identification, and in particular, relates to an application method for low-voltage station area topology identification based on an energy gateway. Background Art

[0002] Due to the influence of various factors, the topological structure data of low-voltage substations often do not match the actual situation. The overall layout of the power system distribution network is relatively complex. The documents and materials related to the distribution network mainly rely on paper archives, resulting in the general lack of substation data. The low-voltage substation distribution network is large in scale, with high expansion and reconnection rates, and frequent changes in phase relationships and users, resulting in delayed updates of substation information. Accurately identifying the topological structure of the problem substation is crucial to the smooth implementation of various tasks in the low-voltage substation. It is the basis for mastering the operating status of the power system, improving power supply reliability, and realizing intelligent monitoring and refined management.

[0003] The traditional method of low-voltage substation topology identification mainly uses a topology identifier for identification. The topology identifier mainly uses power carrier signals and pulse current signals for identification. The topology identifier based on power carrier signals requires point-to-point communication, but when crossing substations, due to signal isolation and loss between the two transformers, the carrier communication performance will decrease, resulting in poor communication. The topology identifier based on pulse current signals needs to be used with flexible coils or current clamps. However, pulse current technology cannot yet achieve two-way communication and is still a point-to-point communication mode, resulting in low substation identification efficiency and high workload. In practical applications, the topology identifier also faces problems such as common high-voltage crosstalk and common ground crosstalk, which affect the accuracy of the identification results. In addition, the topology identifier can only be used for topology identification and cannot be applied to other scenarios. Its use cost is high and the equipment operation is complex, usually requiring multiple people to work together.

[0004] Traditional topology identifiers are limited by device connections and network complexity and have low accuracy. Traditional identifiers rely on offline analysis and manual intervention and cannot track in real time. Traditional instruments rely on fixed collection points and have limited coverage, which affects identification results. Traditional equipment requires dedicated hardware and frequent maintenance, which is costly. Summary of the invention

[0005] In order to solve the above technical problems, the present invention provides a low-voltage area topology identification application method based on an energy gateway, comprising:

[0006] Calculate the geographical distance between users based on their geographic information data, determine the user's geographical location, and classify multi-area coupled users and single-area coupled users. Set up an energy gateway at each coupled user and transformer node for data collection.

[0007] According to the classification results, the voltage data of the transformer in the production area at the time of three-phase voltage imbalance are screened, and the voltage data are screened based on the phase voltage imbalance rate to obtain the voltage data of several moments with the largest phase voltage imbalance rate;

[0008] Calculate the Pearson correlation coefficient and Spearman correlation coefficient between the voltage series of all coupled users in the substation and the three-phase voltage of the transformer in the corresponding substation, and interpolate the missing data;

[0009] All coupled users are clustered using the K-means clustering algorithm to obtain clustering results of coupled users in several stations;

[0010] Compare the clustering results and use the principle of power balance to test and screen the clustering results;

[0011] Calculate the average correlation coefficient between the in-phase coupled users in each area by group;

[0012] The final topological structure of the low-voltage area is determined based on the average correlation coefficient and voltage level.

[0013] Based on the above technical solution, the present invention can also be improved as follows.

[0014] Further, according to the classification results, the voltage data at the time when the three-phase voltage of the transformer in the substation is unbalanced is screened, and the voltage data is screened based on the phase voltage unbalance rate to obtain the voltage data at several times when the phase voltage unbalance rate is the largest, including:

[0015] According to the voltage data collected during the day, for low-voltage substations that only contain single-substation coupled users, several moments with the largest phase voltage imbalance rate are screened; for low-voltage substations that contain multiple substation coupled users, several common moments with the largest phase voltage imbalance rate are first screened. When the common moments are less than the set number, the moment with the largest phase voltage imbalance rate in the substation is selected based on the remaining required number of moments, and the voltage values ​​of all substation coupled users at the corresponding moments are derived as the input voltage sequence.

[0016] Furthermore, let η be the phase voltage unbalance rate, U A , U B and U C are the effective values ​​of the three-phase voltages, U avg is the average value of the effective value of the three-phase voltage, and the calculation formula of the phase voltage unbalance rate is:

[0017]

[0018] Furthermore, the Pearson correlation coefficient between the voltage series of all coupled users in the substation and the three-phase voltage of the transformer in the corresponding substation is calculated, including: the Pearson correlation coefficient is the product of the two voltage series and the voltage average deviation; let T and Y be two sequences, cov(T,Y) is the covariance of sequences T and Y, σ T is the standard deviation of the sequence T, σ Y is the standard deviation of the sequence Y, the Pearson correlation coefficient is rp, and the correlation coefficient between T and Y is ρ TY , replace the population covariance with the sample covariance, replace the population standard deviation with the sample standard deviation, T and Y are two normal populations, and the calculated Pearson correlation coefficient rp is used as the correlation coefficient ρ between T and Y TY is an unbiased and consistent estimator of , then:

[0019]

[0020] Furthermore, let the three-phase voltage sequence be U A , U B with U C , the single-phase user voltage sequence is U X , U A The elements of A(k), U B The elements of B(k), U C The element of is C(k), the element of the single-phase user voltage sequence is X(k), k is the discrete sampling point of the sequence, N is the total number of discrete sampling points of the sequence, rp XA For U A The Pearson correlation coefficient, rp XB For U B The Pearson correlation coefficient, rp XC For U C Pearson correlation coefficient, single-phase user voltage series U X With the three-phase voltage sequence U A , U B and U C The Pearson correlation coefficient calculation formula is:

[0021]

[0022] Furthermore, the Spearman correlation coefficient between the voltage sequence of all coupled users in the substation and the three-phase voltage of the transformer in the corresponding substation is calculated, including: Let X and Y be two sequences, each sequence has N elements, X i represents the i-th value in the sequence X, Y i Respectively represent the i-th value in sequence Y, 1≤i≤N. After arranging sequences X and Y in ascending or descending order, we get sets x and y. Let x i Represents X i Rank in sequence X, yi Represents Y i Rank in sequence Y, is the mean of the sequence X, is the mean of sequence Y, and the ranking difference set is set to d, which is defined as the difference between the corresponding elements in set x and set y, then d i =x i -y i , the Spearman correlation coefficient rs between the sequences X and Y is calculated by the set x and the set y or d, then:

[0023]

[0024] Furthermore, the K-means clustering algorithm is used to cluster all coupled users, and the clustering results of coupled users in several stations are obtained, including:

[0025] Classify the input samples into K clusters, select K elements as the initial centroids of the clusters, calculate the distances between the remaining input sample elements and all centroids, and classify the elements into the cluster to which the nearest centroid belongs, completing the first iteration;

[0026] Calculate the average value of all elements in the same cluster as the new centroid, and then recalculate the distance between all sample elements and the new centroid to complete the classification of the new cluster;

[0027] For the new cluster, repeatedly calculate the new centroid, calculate the distance between all sample elements and the new centroid, complete the classification of the new cluster, and continue iterating;

[0028] When there is no change in the classification of new elements or all clusters no longer change or the number of iterations reaches the set threshold, the iteration ends and the clustering result is obtained.

[0029] Furthermore, let R e is the radius of the earth, the longitude of transformer A is jA, the latitude of transformer A is ωA, the longitude and latitude of transformer A are (jA, ωA), the longitude of user B is jB, the latitude of user B is ωB, the longitude and latitude of user B are (jB, ωB), the distance between the input sample element and all the centroids, that is, the geographical distance between the user and the transformer in the substation is d AB ,but:

[0030] d AB =R e cos -1 [cosωA·cosωB·cos(jA-jB)+sinωA·sinωB].

[0031] The beneficial effects of the present invention are:

[0032] (1) The energy gateway-based method significantly improves the accuracy of low-voltage substation topology identification and reduces errors through real-time data collection and analysis. The energy gateway solution can collect electrical parameters in real time, dynamically adjust and update the topology structure, and ensure the continuous accuracy of identification. The energy gateway achieves extensive data collection through smart meters and wireless communications, making up for the insufficient coverage of traditional methods. The data collection method based on the energy gateway utilizes existing smart meters and communication equipment, reduces hardware investment, and reduces operation and maintenance costs through a remote platform.

[0033] (2) The present invention is based on the topology recognition of voltage correlation and energy gateway data, obtains more high-precision voltage data through the energy gateway, screens the voltage imbalance rate at critical moments, and improves the accuracy of low-voltage area topology recognition;

[0034] (3) The present invention combines the Pearson and Spearman correlation coefficients to comprehensively evaluate the voltage sequence similarity from the perspectives of linearity and monotonicity, thereby enhancing the reliability of the recognition results;

[0035] (4) The K-means clustering algorithm is used to optimize the topology recognition results through power balance inspection, and the positioning function is used to improve the computing efficiency and user classification accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 A schematic diagram of a low-voltage area topology identification application method based on an energy gateway provided in Example 1 of the present invention;

[0037] Figure 2 It is a simplified low voltage area topology tree;

[0038] Figure 3 It is a simulation model of the distribution station area;

[0039] Figure 4 To couple users PC5, PC10 with the three-phase voltage curve;

[0040] Figure 5 It is the correlation coefficient and scatter plot of coupling users PC5, PC10 and three-phase voltage;

[0041] Figure 6 After data screening, user PC5, PC10 and three-phase voltage curves are coupled;

[0042] Figure 7 The correlation coefficient and scatter plot of coupled users PC5, PC10 and three-phase voltage after data screening;

[0043] Figure 8 The correlation coefficient and scatter plot of user 5, user 10 and three-phase voltage after adding noise;

[0044] Fig. 9It is a schematic diagram of the multi-station area topology model;

[0045] Fig.10 This is the schematic diagram of the modified multi-station area topology. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0047] As an example, Figure 1 As shown, in order to solve the above technical problems, this embodiment provides a low-voltage area topology identification application method based on an energy gateway, including:

[0048] Calculate the geographical distance between users based on their geographic information data, determine the user's geographical location, and classify multi-area coupled users and single-area coupled users. Set up an energy gateway at each coupled user and transformer node for data collection.

[0049] According to the classification results, the voltage data of the transformer in the production area at the time of three-phase voltage imbalance are screened, and the voltage data are screened based on the phase voltage imbalance rate to obtain the voltage data of several moments with the largest phase voltage imbalance rate;

[0050] Calculate the Pearson correlation coefficient and Spearman correlation coefficient between the voltage series of all coupled users in the substation and the three-phase voltage of the transformer in the corresponding substation, and interpolate the missing data;

[0051] All coupled users are clustered using the K-means clustering algorithm to obtain clustering results of coupled users in several stations;

[0052] Compare the clustering results and use the principle of power balance to test and screen the clustering results;

[0053] Calculate the average correlation coefficient between the in-phase coupled users in each area by group;

[0054] The final topological structure of the low-voltage area is determined based on the average correlation coefficient and voltage level.

[0055] Specifically, the voltage characteristics of users are closely related to the topology and lines of the power grid. When the electrical distance between users is close, the voltage fluctuation law is relatively similar; however, due to the long electrical distance between users in different substations, the similarity of voltage fluctuation is low. In the same substation, if two users are on the same feeder and the electrical distance is short, their voltage fluctuation curves have the highest correlation. Therefore, the correlation of voltage distribution reflects the electrical distance between users to a certain extent, and the voltage level can indicate the upstream and downstream positions of users in the low-voltage substation. The upstream and downstream positions of users in the same level of feeder can be distinguished by the difference in voltage level. Within a short electrical distance, the voltage waveforms of users are similar and the voltage correlation is high. Generally, the best correlation indicates that users are located in the upstream and downstream parts of the same level of feeder. If the voltage correlation is low, the users do not belong to the same substation. The voltage value gradually decreases along the feeder, so the voltage of users closer to the transformer is higher.

[0056] As attached Figure 2 As shown in the figure, the simplified low-voltage substation topology tree includes two substations, TX1 and TX2, and the terminal users are M1 to M25. When multiple electrical users are within the set range, the voltage changes slightly, and they are usually in the same phase of the same substation, so they can be coupled as one user for processing. At this time, the topological identification of multiple users (such as PC1 to PC12) can be completed by judging the topological position of the coupled users. If the distance between adjacent users is far, each user is measured separately, and the coupled users represent a single user (such as PC8 and PC9). The energy gateway is placed at each coupled user and transformer node for data collection. Treating multiple users as coupled users can significantly reduce the amount of algorithm calculations and the number of energy gateways used, thereby reducing costs and workload.

[0057] According to the principle of voltage correlation, the voltage correlation between coupled users in the same area (such as PC1~PC6 and PC7~PC12) is high, while the voltage correlation between coupled users in different areas is low. The correlation between coupled users in the same feeder is better. For example, in the area TX1, the correlation between PC1~PC3 and PC4~PC6 is stronger. The voltage levels corresponding to coupled users PC1~PC12 are UPC1~UPC12, respectively. According to the upstream and downstream relationship of the line, Figure 2 The relationship between the voltage levels UPC1~UPC12 corresponding to the coupling users PC1~PC12 is:

[0058]

[0059] The voltage correlation principle determines the topology based on the similarity between voltage curves. Therefore, when the similarity of voltage curves is quite different, the topology identification method based on voltage correlation is more accurate.

[0060] Optionally, according to the classification result, the voltage data at the time when the three-phase voltage of the transformer in the stage area is unbalanced is screened, and the voltage data is screened based on the phase voltage unbalance rate to obtain the voltage data at several times when the phase voltage unbalance rate is the largest, including:

[0061] According to the voltage data collected during the day, for low-voltage substations that only contain single-substation coupled users, several moments with the largest phase voltage imbalance rate are screened; for low-voltage substations that contain multiple substation coupled users, several common moments with the largest phase voltage imbalance rate are first screened. When the common moments are less than the set number, the moment with the largest phase voltage imbalance rate in the substation is selected based on the remaining required number of moments, and the voltage values ​​of all substation coupled users at the corresponding moments are derived as the input voltage sequence.

[0062] In power system data monitoring, voltage data can only be collected at 96 moments per day. The data capacity and time coverage are small, and it is difficult to screen out voltage time series with sufficient differentiation. The energy gateway has an adjustable cloud data upload function, which can independently select the upload frequency, so as to collect more data per unit time. Assuming that the energy gateway uploads data every 3 minutes, 480 moments of data can be obtained in a day, which is 5 times the amount of data of the traditional method. By calculating the phase voltage imbalance rate of these 480 moments, the 30 moments with the largest imbalance rate are screened out, and the voltage data of the users in the substation at these moments are exported as the input sequence for voltage correlation calculation.

[0063] Optionally, let η be the phase voltage unbalance rate, U A , U B and U C are the effective values ​​of the three-phase voltages, U avg is the average value of the effective value of the three-phase voltage, and the calculation formula of the phase voltage unbalance rate is:

[0064]

[0065] Optionally, the Pearson correlation coefficient between the voltage series of all coupled users in the substation and the three-phase voltage of the transformer in the corresponding substation is calculated, including: the Pearson correlation coefficient is the product of the two voltage series and the voltage average deviation; let T and Y be two sequences, cov(T,Y) is the covariance of the sequences T and Y, σ T is the standard deviation of the sequence T, σ Y is the standard deviation of the sequence Y, the Pearson correlation coefficient is rp, and the correlation coefficient between T and Y is ρ TY , replace the population covariance with the sample covariance, replace the population standard deviation with the sample standard deviation, T and Y are two normal populations, and the calculated Pearson correlation coefficient rp is used as the correlation coefficient ρ between T and Y TY is an unbiased and consistent estimator of , then:

[0066]

[0067] In order to better implement the Pearson correlation coefficient on a microcomputer, the calculation expression can be further rewritten for discrete samples such as voltage sequences. For voltage correlation discrimination in topology recognition scenarios, assuming that the three-phase voltage sequence U A , U B , U C And the single-phase user voltage sequence U X The elements are A(k), B(k), C(k) and X(k), k is the discrete sampling point of the sequence, and N is the total number of discrete sampling points of the sequence, then:

[0068]

[0069] Optionally, let the three-phase voltage sequence be U A , U B with U C , the single-phase user voltage sequence is U X , U A The elements of A(k), U B The elements of B(k), U C The element of is C(k), the element of the single-phase user voltage sequence is X(k), k is the discrete sampling point of the sequence, N is the total number of discrete sampling points of the sequence, rp XA For U A The Pearson correlation coefficient, rp XB For U B The Pearson correlation coefficient, rp XC For U C Pearson correlation coefficient, single-phase user voltage series U X With the three-phase voltage sequence U A , U B and U C The Pearson correlation coefficient calculation formula is:

[0070]

[0071] The Pearson correlation coefficient rp mainly measures the similarity of two sequences from a linear perspective. The positive and negative correlations between samples are determined by the positive and negative values ​​of rp. If |rp| is closer to 1, the similarity between the two samples is higher.

[0072] Optionally, the Spearman correlation coefficient between the voltage sequence of all coupled users in the substation and the three-phase voltage of the transformer in the corresponding substation is calculated, including: assuming that X and Y are two sequences, each sequence has N elements, X i represents the i-th value in the sequence X, Y iRespectively represent the i-th value in sequence Y, 1≤i≤N. After arranging sequences X and Y in ascending or descending order, we get sets x and y. Let x i Represents X i Rank in sequence X, y i Represents Y i Rank in sequence Y, is the mean of the sequence X, is the mean of sequence Y, and the ranking difference set is set to d, which is defined as the difference between the corresponding elements in set x and set y, then d i =x i -y i , the Spearman correlation coefficient rs between the sequences X and Y is calculated by the set x and the set y or d, then:

[0073]

[0074] In actual use, the relationship between different sequences can be ignored, and the Spearman correlation coefficient can be calculated directly from the ranking difference set d:

[0075]

[0076] The corresponding relationship between the values ​​of the Pearson correlation coefficient and the Spearman correlation coefficient and the closeness of the correlation is generally shown in Table 1:

[0077] Table 1 Correspondence between the value of the correlation coefficient and the degree of closeness of the correlation

[0078] Absolute value of correlation coefficient Serial Correlation 0 Not relevant (0,0.3) Weak correlation [0.3,0.5) Low correlation [0.5,0.8) Significant correlation [0.8,1) Highly correlated 1 Completely relevant

[0079] Missing data is a common problem when calculating Pearson and Spearman correlation coefficients. Due to communication errors or other anomalies, the voltage data of the low-voltage area collected by the energy gateway may be missing, which will lead to inconsistent sequence lengths. In order to ensure the normal calculation of the correlation coefficient, the missing data needs to be interpolated.

[0080] Compared with Lagrange interpolation, Newton interpolation has better inheritance. When adding new data points, only one item needs to be added to the original polynomial, which simplifies the calculation process and reduces the amount of calculation. Suppose the voltage sequence is {(x1, f(x1)),…,(x n ,f(x n ))}, the missing point is (x i ,f(x i ), the interpolation expression is:

[0081] f(x i )=f(x1)+f[x2,x1](x i -x1)+…+

[0082] =f[x n ,x n-1 ,…,x1](x i -x1)…(x i -x n-1 )+

[0083] =f[x n ,x n-1 ,…,x1+x i ](x i -x1)…(x i -x n ).

[0084] Optionally, a K-means clustering algorithm is used to cluster all coupled users to obtain clustering results of coupled users in several stations, including:

[0085] Classify the input samples into K clusters, select K elements as the initial centroids of the clusters, calculate the distances between the remaining input sample elements and all centroids, and classify the elements into the cluster to which the nearest centroid belongs, completing the first iteration;

[0086] Calculate the average value of all elements in the same cluster as the new centroid, and then recalculate the distance between all sample elements and the new centroid to complete the classification of the new cluster;

[0087] For the new cluster, repeatedly calculate the new centroid, calculate the distance between all sample elements and the new centroid, complete the classification of the new cluster, and continue iterating;

[0088] When there is no change in the classification of new elements or all clusters no longer change or the number of iterations reaches the set threshold, the iteration ends and the clustering result is obtained.

[0089] Optionally, let R e is the radius of the earth, the longitude of transformer A is jA, the latitude of transformer A is ωA, the longitude and latitude of transformer A are (jA, ωA), the longitude of user B is jB, the latitude of user B is ωB, the longitude and latitude of user B are (jB, ωB), the distance between the input sample element and all the centroids, that is, the geographical distance between the user and the transformer in the substation is d AB ,but:

[0090] d AB =R e cos -1 [cosωA·cosωB·cos(jA-jB)+sinωA·sinωB].

[0091] According to the geographical distance information, the single-station area coupling user is defined as the coupling user that only exists within the power supply radius of one transformer in this type of power supply area; the multi-station area coupling user is defined as the coupling user that exists within the power supply radius of multiple transformers in this type of power supply area. The number of transformers in the cluster number of multi-station area coupling users is the number of stations within the power supply radius, and the number of single-station area coupling users is 1. Different coupling users only need to be clustered with the cluster of the station area where they are located. Through user classification processing, the overall algorithm calculation complexity can be effectively reduced and the calculation efficiency can be improved.

[0092] The present invention can perform power balance inspection and feeder identification based on the improved K-means clustering topology recognition algorithm using the average correlation coefficient method. In the low-voltage area, the sum of the power at the transformer headend and the power of the same-phase user is basically balanced. The energy gateway measures the power of the coupled user, so the balance is verified by comparing the power data of each user with the transformer.

[0093] After completing the substation attribution and phase identification, further identify the internal topology of the substation. Use voltage correlation to determine the correlation between coupled users on the same feeder, and determine the upstream and downstream connections based on the voltage level. To avoid misjudgment, take advantage of the energy gateway data and divide the voltage data of 480 moments in a day into 5 groups, each with 96 moments. Calculate the Pearson and Spearman correlation coefficients of each group and take the average value. Use the correlation coefficient and voltage level to complete feeder identification.

[0094] In the traditional correlation coefficient topology identification method, a single low-voltage area topology is taken as an example to verify the phase difference identification effect when users in a single area are connected. Figure 3 As shown in the figure, a distribution station simulation model is built through MATLAB / Simulink. The three-phase voltages on the low-voltage side of the transformer in the model are U A , U B and U C The substation has 15 single-phase coupled users, with A-phase coupled users PC1 to PC4, B-phase coupled users PC5 to PC9, and C-phase coupled users PC10 to PC15. The active power of each coupled user is set to 2kW to 5kW, and the power factor is 0.95. Through simulation, the energy gateway measures and records the voltage on the low-voltage side of the transformer and the voltage data of the users in the substation within a day.

[0095] In the traditional correlation coefficient topology identification method, the Pearson correlation coefficient is directly calculated through the measured data to perform phase discrimination or substation area identification. Due to the lack of data screening and iterative clustering process, its reliability is poor, the discrimination is low, and it may face the problem of difficulty in determining the threshold. Figure 4Taking the substation in as an example, the horizontal axis is time, the vertical axis is voltage, UA is L1, UB is L2, UC is L3, UPC5 is L4, UPC10 is L5, and the phase identification process of the B-phase coupled user PC5 and the C-phase coupled user PC10 can show the shortcomings of this method. In this substation, the three-phase voltages on the low-voltage side of the transformer are UA, UB and UC, as shown in the attached figure. Figure 5 As shown, the voltages of coupled users PC5 and PC10 are UPC5 and UPC10 respectively. Figure 5 In (a), rp=0.812, rs=0.663, Figure 5 In (b), rp=0.789, rs=0.711, Figure 5 In (c), rp=0.934, rs=0.917, Figure 5 In (d), rp=0.742, rs=0.666, Figure 5 In (e), rp=0.843, rs=0.828, Figure 5 In (f), rp = 0.819, rs = 0.794. The correlation between the coupled user voltage and the three-phase voltage and the attached Figure 5 The scatter plot shown further demonstrates the effectiveness of this approach when discrimination is insufficient.

[0096] Taking a single-station simulation example as an example, according to the algorithm proposed in this paper, data is first screened based on the phase voltage unbalance rate, and the 30 points with the largest unbalance degree are obtained from the 480 points of data, and then the correlation coefficient is calculated. The coupled users PC5, PC10 and the three-phase voltage curves at the 30 moments after screening are shown in the attached figure. Figure 8 shown.

[0097] As attached Figure 6 As shown, UA is L1, UB is L2, UC is L3, UPC5 is L4, and UPC10 is L5. Figure 6 It can be seen from the voltage curve that compared with Figure 4 After voltage screening, the difference between the three-phase voltage curves of A, B, and C increases significantly, which is more conducive to accurate identification. Based on the screened data, such as Figure 7 As shown, attached Figure 7 (a) is the correlation coefficient and scatter plot of user PC5 and phase A voltage without noise. Figure 7 (b) is the correlation coefficient and scatter plot of user PC10 and phase A voltage without noise. Figure 7 (c) is the correlation coefficient and scatter plot of user PC5 and phase B voltage without noise. Figure 7 (d) is the correlation coefficient and scatter plot of user PC10 and phase B voltage without noise. Figure 7(e) is the correlation coefficient and scatter plot of user PC5 and phase C voltage without noise. Figure 7 (f) is the correlation coefficient and scatter plot of user PC10 and phase C voltage without noise. The horizontal axis is the user phase voltage, and the vertical axis is the three-phase voltage of A, B, and C. Figure 7 In (a), rp=0.255, rs=0.147, Figure 5 In (b), rp=0.155, rs=0.273, Figure 5 In (c), rp=0.896, rs=0.839, Figure 5 In (d), rp=0.297, rs=0.336, Figure 5 In (e), rp=0.716, rs=0.740, Figure 5 In (f), rp=0.753, rs=0.862. The correlation coefficient and scatter plot of coupled users PC5 and PC10 with the three-phase voltage change significantly. Figure 5 In contrast, the filtered data resulted in significant changes in the Pearson and Spearman correlation coefficients of coupled users PC5 and PC10 relative to phases A, B, and C.

[0098] Taking the Pearson correlation coefficient as an example, the correlation coefficient between the coupled user PC5 and phase A dropped from 0.812 to 0.255, the correlation coefficient with phase C dropped from 0.843 to 0.716, and the correlation coefficient with phase B dropped from 0.934 to 0.896, and the gap increased. For PC10, the original correlation coefficient difference with phase A and phase C was 0.03, which increased to 0.598 after screening, and the Pearson correlation coefficient with phase A dropped from 0.789 to 0.155. The Spearman correlation coefficient also changed similarly. Through voltage screening, the discrimination of the correlation coefficient was significantly improved, and the topology recognition effect was optimized.

[0099] To verify the algorithm's anti-noise capability, a random noise of [-0.5V, 0.5V] is added to the coupled users PC5, PC10 and the three-phase voltages A, B, and C. Figure 8 (a) shows the correlation coefficient and scatter plot of user PC5 and phase A voltage under noise addition. Figure 8 (b) shows the correlation coefficient and scatter plot of user PC10 and phase A voltage under noise addition. Figure 8 (c) shows the correlation coefficient and scatter plot of user PC5 and phase B voltage under noise addition. Figure 8 (d) is the correlation coefficient and scatter plot of user PC10 and phase B voltage under noise addition. Figure 8 (e) is the correlation coefficient and scatter plot of user PC5 and phase C voltage under noise addition. Figure 8(f) is the correlation coefficient and scatter plot of user PC10 and phase C voltage under noise addition. The horizontal axis is the user phase voltage, and the vertical axis is the three-phase voltage of A, B, and C. Figure 7 In (a), rp=0.498, rs=0.332, Figure 5 In (b), rp=0.511, rs=0.426, Figure 5 In (c), rp=0.920, rs=0.857, Figure 5 In (d), rp=0.497, rs=0.379, Figure 5 In (e), rp=0.775, rs=0.751, Figure 5 In (f), rp = 0.782, rs = 0.807. After adding noise, data screening based on phase voltage imbalance rate is still performed. The obtained correlation coefficient and scatter plot show that although noise has a certain impact on the results, after screening, the correlation coefficients between coupled users PC5 and B phase and PC10 and C phase are significantly higher than those of other phases, proving that noise interference has little effect on the algorithm and the algorithm has good anti-noise ability.

[0100] According to the topology recognition algorithm implementation steps provided by the present invention, the number of clusters is set to 3, and K-means clustering based on the Pearson correlation coefficient and the Spearman correlation coefficient is performed on all coupled users according to the screened voltage sequence. The two clustering results are the same, as shown in Table 2.

[0101] Table 2 Clustering results of single station area example

[0102] Separation Coupling User A PC1, PC2, PC3, PC4 B PC5, PC6, PC7, PC8, PC9 C PC10, PC11, PC12, PC13, PC14, PC15

[0103] The clustering results are tested for power balance. The power data of the substation for 24 hours a day is shown in Table 3. From the data in Table 3, it can be seen that the power of each phase is basically equal to the sum of the power of each user, and the line loss rate is within a reasonable range, so the clustering result is output. Finally, the feeders and upstream and downstream in the substation are sorted. Since there is only one feeder per phase in the substation model, the final topology identification can be completed mainly through the voltage level of each phase user, and the final identification result is consistent with the simulation model.

[0104] Table 3 Single-station area example power balance test

[0105] Separation Total power / kWh Accumulated electric energy / kWh Line loss rate / % Reasonable degree A 310.3 301.4 2.87 Reasonable B 417.6 406.6 2.63 Reasonable C 548.2 532.8 2.81 Reasonable

[0106] Through the simulation example of a single substation, the shortcomings of the traditional correlation coefficient method and distance feature method were revealed, and the identification effect after a large amount of voltage data and screening based on the energy gateway was proved. This method successfully completed the household phase identification in a single substation area and verified the feasibility of the energy gateway in topology identification.

[0107] The effectiveness of the energy gateway and the corresponding algorithm in identifying household transformer relationships and feeder lines has been verified through multiple area examples. Fig. 9 As shown in the figure, a simulation model including two substations T1 and T2 and 24 single-phase coupled users was built using MATLAB / Simulink. The users of substations T1 and T2 are PC1~PC9, PC10~PC12, PC13~PC15 (substation T1) and PC16~PC18, PC19~PC21, PC22~PC24 (substation T2), PC14 and PC17 are the original topological positions, and PC1~PC9 are the A-phase feeder branches in the original topology. The power of all users is 2~5kW, and the power factor is 0.95. The 24 coupled users are connected to the energy gateway and data is collected every 3 minutes. In the simulation, the A-phase PC1PC9 users of substation T1 are divided into two feeders, simulating the branching situation, and some topological structures do not match the actual situation, which needs to be corrected through topology identification.

[0108] For household transformer identification and household phase identification, the 24 coupled users are all multi-zone coupled users. The number of clusters is set to 6, and cluster comparison is performed with the three-phase voltages of the two transformers. First, the phase voltage imbalance rate of the two transformers is calculated, and 30 moments are selected to obtain the three-phase voltage (U A1 , U B1 , U C1 , U A2 , U B2 , U C2 ) and the voltage of 24 coupled users (U PC1 ~U PC24 ). The 24 coupled users were clustered using K-means based on the Pearson and Spearman correlation coefficients, and the results were consistent, as shown in Table 4. According to the clustering results, the identification results of coupled users PC14 and PC17 were different from the original topology: PC14 was originally in phase C of station area T1, but was identified as phase A of T2; PC17 was originally in phase A of T2, but was identified as phase C of T1.

[0109] Table 4 Clustering results of multiple station area examples

[0110] Separation Coupling User Area T1: Phase A PC1, PC2, PC3, PC4, PC5, PC6, PC7, PC8, PC9 Area T1: Phase B PC10, PC11, PC12 Area T1: Phase C PC13, PC15, PC17 Area T2: Phase A PC14, PC16, PC18 Area T2: Phase B PC19, PC20, PC21 Area T2: Phase C PC22, PC23, PC24

[0111] The energy balance of the clustering results is verified, as shown in Table 5. From the data in the table, it can be seen that when the clustering results are used as the topological structure, the sum of the head-end power of the six phases in the two substations for 24 hours a day and the power of all users is basically balanced, and the line loss rate is within the normal range, which proves the correctness of the clustering results.

[0112] Table 5. Clustering results of multi-station area example for power balance verification

[0113] Separation Total electricity / kWh Accumulated electric energy / kWh Line loss rate / % Reasonable degree Area T1: Phase A 840.4 816.5 2.84 Reasonable Area T1: Phase B 266.3 260.4 2.22 Reasonable Area T1: Phase C 220.6 214.6 2.74 Reasonable Area T2: Phase A 303.9 295.9 2.64 Reasonable Area T2: Phase B 253.7 246.2 2.95 Reasonable Area T2: Phase C 237.6 231.8 2.45 Reasonable

[0114] For feeder identification, according to the algorithm flow, it is necessary to determine the feeder and upstream and downstream relationships of the coupled users of the same phase in the same substation. Phase B and phase C of substation T1 and phase A, phase B, and phase C of substation T2 are all single feeder structures, which can be sorted according to the voltage level. Phase A of substation T1 has two feeders. Taking the 9 coupled users of this phase as an example, 5 groups of Pearson and Spearman correlation coefficients are calculated, and the average correlation coefficient is obtained. The results are shown in Tables 6 and 7.

[0115] Table 6 Average Pearson correlation coefficient of coupled users PC1~PC9

[0116] node 1 2 3 4 5 6 7 8 9 1 1 0.720 0.710 0.704 0.727 0.717 0.981 0.983 0.978 2 0.720 1 0.987 0.983 0.985 0.984 0.718 0.712 0.702 3 0.710 0.987 1 0.962 0.969 0.971 0.709 0.688 0.691 4 0.704 0.983 0.962 1 0.968 0.964 0.700 0.691 0.695 5 0.727 0.985 0.969 0.968 1 0.962 0.725 0.706 0.705 6 0.717 0.984 0.971 0.964 0.962 1 0.720 0.695 0.701 7 0.981 0.718 0.709 0.700 0.725 0.720 1 0.966 0.656 8 0.983 0.712 0.688 0.691 0.706 0.695 0.966 1 0.966 9 0.978 0.702 0.691 0.695 0.705 0.701 0.956 0.966 1

[0117] Table 7 Average Spearman correlation coefficients of coupled users PC1ˉPC9

[0118] node 1 2 3 4 5 6 7 8 9 1 1 0.599 0.549 0.596 0.610 0.611 0.987 0.990 0.983 2 0.599 1 0.987 0.988 0.984 0.986 0.568 0.576 0.583 3 0.549 0.987 1 0.970 0.971 0.965 0.511 0.523 0.539 4 0.596 0.988 0.970 1 0.984 0.970 0.564 0.578 0.583 5 0.610 0.984 0.971 0.984 1 0.968 0.576 0.582 0.593 6 0.611 0.986 0.965 0.970 0.968 1 0.583 0.588 0.599 7 0.987 0.568 0.511 0.564 0.576 0.583 1 0.981 0.977 8 0.990 0.576 0.523 0.578 0.582 0.588 0.981 1 0.978 9 0.983 0.583 0.539 0.583 0.593 0.599 0.977 0.978 1

[0119] Taking the average Pearson correlation coefficient as an example, it can be seen from the data in the analysis table that the correlation coefficient of coupled users PC1 with PC7 and PC8 is significantly higher than the correlation coefficient with PC2ˉPC6, indicating that PC1, PC7, and PC8 may be on the same feeder. Similarly, the correlation coefficient of PC7 with PC1, PC8, and PC9 is high, the correlation coefficient of PC8 with PC1, PC7, and PC9 is high, and the correlation coefficient of PC9 with PC1, PC7, and PC8 is also high. The correlation coefficient between PC2ˉPC6 is high, indicating that PC1 and PC7ˉPC9 are not on the same feeder. According to the voltage level, the coupled users on the feeder are further sorted upstream and downstream, and the final corrected topology is consistent with the simulation model, as shown in the attached figure. Fig.10 shown.

[0120] The present invention has the following advantages:

[0121] (1) The energy gateway-based method significantly improves the accuracy of low-voltage substation topology identification and reduces errors through real-time data collection and analysis. The energy gateway solution can collect electrical parameters in real time, dynamically adjust and update the topology structure, and ensure the continuous accuracy of identification. The energy gateway achieves extensive data collection through smart meters and wireless communications, making up for the insufficient coverage of traditional methods. The data collection method based on the energy gateway utilizes existing smart meters and communication equipment, reduces hardware investment, and reduces operation and maintenance costs through a remote platform.

[0122] (2) The present invention is based on the topology recognition of voltage correlation and energy gateway data, obtains more high-precision voltage data through the energy gateway, screens the voltage imbalance rate at critical moments, and improves the accuracy of low-voltage area topology recognition;

[0123] (3) The present invention combines the Pearson and Spearman correlation coefficients to comprehensively evaluate the voltage sequence similarity from the perspectives of linearity and monotonicity, thereby enhancing the reliability of the recognition results;

[0124] (4) The K-means clustering algorithm is used to optimize the topology recognition results through power balance inspection, and the positioning function is used to improve the computing efficiency and user classification accuracy.

[0125] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A low voltage area topology identification application method based on energy gateway, characterized in that: include: Calculate the geographical distance between users based on their geographic information data, determine the user's geographical location, and classify multi-area coupled users and single-area coupled users. Set up an energy gateway at each coupled user and transformer node for data collection. According to the classification results, the voltage data of the transformer in the production area at the time of three-phase voltage imbalance are screened, and the voltage data are screened based on the phase voltage imbalance rate to obtain the voltage data of several moments with the largest phase voltage imbalance rate; Calculate the Pearson correlation coefficient and Spearman correlation coefficient between the voltage series of all coupled users in the substation and the three-phase voltage of the transformer in the corresponding substation, and interpolate the missing data; All coupled users are clustered using the K-means clustering algorithm to obtain clustering results of coupled users in several stations; Compare the clustering results and use the principle of power balance to test and screen the clustering results; Calculate the average correlation coefficient between the in-phase coupled users in each area by group; The final topological structure of the low-voltage area is determined based on the average correlation coefficient and voltage level.

2. According to claim 1, a low-voltage area topology identification application method based on energy gateway is characterized in that: According to the classification results, the voltage data of the transformer in the substation at the time of three-phase voltage imbalance are screened, and the voltage data are screened based on the phase voltage imbalance rate to obtain the voltage data at several times when the phase voltage imbalance rate is the largest, including: According to the voltage data collected during the day, for low-voltage substations that only contain single-substation coupled users, several moments with the largest phase voltage imbalance rate are screened; for low-voltage substations that contain multiple substation coupled users, several common moments with the largest phase voltage imbalance rate are first screened. When the common moments are less than the set number, the moment with the largest phase voltage imbalance rate in the substation is selected based on the remaining required number of moments, and the voltage values ​​of all substation coupled users at the corresponding moments are derived as the input voltage sequence.

3. According to claim 1, a low-voltage area topology identification application method based on energy gateway is characterized in that: set up η is the phase voltage unbalance rate, U A , U B and U C are the effective values ​​of the three-phase voltages, U avg is the average value of the effective value of the three-phase voltage, and the calculation formula of the phase voltage unbalance rate is:

4. According to claim 1, a low-voltage area topology identification application method based on energy gateway is characterized in that: Calculate the Pearson correlation coefficient between the voltage series of all coupled users in the substation and the three-phase voltage of the transformer in the corresponding substation, including: the Pearson correlation coefficient is the product of the two voltage series and the voltage average deviation; let T and Y be two sequences, cov(T,Y) is the covariance of the sequences T and Y, σ T is the standard deviation of the sequence T, σ Y is the standard deviation of the sequence Y, the Pearson correlation coefficient is rp, and the correlation coefficient between T and Y is ρ TY , replace the population covariance with the sample covariance, replace the population standard deviation with the sample standard deviation, T and Y are two normal populations, and the calculated Pearson correlation coefficient rp is used as the correlation coefficient ρ between T and Y TY is an unbiased and consistent estimator of , then:

5. According to claim 4, a low-voltage area topology identification application method based on energy gateway is characterized in that: set up The three-phase voltage sequence is U A , U B with U C , the single-phase user voltage sequence is U X , U A The elements of A(k), U B The elements of B(k), U C The element of is C(k), the element of the single-phase user voltage sequence is X(k), k is the discrete sampling point of the sequence, N is the total number of discrete sampling points of the sequence, rp XA For U A The Pearson correlation coefficient, rp XB For U B The Pearson correlation coefficient, rp XC For U C Pearson correlation coefficient, single-phase user voltage series U X With the three-phase voltage sequence U A , U B and U C The Pearson correlation coefficient calculation formula is:

6. According to claim 1, a low-voltage area topology identification application method based on energy gateway is characterized in that: Calculate the Spearman correlation coefficient between the voltage sequence of all coupled users in the substation and the three-phase voltage of the transformer in the corresponding substation, including: Let X and Y be two sequences, each with N elements, X i represents the i-th value in the sequence X, Y i Respectively represent the i-th value in sequence Y, 1≤i≤N. After arranging sequences X and Y in ascending or descending order, we get sets x and y. Let x i Represents X i Rank in sequence X, y i Represents Y i Rank in sequence Y, is the mean of the sequence X, is the mean of sequence Y, and the ranking difference set is set to d, which is defined as the difference between the corresponding elements in set x and set y, then d i =x i -y i , the Spearman correlation coefficient rs between the sequences X and Y is calculated by the set x and the set y or d, then:

7. According to claim 1, a low-voltage area topology identification application method based on energy gateway is characterized in that: All coupled users are clustered using the K-means clustering algorithm to obtain clustering results of coupled users in several areas, including: Classify the input samples into K clusters, select K elements as the initial centroids of the clusters, calculate the distances between the remaining input sample elements and all centroids, and classify the elements into the cluster to which the nearest centroid belongs, completing the first iteration; Calculate the average value of all elements in the same cluster as the new centroid, and then recalculate the distance between all sample elements and the new centroid to complete the classification of the new cluster; For the new cluster, repeatedly calculate the new centroid, calculate the distance between all sample elements and the new centroid, complete the classification of the new cluster, and continue iterating; When there is no change in the classification of new elements or all clusters no longer change or the number of iterations reaches the set threshold, the iteration ends and the clustering result is obtained.

8. According to claim 7, a low-voltage area topology identification application method based on energy gateway is characterized in that: Assume R e is the radius of the earth, the longitude of transformer A is jA, the latitude of transformer A is ωA, the longitude and latitude of transformer A are (jA, ωA), the longitude of user B is jB, the latitude of user B is ωB, the longitude and latitude of user B are (jB, ωB), the distance between the input sample element and all the centroids, that is, the geographical distance between the user and the transformer in the substation is d AB ,but: d AB =R e body -1 [cosωA·cosωB·cos(jA-jB)+sinωA·sinωB]。

Images