Abnormal diagnosis method of line loss in transformer area

Abstract

The present invention discloses a method for diagnosing abnormal line loss in a substation, which relates to the technical field of power system analysis and is used to solve the problems of low accuracy and non-intuitiveness of existing line loss analysis. The method comprises the following steps: obtaining substation data; generating a situation base map based on the low-voltage substation topology relationship in the substation data; calculating the three-phase line loss of the low-voltage substation and rendering the situation base map; performing image recognition and cluster analysis based on the rendered situation base map to obtain the abnormal location and preliminary cause classification of the line loss in the substation; performing abnormal factor diagnosis based on the abnormal location and preliminary cause classification; analyzing the correlation between the line loss abnormality and the abnormal factors to obtain the abnormal line loss analysis results of the substation. The present invention performs abnormal analysis on the substation line loss and displays it intuitively through situation perception and image recognition clustering technology.

Description

Technical Field

[0001] The present invention relates to the technical field of power system analysis, and in particular to a method for diagnosing abnormal line loss in a low-voltage substation based on situational awareness visualization technology. Background Art

[0002] With the continuous acceleration of urban construction, the scale of low-voltage distribution networks has become increasingly large and their structure has become increasingly complex. With the substantial increase in electricity demand from residents and industry, line loss and energy supply quality issues in low-voltage distribution areas have gradually been exposed. Among them, line loss is an important comprehensive technical indicator reflecting the operation of the distribution network. It is one of the main signs and means of comprehensively measuring the management level of power companies. It can also directly reflect the power supply company's grid planning and construction, operation and maintenance, production technology, and business management level. Improper management of line loss in low-voltage distribution areas will not only affect the production and operating efficiency of power supply companies, but also result in a waste of resources. Therefore, reducing the line loss rate has become the most effective way for power supply companies to reduce production costs, improve economic benefits, and save energy and reduce emissions. Pursuing the lowest possible line loss rate is an important means to improve corporate economic benefits and is also one of the goals of power companies.

[0003] Currently, line loss analysis is primarily conducted manually. Due to the involvement of multiple professionals, the low-voltage distribution network structure, the complex nature of customers, and its reliance on hardware facilities and personnel expertise, line loss analysis results are inaccurate and unable to reveal specific line loss issues. Different analysts also use inconsistent standards for line loss compliance, resulting in a low input-output ratio for substation upgrades. This extensive management and passive supervision hinders precise investment and lean management. Therefore, there is an urgent need to promote lean management of line losses across substations, improve the accuracy of line loss analysis, and achieve loss reduction and efficiency improvement for the power grid. Summary of the Invention

[0004] In order to overcome the deficiencies of the prior art, one of the objectives of the present invention is to provide a method for diagnosing abnormal line loss in a substation, which uses situational awareness technology in combination with equipment parameters to obtain substation diagnosis results.

[0005] A method for diagnosing abnormal line loss in a transformer area comprises the following steps:

[0006] Get station area data;

[0007] Generate a situation base map based on the low-voltage substation area topological relationship in the substation area data;

[0008] Calculating the three-phase line loss of the low-voltage substation area and rendering the line loss on the situation base map;

[0009] Based on the rendered situation base map, image recognition and cluster analysis are performed to obtain the abnormal location of line loss in the substation area and the preliminary cause of the abnormality;

[0010] Conduct further diagnosis of abnormal factors based on the abnormal location and preliminary causes of the abnormality;

[0011] The correlation between the line loss and the abnormal factors is analyzed to obtain the abnormal analysis results of the substation line loss.

[0012] Furthermore, generating a situation base map based on the low-voltage substation topological relationship in the substation data includes the following steps:

[0013] Acquiring area topology line information from the area data;

[0014] Traverse the nodes in the line information and calculate the gravity of the nodes connected to each node. i The calculation of the gravitational force satisfies:

[0015]

[0016] Where m is the total number of nodes in the graph; For node N i and N j The distance between a For other nodes to point N i The gravitational coefficient, L is the ideal length of the branch line, F r (N i ) represents node N i The net gravitational force, F a (N i ,N j ) represents node N i and N j The gravitational force between them, i and j are the node numbers;

[0017] According to the point-edge avoidance and edge-edge avoidance algorithms, the repulsion force of each node is calculated. i The total repulsive force calculation satisfies:

[0018]

[0019] Where m1 is the number of nodes not on this feeder; n1 and q are the number of nodes on the previous level branch of the current node of this feeder and the number of remaining nodes respectively; k r,1 For node pair N on other feeders i The repulsive force weight, k r,2 Point N i The repulsive force weight of the nodes on the previous level branch, k r,3 For the remaining node pairs N i The repulsive weight of

[0020] Adding the gravitational resultant force and the total repulsive force to obtain a final resultant force;

[0021] According to the resultant force and the node motion equation, the node is moved, and the node motion equation satisfies:

[0022]

[0023] Where k is the node N i The number of iterations; Node N i The projections of the resultant gravitational force, gravitational force, and total repulsive force on the x-axis and y-axis; d k is the distance from the node to the center point in k steps, and g is the gravity coefficient;

[0024] Repeated iterations are performed based on the positions of the nodes after movement to obtain the final positions of the nodes and lines, and generate a situation base map.

[0025] Furthermore, the three-phase line loss of the low-voltage area is calculated, the line loss is rendered on the situation base map, and the key data of the low-voltage area is intuitively displayed on the situation map, including the following steps:

[0026] Obtain low-voltage substation topology parameters, line parameters, voltage, active and reactive power output data;

[0027] Calculate line loss by phase and obtain three-phase line loss data of low-voltage substation;

[0028] Color rendering is performed based on the situation base map and the line loss data to obtain a line loss situation map, a voltage situation map, and a line power situation map.

[0029] Furthermore, color rendering is performed according to the situation base map and the line loss data, including:

[0030] Classifying line loss levels according to the line loss data, matching corresponding line segment colors according to the line loss levels, and rendering the line segment colors to obtain the line loss situation map;

[0031] Dividing the voltage levels, matching the corresponding line segment colors according to the voltage levels, and rendering the line segment colors to obtain the voltage situation map;

[0032] The line power levels are divided, the corresponding line segment colors are matched according to the power levels, and the line segment colors are rendered to obtain the line power situation diagram.

[0033] Furthermore, based on the rendered situation base map, image recognition and cluster analysis of relevant subject data situation maps are performed to obtain the abnormal location and preliminary cause of line loss in the substation area, including the following steps:

[0034] To reflect the image's color space distribution, the line loss, power, and voltage situation maps are segmented, dividing the global color features of the image into several localized color features. Since the situation maps all display the three-phase A, B, and C lines simultaneously in a 120° sector, the segmentation is considered by dividing the distribution transformer into nine segments with 120° intervals from the inside out.

[0035] Extract the characteristic vectors in the line loss situation diagram, power situation diagram and voltage situation diagram respectively, and the characteristic vectors include color histogram and color density histogram, which are recorded as {N(I i ,C j ),D(I i ,C j )|i∈[1,…,m],j∈[1,…,n]};where, N(I i ,C j ) is the image block I i Falls into color C j The sum of the number of pixels, m is the number of color blocks in image I, n is the number of color levels in color space C; color density D(I i ,C j )=N(I i ,C j ) / L j ,L j is the length of the color segment with color level j.

[0036] Extracting historical data of the substation area from the substation area data, performing cluster fitting on the historical data, and obtaining typical line loss, power, and voltage situation characteristics of the substation area; performing feature matching based on the feature vector, the typical line loss, and the voltage situation characteristics, including the following steps:

[0037] The similarity measurement is calculated by Euclidean distance to obtain Euclidean distance, which satisfies the following conditions:

[0038] Among them, d(I1,I T ) represents the Euclidean distance between the current situation feature vector and the typical situation feature vector, I1 represents the situation feature vector of the current situation graph, I T Represents the situation feature vector of a typical situation graph, D(I 1i ,C j ) indicates that the i-th block of the current situation graph I1 is in color level C j Color density value, D(I Ti ,C j ) represents a typical situation diagram I T The i-th block image is in color level C j Color density value, N(I1i ,C j ) indicates that the i-th block of the current situation graph I1 is in color level C j The number of pixels, N(I Ti ,C j ) represents a typical situation diagram I T The i-th block image is in color level C j The number of pixels;

[0039] According to the Euclidean distance, the feature comparison is performed, and the situation feature of the current station area is compared with the typical situation feature of the station area to determine d(I1,I T ) is greater than the upper limit of the Euclidean distance of this type. If it is less than, the current substation has typical situation characteristics of the substation; otherwise, it is considered that the current substation has abnormal line loss situation characteristics;

[0040] If there are abnormal line loss features, compare the number of pixels and color density of each block to lock the abnormal area and accurately locate the abnormal point.

[0041] Based on the result of the feature matching and the situation base map, the abnormal position of the line loss and the preliminary cause of the abnormality are located;

[0042] According to the result of the feature matching and the situation base map, the abnormal position of the line loss and the preliminary cause of the abnormality are located.

[0043] Furthermore, cluster fitting is performed on the historical data, comprising the following steps:

[0044] Screening and cleaning the characteristic vector data to obtain line loss stable area characteristic data;

[0045] According to the preset number of clusters and termination threshold, cluster calculation is performed using the K-means algorithm;

[0046] Calculating the silhouette coefficient according to the cluster calculation result;

[0047] According to the optimal contour system, a clustering result is output.

[0048] Furthermore, based on the abnormality location, abnormality factors are diagnosed, including three-phase imbalance, distributed photovoltaic access, load concentration at the end of the line, and line problems. Voltage anomalies can be further diagnosed based on three-phase imbalance and distributed photovoltaic access, while power anomalies can be diagnosed based on load concentration at the end of the line and excessively long power supply paths.

[0049] Furthermore, the three-phase imbalance factors include three-phase voltage imbalance and three-phase current imbalance, which are analyzed and judged through the symmetrical component decomposition method; the factor of load concentration at the end of the line is analyzed and judged through the algorithm of finding the center of mass in three-dimensional space; the factor of the line power supply path being too long is analyzed and judged by calculating the maximum voltage drop in the substation and comprehensively displaying it with the situation diagram; the line problem factor is analyzed and judged through line aging diagnosis and wire cross-sectional area too small diagnosis.

[0050] Furthermore, the correlation between the line loss and the abnormal factors is analyzed to obtain the abnormal analysis results of the substation line loss, including:

[0051] The correlation coefficient between the line loss abnormality feature vector and the abnormal factor vector is calculated to determine the correlation. The correlation coefficient is the Pearson correlation coefficient, which satisfies:

[0052] Among them, the line loss abnormal feature vector is P, the i-th abnormal factor vector is Ri, P=[P1,P2…P n ],Ri=[Ri1,Ri2…Ri n ],μ P is the mean of the line loss anomaly feature vector elements, μ Ri is the mean of the elements of the i-th abnormal factor vector;

[0053] The main causes of abnormal line loss are matched according to the correlation coefficient. When |p(P,Ri)|<0.4, all abnormal factors are not the main causes of abnormal line loss. When 0.4≤|p(P,Ri)|<0.7, the factor with the largest |p(P,Ri)| value is selected as the main cause of abnormal line loss. When 0.7≤|p(P,Ri)|<1, the factors with the largest and second largest |p(P,Ri)| values ​​are selected as the main causes of abnormal line loss.

[0054] Compared with the prior art, the present invention has the following beneficial effects:

[0055] The present invention uses situational awareness technology, combined with various substation equipment parameters and operating parameters, to conduct scientific and objective diagnostic analysis to improve the accuracy of line loss research and judgment; through situation diagrams, the substation data is intuitively displayed, and image analysis technology is combined with cluster analysis technology to perform anomaly analysis, converting the large-scale data analysis problem of long time series and multiple data themes into an image recognition problem, which can achieve rapid positioning and diagnosis of line loss anomalies, and has the advantages of occupying less computing resources and fast comparison speed. This method can not only display the substation data characteristics and anomaly positioning, but also analyze and display the specific routes and causes of the occurrence of substation line loss anomalies. There is no need for manual abnormal line loss analysis, which improves the accuracy and uniformity of the analysis and realizes lean management of substation line losses and loss reduction and efficiency improvement. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 This is a flow chart of a method for diagnosing abnormal line loss in a single-station area according to an embodiment;

[0057] Figure 2 1 is a schematic diagram of a situation base map of the first embodiment;

[0058] Figure 3 This is a schematic diagram of the situation diagram after line loss rendering in Example 1;

[0059] Figure 4 It is a block diagram of the situation map;

[0060] Figure 5 This is a line loss cause analysis path diagram. DETAILED DESCRIPTION

[0061] The present invention will be described in more detail below with reference to the accompanying drawings. It should be noted that the following description of the present invention with reference to the accompanying drawings is merely illustrative and non-limiting. Various embodiments may be combined with each other to form other embodiments not shown in the following description.

[0062] Example 1

[0063] Example 1 provides a method for diagnosing abnormal line loss in a substation based on situational awareness visualization technology, which aims to diagnose technical line loss in abnormal line loss in a substation through situational awareness visualization technology, and perform image analysis and mining on the abnormal line loss situation map of the substation to obtain a typical line loss situation map of the substation.

[0064] Currently, line loss is typically assessed by defining a qualified range of line loss values ​​for a given substation over the same period. This criterion is relatively simple, lacks flexibility, and fails to directly demonstrate specific line loss issues or locate them. The widespread application of technologies such as big data and cloud computing has created conditions for achieving simultaneous line loss management, as well as locating and visualizing line loss issues across substations.

[0065] Please refer to Figure 1 As shown, a method for diagnosing abnormal line loss in a transformer area includes the following steps:

[0066] S1. Obtaining station area data;

[0067] The above-mentioned substation data includes substation topology line information, line parameters, voltage data, etc. The data is obtained by conventional technical means. Each device in the substation has a recording device for relevant data. This embodiment will not elaborate on the acquisition method.

[0068] S2. Generate a situation base map based on the low-voltage substation area topology relationship in the substation area data;

[0069] Please refer to the schematic diagram of the S2 situation base map Figure 2This shows a base map of the low-voltage substation in a county-level city in Zhejiang Province. Step S2 provides a clear and beautiful wiring diagram base for the next step of rendering the situation based on smart meter measurements and line loss calculation results. It specifically includes:

[0070] Acquiring area topology line information from the area data;

[0071] Traverse the nodes in the line information and calculate the gravity of the nodes connected to each node. i The calculation of the gravitational force satisfies:

[0072]

[0073] Where m is the total number of nodes in the graph; For node N i and N j The distance between a For other nodes to point N i The gravitational coefficient, L is the ideal length of the branch line, F r (N i ) represents node N i The net gravitational force, F a (N i ,N j ) represents node N i and N j The gravitational force between them, i and j are the node numbers;

[0074] According to the point-edge avoidance and edge-edge avoidance algorithms, the repulsion force of each node is calculated. i The total repulsive force calculation satisfies:

[0075]

[0076] Where m1 is the number of nodes not on this feeder; n1 and q are the number of nodes on the previous level branch of the current node of this feeder and the number of remaining nodes respectively; k r,1 For node pair N on other feeders i The repulsive force weight, k r,2 Point N i The repulsive force weight of the nodes on the previous level branch, k r,3 For the remaining node pairs N i The repulsive weight of

[0077] Adding the attractive force and the repulsive force to obtain a resultant force;

[0078] According to the resultant force and the node motion equation, the node is moved, and the node motion equation satisfies:

[0079]

[0080] Where k is the node N i The number of iterations; Node N i The projections of the attraction and repulsion, the attraction, and the total repulsion on the x-axis and the y-axis; d k is the distance from the node to the center point in k steps, and g is the gravity coefficient;

[0081] Repeated iterations are performed based on the positions of the nodes after movement. Specifically, iterations are performed based on the resultant force of the latest node positions. In each iteration, the node is displaced by the resultant force. As the iterations proceed, the entire system reaches a force balance, and ultimately the final positions of each node and line are obtained, generating a situation base map.

[0082] S3. Calculate the three-phase line loss of the low-voltage substation area and render the line loss on the situation base map;

[0083] Please refer to the line loss situation diagram rendered in S3 Figure 3 The line loss situation diagram of a low-voltage substation in a county-level city in Zhejiang Province is shown in FIG. Step S3 can intuitively display the key data of the low-voltage substation on the situation diagram. S3 specifically includes:

[0084] Obtain low-voltage substation topology parameters, line parameters, voltage, active and reactive power output data;

[0085] Calculate line loss by phase and obtain three-phase line loss data of low-voltage substation;

[0086] Color rendering is performed based on the situation base map and the line loss data to obtain a line loss situation map, a power situation map, and a voltage situation map.

[0087] Since smart meters collect data every 15 minutes, the above data can be obtained directly.

[0088] The principle of the above phase calculation is as follows: assuming that node j is a terminal user, that is, a node representing a single-phase user, and node i is a trunk node connecting single-phase users, the power of node i can be obtained from the three parameters of node j, active power, reactive power, and voltage, as follows:

[0089]

[0090] Among them, P loss·ij and Q loss·ij are the active and reactive losses of the line, P i and Q i are the active and reactive power of node i, r ij and x ij are line resistance and reactance respectively, Pj , Q j and U j are the active power, reactive power and node voltage of node j respectively.

[0091] In this way, the line loss of each section is calculated section by section.

[0092] It should be noted that the three-phase line loss, power and voltage situation diagrams are generated separately, and then the correlation analysis is performed.

[0093] The above color rendering according to the situation base map includes:

[0094] The line loss data is divided into line loss levels, and the corresponding line segment colors are matched according to the line loss levels. The line loss situation diagram is obtained by rendering the line segment colors. The voltage level is divided into voltage levels, and the corresponding line segment colors are matched according to the voltage level. The voltage situation diagram is obtained by rendering the line segment colors. The line power level is divided into line power levels, and the corresponding line segment colors are matched according to the power level. The line power situation diagram is obtained by rendering the line segment colors.

[0095] Different colors are used to visually display the distribution changes of the data features of interest within the low-voltage substation, transforming the problem of analyzing line loss anomalies in the substation into an image recognition problem. For example, the line loss rate (-∞, 0) is represented by blue, [0, 2) by green, [2, 7) by yellow, and [7, ∞) by red. The voltage (230, 231) is represented by dark blue, [231, 233) by light blue, [233, 235) by cyan, [235, 237) by orange, and [237, 238) by light yellow. The colors gradually change with the voltage. The same applies to line power.

[0096] S4. Performing image recognition and cluster analysis on the rendered situation map to obtain abnormal line loss location in the substation area;

[0097] Cluster analysis and mining of cluster-related subject data of S4. By mining typical situation features and matching the situation features of current data, the abnormal line loss location of the substation area is obtained, and the specific line where the abnormal line loss occurs in the substation area is identified. Specifically, the following are included:

[0098] In order to reflect the color space distribution information of the image, the line loss, power and voltage situation diagrams are divided into blocks respectively, and the global color features of the image are divided into several local area color features. Since the situation diagrams all display the A, B, and C three-phase substation lines in a 120° sector shape, when considering the block division, the distribution transformer is taken as the center and divided into 9 blocks with equal intervals of 120° from the inside to the outside, such as Figure 4 shown.

[0099] Extract the characteristic vectors in the line loss situation diagram, power situation diagram and voltage situation diagram respectively, and the characteristic vectors include color histogram and color density histogram, which are recorded as {N(I i ,C j ),D(I i ,C j )|i∈[1,…,m],j∈[1,…,n]};where, N(I i ,C j ) is the image block I i Falls into color C j The sum of the number of pixels, m is the number of color blocks in image I, n is the number of color levels in color space C; color density D(I i ,C j )=N(I i ,C j ) / L j ,L j is the length of the color segment with color level j.

[0100] Extracting historical data of the substation area from the substation area data, performing cluster fitting on the historical data, and obtaining typical line loss, power and voltage situation characteristics of the substation area;

[0101] Performing feature matching on the typical line loss, power and voltage situation features according to the feature vector;

[0102] The abnormal position of line loss is located according to the result of the feature matching and the situation base map.

[0103] In order to more accurately obtain the typical situation characteristics of each data theme such as line loss, voltage, and current in the substation area, this embodiment also performs cluster fitting of historical data, specifically including:

[0104] Screening and cleaning the characteristic vector data to obtain line loss stable area characteristic data;

[0105] According to the preset number of clusters and termination threshold, cluster calculation is performed using the K-means algorithm;

[0106] Calculating the silhouette coefficient according to the cluster calculation result;

[0107] According to the optimal contour system, a clustering result is output.

[0108] Clustering calculation is performed using the K-means algorithm, and the steps are as follows:

[0109] 1) Screen and clean the characteristic data to obtain the characteristic data of the line loss stable area;

[0110] 2) Set the number of clusters K = n, where n is a positive integer, and set the cluster calculation iteration termination threshold;

[0111] 3) Perform clustering calculation to obtain the clustering result when the number of clusters is K;

[0112] 4) Calculating the overall silhouette coefficient of the clustering results in step 3);

[0113] The silhouette coefficient is mainly used to evaluate the effectiveness of clustering. The value is between [-1, 1]. The larger the value, the better the clustering effect. The specific calculation method is as follows:

[0114] a) For the i-th element x i , calculate x i The average distance to all other elements in the same category A, denoted as a i , used to quantify the degree of cohesion within a category.

[0115] b) Select x i A category B outside, calculate x i The average distance to all points in B, traverse all other categories, find the nearest average distance, recorded as b i , used to quantify the degree of separation between categories.

[0116] c) For element x i , silhouette coefficient s i =(b i -a i ) / max(a i , b i ).

[0117] d) Calculate the silhouette coefficient of all elements x in A, and the average value is the overall silhouette coefficient of the current cluster.

[0118] e) If s i Less than 0, indicating x i The average distance between the elements in its category is greater than that of the nearest other categories, indicating that the clustering effect is not good. i tends to 0, or b i Big enough, then s i

[0119] It is close to 1, indicating that the clustering effect is good.

[0120] 5) Determine whether K is less than n+p-1, where p is a positive integer. If so, K=n+1, and go to step

[0121] 3), otherwise go to step 6);

[0122] 6) Compare p overall silhouette coefficients, determine the optimal overall silhouette coefficient, and obtain the optimal clustering result;

[0123] In addition, in order to analyze the abnormal situation of the subject data and clearly locate the abnormality, feature matching is performed based on the feature vector situation characteristics, including:

[0124] The similarity measurement calculation formula is constructed by Euclidean distance, and the calculation of the Euclidean distance satisfies:

[0125]

[0126] Among them, d(I1,I T ) represents the Euclidean distance between the current situation feature vector and the typical situation feature vector, I1 represents the situation feature vector of the current situation graph, I T Represents the situation feature vector of a typical situation graph, D(I 1i ,C j ) indicates that the i-th block of the current situation graph I1 is in color level C j Color density value, D(I Ti ,C j ) represents a typical situation diagram I T The i-th block image is in color level C j Color density value, N(I 1i ,C j ) indicates that the i-th block of the current situation graph I1 is in color level C j The number of pixels, N(I Ti ,C j ) represents a typical situation diagram I T The i-th block image is in color level C j The number of pixels.

[0127] After calculating the Euclidean distance, feature comparison is required. According to the Euclidean distance formula, d(I1,I T ) is smaller, indicating that the color distribution of the current situation map and the typical situation map is more consistent. Specifically, the situation characteristics of the current area are compared with the typical situation characteristics of the area to determine d(I1,I T ) is greater than the upper limit of the Euclidean distance of this class. If the comparison results are similar enough (d(I1,I T ) is less than the upper limit of the Euclidean distance for this category, the current substation is considered to have typical substation situation characteristics. Otherwise, it is considered to have abnormal line loss situation characteristics. This method can quickly compare the current situation diagram with the typical situation diagram, and can calculate a large amount of multiple subject data such as line loss, power, and voltage over a long time series and multiple nodes, with the advantages of low computing resource consumption and fast comparison speed.

[0128] If there are abnormal line loss trends, the abnormal area can be further locked by further comparing the number of pixels and color density of the blocks, and the abnormal situation point can be accurately located. Through S4, a typical theme situation map at each abnormal situation moment can be obtained. By comparing the size of the Euclidean distance of different theme situation maps, the basic cause of the line loss abnormality can be determined. For example, if there is an abnormality in the line loss situation map at time t, if the line loss abnormal situation map - voltage abnormal situation map - power abnormal situation map at that moment are retrieved, and it is found that the Euclidean distance of the voltage situation map is much larger than the Euclidean distance of the power situation map, it can be preliminarily determined that the line loss abnormality at time t is caused by voltage abnormality, which narrows the scope for accurate judgment in the next step and reduces the amount of calculation.

[0129] S5. diagnose abnormal factors based on the abnormal location;

[0130] Based on the above abnormal location and basic causes of abnormality, abnormal factor diagnosis is performed to analyze the specific causes of line loss abnormality. S5 specifically includes: three-phase imbalance factor analysis, distributed photovoltaic access factor analysis, load concentration at the end of the line factor and line problem factor analysis. Among them, voltage abnormalities can be further analyzed from two aspects: three-phase imbalance factor and distributed photovoltaic access factor. Power abnormalities can be further analyzed from two aspects: load concentration at the end of the line factor and excessively long line power supply path factor. Figure 5 shown.

[0131] The above-mentioned three-phase unbalance factors are divided into three-phase voltage imbalance and three-phase current imbalance. The three-phase voltage imbalance includes the three-phase voltage zero-sequence imbalance and the three-phase voltage negative-sequence imbalance. When the amplitude and phase of the three-phase voltage are known, the symmetrical component method decomposition method is used to calculate the positive-sequence, zero-sequence and negative-sequence voltage components, and calculate the zero-sequence and negative-sequence voltage imbalance. Similarly, the zero-sequence and negative-sequence current imbalance can be calculated. Among them, the component method decomposition method satisfies:

[0132]

[0133] in, and are the positive sequence, negative sequence and zero sequence voltages respectively; α is the rotation factor.

[0134] The voltage negative sequence unbalance is:

[0135]

[0136] The voltage zero-sequence unbalance is:

[0137] Calculate the voltage negative sequence unbalance ε based on the three-phase voltage at 96 moments U2 and voltage zero sequence unbalance ε U0, the voltage negative sequence imbalance is greater than the set limit K U2 Or the voltage zero sequence unbalance is greater than the set limit K U0 In this case, if it continues for M days, it can be determined that the voltage is unbalanced.

[0138] The aforementioned factors contributing to the concentration of load at the end of the line are analyzed through situational awareness. Specifically, situational awareness technology can be used to detect whether the load is concentrated at the end of the line. The low-voltage distribution network situation base map is a uniform radial distribution line diagram centered on the distribution transformer. It has the following characteristics: 1) The radial distribution network wiring diagram centered on the distribution transformer can reflect the relative geographical location of each power terminal; 2) The power terminals are evenly distributed and clearly distinguished; 3) They are generally single-phase; they can be expanded to three-phase; 4) The graphics are clear and the situational awareness is high. The situation base map clearly shows the relative and causal relationship between each load, and it can be used to determine whether the load is concentrated at the end of the line.

[0139] Taking the distribution transformer as the center O, the algorithm for finding the centroid in three-dimensional space can be used to find the load center O0'(X0', Y0', Z0') of the distribution transformer area, then:

[0140]

[0141]

[0142] Among them, W i The sum of the electricity consumption of each user in the meter box. If OO0' is greater than the set limit K L , it can be judged that the load is concentrated at the end of the line.

[0143] The above-mentioned factors affecting excessive power supply paths are determined by calculating the maximum voltage drop in the substation area and combining them with a situation diagram to determine if the power supply path is too long. The difference between the voltage at the low-voltage output terminal of the transformer and the load voltage at the same time as peak power demand is calculated. If the voltage difference exceeds a certain threshold, the power supply path may be too long. Combined with the voltage situation diagram, if the situation diagram also shows a long relative path, the power supply line in the substation area is considered too long.

[0144] The specific mechanism for the impact of distributed household PV installation locations on substation line losses is as follows: when the PV capacity connected to the substation is at a level sufficient to reduce line losses, the PV power access point is generally selected at a node with a high load factor and close to the load zone. By analyzing the substation's voltage profile, if line losses are abnormal and the voltage at the installed PV user and surrounding users increases, even exceeding a threshold, and using a three-dimensional centroid algorithm to calculate the load center O0', a problem is determined if the distance between the PV installation location and the load center exceeds the threshold.

[0145] The impact of distributed household PV installation capacity on substation line losses is explained as follows: when the access capacity of distributed PV is less than twice the load capacity, distribution network line losses are reduced; when the access capacity of distributed PV is equal to twice the load capacity, distribution network line losses remain unchanged before and after access; and when the access capacity of distributed PV is greater than twice the load capacity, distribution line losses increase. By analyzing the substation's voltage and power flow diagrams, if the power flow is reversed, the voltage at the PV access point is high, and the voltage at surrounding users also increases significantly, and if the line loss rate is also abnormal, it can be determined that the PV installation capacity does not match the substation capacity.

[0146] The principle of the above line aging diagnosis method is:

[0147] If both of the following conditions are met at the same time, the line has an aging problem:

[0148] 1. Line age determination: If the line age is greater than the threshold, the line is aging.

[0149] 2. Analyze the long-term trend line loss diagram. If the line loss carrying the same load increases significantly, it is determined that the line loss is abnormal due to line aging.

[0150] The principle of the above-mentioned diagnosis and judgment of small wire cross-sectional area is as follows: Diagnosis method of small wire cross-sectional area based on situational awareness:

[0151] First, determine whether the line loss rate exceeds the normal range. Second, perform cluster analysis on the power flow situation diagram of the substation. If the line diameter is smaller than the load-bearing substation, it is considered that the line diameter is too small.

[0152] S6. Analyze the correlation between the line loss and the abnormal factors to obtain an abnormal analysis result of the substation line loss.

[0153] The purpose of S6 is to analyze the main causes of abnormal line loss on specific lines, including:

[0154] The correlation coefficient between the line loss abnormality feature vector and the abnormal factor vector is calculated to determine the correlation. The correlation coefficient is the Pearson correlation coefficient, which satisfies:

[0155] Among them, the line loss abnormal feature vector is P, the i-th abnormal factor vector is Ri, P=[P1,P2…P n ],Ri=[Ri1,Ri2…Ri n ],μ P is the mean of the line loss anomaly feature vector elements, μ Ri is the mean of the elements of the i-th abnormal factor vector;

[0156] The main causes of abnormal line loss are matched according to the correlation coefficient. When |p(P,Ri)|<0.4, all abnormal factors are not the main causes of abnormal line loss. When 0.4≤|p(P,Ri)|<0.7, the factor with the largest |p(P,Ri)| value is selected as the main cause of abnormal line loss. When 0.7≤|p(P,Ri)|<1, the factors with the largest and second largest |p(P,Ri)| values ​​are selected as the main causes of abnormal line loss.

[0157] In summary, this method for diagnosing abnormal line loss in a substation area is based on situational awareness visualization technology. It diagnoses technical line loss within abnormal line loss in the substation area and performs image recognition and cluster analysis on the abnormal line loss situation map. Using image recognition methods, it quickly locates the abnormal location and preliminary cause of the abnormality, narrowing the scope and reducing the computational effort for the next step of precise analysis. Correlation analysis between the abnormal line loss feature vector and the abnormal factor vector is used to identify the primary cause of the abnormal line loss in the substation area.

[0158] Those skilled in the art can make various other corresponding changes and deformations based on the technical solutions and concepts described above, and all of these changes and deformations should fall within the scope of protection of the claims of the present invention.

Claims

1. A method for diagnosing abnormal line loss in a transformer area, characterized in that: The following steps are involved: Get station area data; Generate a situation base map based on the low-voltage substation area topological relationship in the substation area data; Calculating the three-phase line loss of the low-voltage substation area and rendering the line loss on the situation base map; Based on the rendered situation base map, image recognition and cluster analysis are performed to obtain the abnormal location of line loss in the substation area and the preliminary cause of the abnormality; Based on the rendered situation base map, image recognition and cluster analysis of relevant subject data are performed to obtain the abnormal location of line loss in the substation area and the preliminary abnormal cause, including the following steps: The line loss situation map, power situation map, and voltage situation map are divided into blocks, and the global color features of the image are divided into multiple local area color features. The first block is divided into nine sectors with 120° equal spacing from the inside to the outside, centered on the distribution transformer. Extracting feature vectors from the line loss situation diagram, the power situation diagram, and the voltage situation diagram respectively, wherein the feature vectors include a color histogram and a color density histogram; Extracting historical data of the substation area from the substation area data, performing cluster fitting on the historical data, and obtaining typical line loss and voltage situation characteristics of the substation area; Performing feature matching based on the feature vector, the typical line loss, and the voltage situation feature includes the following steps: The similarity measurement is calculated by Euclidean distance to obtain Euclidean distance, which satisfies the following conditions: ， in, d ( I 1, I T ) represents the Euclidean distance between the current situation feature vector and the typical situation feature vector, I 1 represents the current situation graph situation feature vector, I T Represents the situation feature vector of a typical situation graph, D ( I 1i , C j ) represents the current situation map I 1st i The block diagram is in color level C j The color density value, D ( I Ti , C j ) represents a typical situation diagram I T No. i The block diagram is in color level C j The color density value, N ( I 1i , C j ) represents the current situation map I 1st i The block diagram is in color level C j The number of pixels, N ( I Ti , C j ) represents a typical situation diagram I T No. i The block diagram is in color level C j The number of pixels; According to the Euclidean distance, the characteristics of the current situation in the area are compared with the typical situation characteristics of the area to determine whether the current situation in the area is normal or not. d ( I 1, I T ) is greater than the upper limit of the Euclidean distance of this type. If it is less than, the current substation has typical situation characteristics of the substation; otherwise, it is considered that the current substation has abnormal line loss situation characteristics; If there are abnormal line loss features, compare the number of pixels and color density of each block, lock the abnormal area, and locate the abnormal point; Based on the result of the feature matching and the situation base map, the abnormal position of the line loss and the preliminary cause of the abnormality are located; Diagnose abnormal factors based on the abnormal location and preliminary abnormal causes; The correlation between the line loss and the abnormal factors is analyzed to obtain the abnormal analysis results of the substation line loss.

2. The method for diagnosing abnormal line loss in a transformer area according to claim 1, wherein: Generating a situation base map based on the low-voltage substation area topological relationship in the substation area data includes the following steps: Acquiring area topology line information from the area data; Traverse the nodes in the line information and calculate the gravity of the nodes connected to each node. The calculation of the gravitational force satisfies: ， ， in, is the total number of nodes in the graph; For nodes and The distance between For other nodes The gravitational coefficient, is the ideal length of the branch line, Representation node The net gravitational force, Representation node and The gravitational force between and Number the node; The repulsion force of each node is calculated according to the point-edge avoidance and edge-edge avoidance algorithms. The total repulsive force calculation satisfies: ， in, is the number of nodes that are not on this feeder; and are the number of nodes on the branch of the previous level and the number of remaining nodes of the current node of this feeder respectively; For node-to-point on other feeders The repulsive weight, For nodes The repulsive force weight of the nodes on the previous level branch, For the remaining node pairs The repulsive weight of Adding the gravitational resultant force and the total repulsive force to obtain a final resultant force; According to the final resultant force and the node motion equation, the node is moved, and the node motion equation satisfies: ， ， ， ， ， in, For nodes The number of iterations; 、 , , 、 , Node The total gravitational force, gravitational force, and total repulsive force are Axis and projection on axis; for The distance between the node and the center point, is the gravity coefficient; Repeated iterations are performed based on the positions of the nodes after movement to obtain the final positions of the nodes and lines, and generate a situation base map.

3. The method for diagnosing abnormal line loss in a transformer area according to claim 1, wherein: Calculating the three-phase line loss of the low-voltage area, rendering the line loss on the situation base map, and visually displaying the key data of the low-voltage area on the situation map, including the following steps: Obtain low-voltage substation topology parameters, line parameters, voltage, active and reactive power output data; Calculate line loss by phase and obtain three-phase line loss data of low-voltage substation; Color rendering is performed based on the situation base map and the line loss data to obtain a line loss situation map, a voltage situation map, and a line power situation map.

4. The method for diagnosing abnormal line loss in a transformer area according to claim 3, wherein: Performing color rendering according to the situation base map and the line loss data includes: Classifying line loss levels according to the line loss data, matching corresponding line segment colors according to the line loss levels, and rendering the line segment colors to obtain the line loss situation map; Dividing the voltage levels, matching the corresponding line segment colors according to the voltage levels, and rendering the line segment colors to obtain the voltage situation map; The line power levels are divided, the corresponding line segment colors are matched according to the power levels, and the line segment colors are rendered to obtain the line power situation diagram.

5. The method for diagnosing abnormal line loss in a transformer area according to claim 1, wherein: The feature vector includes a color histogram and a color density histogram, and the feature vector is represented as { N ( I i , C j ), D ( I i , C j )|i∈[1,…, m ], j ∈[1,…, n ]},in, N ( I i , C j ) is the image block I i Falling into color grades C j The sum of the number of pixels, m For images I The number of color blocks, n Color space C The number of color levels; color density D ( I i , C j )= N ( I i , C j ) / L j , L j Color space C The j The length of the color line segment for each color level.

6. The method for diagnosing abnormal line loss in a transformer area according to claim 1, wherein: Performing cluster fitting on the historical data includes the following steps: Screening and cleaning the data of the characteristic vector to obtain characteristic data of the line loss stable area; According to the preset number of clusters and termination threshold, cluster calculation is performed using the K-means algorithm; Calculate the silhouette coefficient based on the clustering calculation results; According to the optimal silhouette coefficient, the clustering result is output.

7. The method for diagnosing abnormal line loss in a transformer area according to claim 1, wherein: Based on the abnormal location, abnormal factor diagnosis is performed, including: three-phase imbalance factor analysis, distributed photovoltaic access factor analysis, load concentration at the line end factor and line problem factor analysis; among them, voltage abnormalities are analyzed from two directions: three-phase imbalance factor and distributed photovoltaic access factor, and power abnormalities are analyzed from two directions: load concentration at the line end factor and excessively long line power supply path factor.

8. The method for diagnosing abnormal line loss in a transformer area according to claim 7, wherein: The three-phase imbalance factors include three-phase voltage imbalance and three-phase current imbalance, which are analyzed and judged through the symmetrical component decomposition method; the factor of load concentration at the end of the line is analyzed and judged through the algorithm of finding the center of mass in three-dimensional space; the factor of the line power supply path being too long is analyzed and judged by calculating the maximum voltage drop in the substation and comprehensively displaying it with the situation diagram; the line problem factor is analyzed and judged through line aging diagnosis and wire cross-sectional area being too small diagnosis.

9. The method for diagnosing abnormal line loss in a transformer area according to claim 1, wherein: Analyze the correlation between the line loss and the abnormal factors to obtain the abnormal analysis results of the substation line loss, including: The correlation coefficient between the line loss abnormality feature vector and the abnormal factor vector is calculated to determine the correlation. The correlation coefficient is the Pearson correlation coefficient, which satisfies: , where the line loss abnormal feature vector is P , No. i' The abnormal factor vector is Ri' , P =[ P 1, P 2… P n ], Ri' =[ Ri' 1, Ri' 2… Ri' n ], is the mean of the line loss anomaly feature vector elements, For the i' The mean of the elements of the abnormal factor vector; Match the main cause of line loss anomaly based on the correlation coefficient. When | p ( P , Ri' )| <0.4, all abnormal factors are not the main cause of line loss abnormality; when 0.4≤| p ( P , Ri' )|<0.7, then select | p ( P , Ri' )|The factor with the largest value is the main cause of abnormal line loss; when 0.7≤| p ( P , Ri' )|<1, then select | p ( P , Ri' The factors with the largest and second largest values ​​are the main causes of abnormal line loss.

Images