A statistical method and system for additional losses due to regional transformer imbalance

Abstract

This invention discloses a method and system for statistical analysis of regional transformer substation unbalanced losses. The method includes: clustering the loss currents of data-rich substations using a clustering algorithm, and minimizing the weighted terms of the profile coefficient and the sum of squares of the clusters using clustering evaluation indices; classifying data-scarce substations into clusters by extracting feature vectors from them; and employing a distance estimation method to statistically analyze regional transformer substation unbalanced losses, considering outliers in the unbalanced loss estimation. This invention obtains detailed time-series current data from selected typical substations and combines this data with the average phase current data from other substations. It utilizes statistical methods, such as Chebyshev's inequality, to evaluate and improve the accuracy and confidence of the unbalanced loss range. This significantly reduces the cost and workload of data acquisition and processing, and successfully achieves a comprehensive assessment of the total unbalanced losses for all substations, including neutral and phase lines.

Description

Technical Field

[0001] This invention relates to the field of low-voltage distribution network research technology, and in particular to a statistical method and system for calculating additional losses due to regional transformer imbalance. Background Technology

[0002] Three-phase imbalance is a common problem in my country's low-voltage distribution network, and assessing the additional losses caused by this imbalance is an urgent issue for power supply companies. However, a key challenge in assessing the additional losses caused by imbalance across a large number of distribution areas is the difficulty in obtaining time-series data for all distribution areas.

[0003] Existing technologies address energy loss caused by imbalance in the UK low-voltage network by combining clustering, classification, and range estimation. This method finds a match between networks with only annual average phase current (data-scarce networks) and network clusters with time-series phase current data (data-rich networks). Then, it estimates the range of energy loss caused by imbalance in data-rich network clusters, similar to data-scarce networks. However, this method only calculates the energy loss due to neutral line imbalance, neglecting the impact of imbalance on phase line losses. Some methods calculate the three-phase imbalance degree based on 24-hour hourly three-phase current data for transformer areas, classify current imbalance using cluster analysis, and then use adversarial generative networks to increase the sample size. For different categories of three-phase load current imbalance, it calculates the line loss increment coefficient, quantifying the impact of three-phase imbalance in line loss theoretical calculations. However, this method requires time-series data of the three-phase current for the transformer area for quantitative analysis of unbalanced line losses. For quantitative analysis of unbalanced line losses in a large area, it is difficult to obtain the required time-series current data for each transformer area, and obtaining time-series current data for a large number of transformer areas increases the operating costs of the power supply company.

[0004] Another method calculates the average load current at the beginning of the line based on data monitored at the load end; then, it calculates the equivalent resistance of the low-voltage distribution line based on the monitored meter readings; next, it calculates the single-phase and three-phase imbalance based on real-time collected data; and finally, it calculates the line loss. This method can accurately represent the impact of each three-phase imbalance on the overall line loss. However, for assessing the line loss of a large number of transformer substations in a region, this method requires obtaining relevant calculation data for each substation individually, resulting in high computational costs. Summary of the Invention

[0005] In view of the aforementioned existing problems, the present invention is proposed.

[0006] Therefore, the present invention provides a method and system for statistical analysis of additional losses due to regional transformer imbalance, which solves the problem that existing technologies cannot obtain and process detailed time-series current data required for a large number of transformer areas in order to accurately assess the additional line losses caused by three-phase imbalance without significantly increasing operating costs.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0008] In a first aspect, the present invention provides a method for statistical analysis of regional transformer substation imbalance losses, comprising: using a clustering algorithm to cluster the loss currents of data-rich transformer substations, and minimizing the weighted terms of the profile coefficient and the sum of squares of the clusters through a clustering evaluation index; classifying the data-scarce transformer substations into the clusters by extracting feature vectors of data-scarce transformer substations; and considering the existence of outliers in the estimation of imbalance losses, using a distance estimation method to statistically analyze the regional transformer substation imbalance losses.

[0009] As a preferred embodiment of the regional transformer area imbalance additional loss statistical method described in this invention, the loss current extracted from the data-rich transformer area includes:

[0010] Based on the relationship between current sequence components and phasors, the line loss when the system has a three-phase imbalance problem is:

[0011]

[0012] Where I1, I2, and I0 are the positive, negative, and zero-sequence currents of the three-phase currents when the system is unbalanced, respectively, R is the line resistance, and I N Neutral current;

[0013] Furthermore, based on the characteristic that the positive sequence current of an unbalanced system is equal to the current of a symmetrical system, the additional losses due to three-phase unbalance are obtained as follows:

[0014]

[0015] Among them, I A I B I C These are the three-phase currents, S is the three-phase apparent power, and U is the line voltage;

[0016] The phase line loss current I is obtained based on the additional losses caused by the three-phase imbalance. pl and neutral line loss current I nl .

[0017] As a preferred embodiment of the regional transformer area imbalance additional loss statistical method described in this invention, the step of clustering the loss current of the extracted data-rich transformer area using a clustering algorithm includes:

[0018] An improved k-means algorithm and hierarchical clustering algorithm are used to analyze the phase line loss current I. pl and neutral line loss current I nl Clustered into k clusters;

[0019] The improved k-means clustering selects k initial cluster centers, calculates the distance from each data point to each cluster center, and assigns it to the nearest cluster. For each cluster, the average distance between all data points within each cluster is used as the new cluster center. The assignment of data points and the recalculation of cluster centers are repeated until the cluster centers no longer change or the iteration converges.

[0020] The hierarchical clustering algorithm treats each data point as a cluster and generates higher-level clusters by merging the clusters with the least similarity to each other.

[0021] As a preferred embodiment of the statistical method for regional transformer area imbalance additional loss described in this invention, the weighted terms for minimizing the silhouette coefficient and the sum of squares of clusters through cluster evaluation indices include:

[0022] By using a bi-objective optimization model as a weighted term to minimize the silhouette coefficient and the sum of squares of clusters as cluster evaluation metrics, the bi-objective optimization model is calculated as follows:

[0023]

[0024] Where C is the weighting factor, s(k) is the silhouette coefficient, and w(k) is the cluster sum of squares.

[0025] As a preferred embodiment of the regional transformer area imbalance additional loss statistical method described in this invention, the extraction of feature vectors for data-scarce transformer areas includes:

[0026] Define virtual loss current I hl and virtual three-phase imbalance ε ub As a feature vector, the formula is expressed as:

[0027]

[0028] Among them, I av I bv I cv These represent the average currents of phases A, B, and C, respectively.

[0029] As a preferred embodiment of the regional transformer area imbalance additional loss statistical method described in this invention, the method includes: classifying the data-scarce transformer areas into clusters using the K-nearest neighbor algorithm and the MSVM algorithm.

[0030] For a sample to be classified, find the K nearest samples in the feature space, vote according to the category of the K samples, and finally classify the sample to be classified into the category with the most votes; the MSVM algorithm transforms the multi-classification problem into multiple binary classification problems.

[0031] The feature vectors and cluster labels are used as training data and input into the classification model. The classification accuracy of the two models is compared, and the classification model with higher accuracy is selected for the classification of data-scarce substations. The feature vectors of the data-scarce substations are input to realize the classification process of the substations.

[0032] As a preferred embodiment of the regional transformer area imbalance additional loss statistical method described in this invention, the method of using distance estimation to statistically analyze the regional transformer area imbalance additional loss includes:

[0033] The data-rich areas are clustered into k clusters, and the maximum range of imbalance-related additional losses for each cluster is determined to be [P]. lossmin P lossmax ], where P lossmin and P lossmax These represent the minimum and maximum values ​​of the additional energy loss due to imbalance, respectively.

[0034] The Chebyshev inequality is used to describe the unbalanced energy loss distribution of clusters, expressed as:

[0035]

[0036] Where X represents the unbalanced energy loss of the platform area in the cluster, μ represents the average value of the unbalanced energy loss, and σ represents the standard deviation of the unbalanced energy loss;

[0037] For a cluster m, the average unbalanced energy loss needs to be calculated. Then, the difference between the unbalanced energy loss of each data-rich area and the average unbalanced energy loss is calculated. After deleting data-rich areas with larger differences, the range of unbalanced energy losses of the remaining areas is the confidence range of the unbalanced energy loss of the cluster m.

[0038] Secondly, the present invention provides a regional transformer area imbalance additional loss statistics system, comprising:

[0039] The clustering module is used to cluster the loss current of the extracted data-rich transformer area using a clustering algorithm, and to minimize the weighted terms of the profile coefficient and the sum of squares of the clusters through clustering evaluation metrics.

[0040] The classification module is used to classify the data-scarce transformer areas into the cluster of the cluster by extracting the feature vectors of the data-scarce transformer areas;

[0041] The loss statistics module is used to account for outliers in the estimation of unbalanced additional losses and uses a distance estimation method to realize the statistics of unbalanced additional losses in regional transformer areas.

[0042] Thirdly, the present invention provides an electronic device, comprising:

[0043] Memory and processor;

[0044] The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, they implement the steps of the method for statistical analysis of regional transformer imbalance additional losses.

[0045] Fourthly, the present invention provides a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the method for statistical analysis of additional losses due to regional transformer imbalance.

[0046] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention provides a method and system for statistical analysis of unbalanced additional losses in regional transformer substations. By selecting a number of typical transformer substations to obtain detailed time-series current data and combining it with the average phase current data of the remaining transformer substations, statistical methods such as Chebyshev's inequality are used to evaluate and improve the accuracy and confidence of the range of unbalanced additional losses. This not only significantly reduces the cost and workload of data acquisition and processing, but also successfully achieves a comprehensive assessment of the total unbalanced additional losses of all transformer substations, including neutral and phase lines, providing power supply companies with an economical, effective, accurate and reliable line loss assessment solution. Attached Figure Description

[0047] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 This is a schematic diagram of the overall process logic of the regional transformer area imbalance additional loss statistical method according to an embodiment of the present invention;

[0049] Figure 2 This is a schematic diagram of the k-means++ elbow clustering result of the statistical method for regional transformer area imbalance additional loss described in an embodiment of the present invention.

[0050] Figure 3 This is a schematic diagram of the hierarchical clustering elbow method results of the statistical method for regional transformer area imbalance additional loss according to an embodiment of the present invention.

[0051] Figure 4 This is a schematic diagram of the characteristic regions of each cluster in the regional transformer area imbalance additional loss statistical method according to an embodiment of the present invention;

[0052] Figure 5 This is a schematic diagram of the confusion matrix of the KNN classification results of the statistical method for additional losses due to regional transformer imbalance described in an embodiment of the present invention.

[0053] Figure 6 This is a schematic diagram illustrating the confidence range of unbalanced additional losses for each cluster in the statistical method for regional transformer area unbalanced additional losses according to an embodiment of the present invention. Detailed Implementation

[0054] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0055] Example 1

[0056] Reference Figure 1 As one embodiment of the present invention, a method for statistical analysis of additional losses due to regional transformer imbalance is provided, such as... Figure 1 The specific steps shown are as follows:

[0057] S100: Clustering algorithm is used to cluster the loss current of the extracted data-rich transformer area, and the weighted terms of the profile coefficient and the sum of squares of the clusters are minimized through clustering evaluation index;

[0058] S200: By extracting the feature vectors of data-scarce transformer areas, data-scarce transformer areas are classified into clusters.

[0059] S300: Considering the existence of outliers in the estimation of unbalanced additional losses, a distance estimation method is used to realize the statistics of unbalanced additional losses in regional transformer areas.

[0060] It should be noted that this invention provides a method and system for statistical analysis of regional transformer substation unbalanced additional losses. By selecting a subset of typical transformer substations to obtain detailed time-series current data and combining it with the average phase current data of the remaining substations, statistical methods such as Chebyshev's inequality are used to evaluate and improve the accuracy and confidence of the unbalanced additional loss range. This not only significantly reduces the cost and workload of data acquisition and processing but also successfully achieves a comprehensive assessment of the total unbalanced additional losses of all transformer substations, including neutral and phase lines. This provides power supply companies with an economical, effective, accurate, and reliable line loss assessment solution.

[0061] In this embodiment of the application, step S100 includes the following sub-steps A1-A3;

[0062] In A1: Extract the loss current of the data enrichment area;

[0063] In A2: a clustering algorithm is used to cluster the loss current of the extracted data-rich transformer areas;

[0064] In A3: Minimize the weighted terms of the silhouette coefficient and the sum of squares of the clusters by using cluster evaluation metrics;

[0065] In one feasible implementation, extracting the loss current of data-rich transformer areas includes:

[0066] Due to the presence of the zero-sequence component of the neutral line, the additional losses caused by the imbalance in a three-phase four-wire system consist of two parts: phase line losses and neutral line losses. The current sequence component and phasor have the following relationship:

[0067]

[0068] Among them, I A I B I C They represent the three-phase currents, with I1, I2, and I0 representing the positive, negative, and zero-sequence currents of the three-phase currents when the system is unbalanced, respectively.

[0069] Assuming the line resistance is R, the line loss when the system is in equilibrium is: P bal_loss =3I 2 R, where I represents the current in each phase; the line loss when the system has a three-phase imbalance problem is: Among them, I N Indicates the neutral line current;

[0070] Based on the relationship between current sequence components and phasors, the line loss when the system has a three-phase imbalance problem is:

[0071]

[0072] Where I1, I2, and I0 are the positive, negative, and zero-sequence currents of the three-phase currents when the system is unbalanced, respectively, R is the line resistance, and I N Neutral current;

[0073] Given that the current in a balanced system contains only a positive-sequence component, the additional losses in an unbalanced system compared to a balanced system are caused by the negative-sequence current, zero-sequence current, and neutral current. Therefore, the additional losses due to unbalance are:

[0074]

[0075] in, neutral current

[0076] The phase current I can be obtained from the power grid measurement data. A I B I CIn a three-phase four-wire system, since the quadrilateral formed by the A, B, and C phase currents and the neutral current is not unique, it is difficult to obtain the positive-sequence current component by measuring the current amplitude in the power grid. Based on the characteristic that the positive-sequence current in an unbalanced system equals the current in a symmetrical system, we have:

[0077]

[0078] Where S is the three-phase apparent power and U is the line voltage;

[0079] Furthermore, based on the characteristic that the positive sequence current of an unbalanced system is equal to the current of a symmetrical system, the additional losses due to three-phase unbalance are obtained as follows:

[0080]

[0081] The phase line loss current I is obtained based on the additional losses caused by the three-phase imbalance. pl and neutral line loss current I nl The formula is:

[0082]

[0083] Phase line loss current and neutral line loss current will be used as the basis for clustering of the substation area for subsequent data enrichment.

[0084] In one feasible implementation, an improved k-means algorithm and hierarchical clustering algorithm are used to analyze the phase line loss current I. pl and neutral line loss current I nl Clustered into k clusters;

[0085] The improved k-means clustering selects k initial cluster centers, calculates the distance from each data point to each cluster center, and assigns it to the nearest cluster. For each cluster, the average distance between all data points within each cluster is used as the new cluster center. The assignment of data points and the recalculation of cluster centers are repeated until the cluster centers no longer change or the iteration converges.

[0086] Hierarchical clustering algorithms treat each data point as a cluster and generate higher-level clusters by merging the clusters with the least similarity to each other.

[0087] In one feasible implementation, a bi-objective optimization model is used as a weighted term for minimizing the silhouette coefficient and the sum of squares of clusters as cluster evaluation metrics. The bi-objective optimization model is calculated as follows:

[0088]

[0089] Where C is the weighting factor (C>0), s(k) is the silhouette coefficient, and w(k) is the cluster sum of squares;

[0090] A higher profile coefficient indicates lower intra-cluster distance and higher inter-cluster distance; therefore, the inter-cluster distance measurement is expressed as:

[0091]

[0092] Where N represents the number of samples, a (i) b represents the average distance between sample i and other samples in the same cluster. (i) This represents the average distance between sample i and other cluster samples;

[0093] The Within-Cluster Sum of Squares (WCSS) is defined to measure the compactness of clusters. Specifically, WCSS measures the sum of the squared distances from each point within a cluster to the cluster center. A smaller WCSS value indicates that the points within the cluster are closer to each other, meaning the cluster is more compact. WCSS is calculated as follows:

[0094]

[0095] Where, x i Let xc represent the i-th sample in the j-th cluster. ,j This represents the cluster center of the j-th cluster.

[0096] It should be noted that step S100 above can effectively identify different transformer area groups with similar loss characteristics. This not only helps improve the quality of clustering results and ensures that data points within the same group are more similar, but also makes the differences between different groups more obvious. Furthermore, by optimizing the silhouette coefficient and the sum of squares of clusters, the effectiveness and stability of clustering can be mathematically quantified and improved. This enables power companies to more effectively locate problem areas, take targeted measures to reduce power loss, and ultimately improve energy efficiency.

[0097] In this embodiment of the application, step S200 includes the following sub-steps B1-B2;

[0098] In B1: Extract the feature vector of the data-sparse transformer area;

[0099] In B2: Classify data-scarce areas into clusters;

[0100] In one feasible implementation, extracting the feature vector of data-scarce transformer areas includes:

[0101] In data-scarce areas, only the three-phase average current data is collected once at a fixed time. However, in data-rich areas, the additional losses due to imbalance are closely related to the three-phase imbalance degree and the loss current. Therefore, a virtual loss current I is defined. hl and virtual three-phase imbalance ε ubAs a feature vector, the formula is expressed as:

[0102]

[0103] Among them, I av I bv I cv These represent the average currents of phases A, B, and C, respectively.

[0104] It should be noted that for data-rich transformer areas, two feature vectors can be easily calculated from the time-series current data and used together with the cluster labels as training data for the classification model.

[0105] In one feasible implementation, the K-Nearest Neighbors (KNN) algorithm and the MSVM algorithm are used to classify data-scarce areas into clusters. KNN is based on the basic assumption that "similar samples have similar labels" and determines the classification of new samples by comparing the distance between new samples and training samples. MSVM separates different classes by finding the maximum margin, and the performance of support vector machines has been widely recognized.

[0106] Specifically, the main characteristic of the KNN method is that it uses distance metrics for classification, without needing to build a classification model. The basic idea is: for a sample to be classified, find the K nearest samples in the feature space, vote on the categories of these K samples, and finally assign the sample to the category with the most votes. This consists of the following steps:

[0107] First, let's assume the training set is... in It is an eigenvector, y i This corresponds to the cluster label. Calculate the feature vector of the area to be classified. With each training sample feature vector Euclidean distance:

[0108]

[0109] Next, select the K nearest samples from the training set, and then select the class that appears most frequently among these K nearest neighbors as the sample to be classified. Category:

[0110]

[0111] Where, N k Let I(·) represent the set of K nearest neighbor samples. I(·) is an indicator function that takes the value 1 when the condition is true and 0 otherwise.

[0112] Specifically, the MSVM (Multi-Class Support Vector Machine) extends the binary support vector machine (SVM) into a multi-class classifier. For each binary classification problem, SVM distinguishes between the two classes of data points by finding an optimal hyperplane in a high-dimensional space. MSVM, on the other hand, transforms the multi-class classification problem into multiple binary classification problems. Its mathematical model is as follows:

[0113]

[0114] Where, ω ij and b ij These are the classifier parameters, σ ijk The value of x represents the degree of misclassification, C represents the penalty parameter, which controls the trade-off between model complexity and misclassification. ijk It is the k-th sample, and its label is y. ijk +1 (if belonging to category i), or -1 (if belonging to category j); n ij f is the total number of samples in category i and category j. ij (x) is the decision function for each binary classification model. This represents the voting mechanism, where I(·) is an indicator function that takes the value 1 when the condition is true and 0 otherwise.

[0115] Specifically, feature vectors and cluster labels are used as training data input into the classification model. The classification accuracy of the two models is compared, and the classification model with higher accuracy is selected for classification of data-scarce substations. The feature vectors of the data-scarce substations are input to realize the classification process of the substations.

[0116] It should be noted that step S200 above, through feature vector extraction, can capture the key attributes of data-scarce transformer areas, making the classification process more scientific and reasonable, and enhancing the model's adaptability to different situations. Classifying data-scarce transformer areas into appropriate clusters helps identify potential connections and differences between these areas and data-rich transformer areas, providing valuable information for further exploration of power loss patterns, thereby improving the overall operating efficiency and energy utilization level of the power grid system.

[0117] In this embodiment of the application, step S300 above considers the existence of outliers in the estimation of unbalanced additional losses, and uses a distance estimation method to realize the statistics of regional unbalanced additional losses, including:

[0118] Additional energy loss due to three-phase imbalance in low-voltage distribution area The 150 data-rich distribution areas were clustered into k clusters, and the maximum range of imbalance-related additional losses for each cluster was determined to be [P]. lossmin P lossmax ], where P lossmin and P lossmax These represent the minimum and maximum values ​​of the additional energy loss due to imbalance, respectively.

[0119] However, this range is easily affected by outliers within the cluster. To reduce the impact of outliers, certain statistical methods are needed to narrow this range and increase the confidence level. Chebyshev's inequality, an important inequality in probability theory and statistics, can describe the dispersion of data points at a certain confidence level for a dataset with known mean and standard deviation, without requiring assumptions about the data distribution. Therefore, we consider using Chebyshev's inequality to describe the unbalanced energy loss distribution of clustering clusters. Its formula is expressed as:

[0120]

[0121] Where X represents the unbalanced energy loss of the cluster, μ represents the average unbalanced energy loss, and σ represents the standard deviation of the unbalanced energy loss; this formula represents the probability of falling within the interval [μ-3σ, μ+3σ], i.e., the confidence level is 1. That is, 89%;

[0122] For cluster m, the average unbalanced energy loss needs to be calculated. Then, the difference between the unbalanced energy loss of each data-rich area and the average unbalanced energy loss is calculated. After deleting data-rich areas with larger differences, the range of unbalanced energy losses of the remaining areas is the confidence range of unbalanced energy loss of cluster m, with a confidence level of 89%.

[0123] To verify the 89% confidence level, a five-fold cross-validation method was used to randomly divide a cluster m into five groups of equal size. One group was used as the validation group, and the remaining four groups were used as the training group. The Chebyshev inequality was applied to narrow down the imbalanced energy loss distribution of the training group to the range of the 89% confidence level. Then, the probability that the validation group data would not fall within this range was calculated. This process was repeated, with each group serving as a validation group. Finally, five probability values ​​were obtained, and their average was calculated to be close to 89%, thus verifying the 89% confidence level.

[0124] It should be noted that step S300, by considering the impact of outliers and applying distance estimation methods, can more accurately measure and analyze the additional unbalanced losses between different distribution areas. This avoids the biases that may arise from outliers in traditional methods, thus providing more robust data support for power system loss assessment. Furthermore, this step enhances adaptability to various uncertainties in complex power grid environments, enabling a true reflection of regional distribution area unbalanced losses even when data exhibits significant fluctuations or anomalies. This, in turn, supports more scientific and rational decision-making and management optimization.

[0125] The above is an illustrative scheme of a method for statistically analyzing additional losses due to regional transformer imbalance in this embodiment. It should be noted that the technical solution of this system for statistically analyzing additional losses due to regional transformer imbalance is based on the same concept as the method described above. Details not described in detail in the technical solution of the system for statistically analyzing additional losses due to regional transformer imbalance in this embodiment can be found in the description of the method described above.

[0126] This embodiment also provides a regional transformer area imbalance additional loss statistics system, including:

[0127] The clustering module is used to cluster the loss current of the extracted data-rich transformer area using a clustering algorithm, and to minimize the weighted terms of the profile coefficient and the sum of squares of the clusters through clustering evaluation metrics.

[0128] The classification module is used to classify data-scarce transformer areas into clusters by extracting feature vectors from these areas.

[0129] The loss statistics module is used to account for outliers in the estimation of unbalanced additional losses and uses a distance estimation method to realize the statistics of unbalanced additional losses in regional transformer areas.

[0130] The above-mentioned unit modules can be embedded in the processor of the computer device in hardware form or independent of it, or they can be stored in the memory of the computer device in software form, so that the processor can call and execute the corresponding operations of the above modules.

[0131] This embodiment also provides an electronic device, which includes a processor, a memory, a communication interface, a display screen, and an input device connected via a system bus. The processor of this computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The communication interface of the computer device is used for wired or wireless communication with external terminals. Wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. When the computer program is executed by the processor, it implements a method for statistical analysis of regional transformer imbalance losses. The display screen of the computer device can be a liquid crystal display (LCD) or an e-ink display. The input device of the computer device can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the computer device casing, or an external keyboard, touchpad, or mouse, etc.

[0132] This embodiment also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method proposed in the above embodiments.

[0133] The storage medium proposed in this embodiment belongs to the same inventive concept as the method proposed in the above embodiments. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0134] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the method of the embodiments of the present invention.

[0135] Example 2

[0136] Reference Figures 2-6 Based on the previous embodiment, this embodiment provides an application example of a method and system for statistical analysis of regional transformer imbalance additional losses, in order to verify and illustrate the technical effects adopted in this method.

[0137] This embodiment selects 800 transformer substations in a high-tech zone of a city for verification, and selects 150 typical substations to obtain time-series data. These substations are designated as data-rich substations. The phase line loss current and neutral line loss current of the data-rich substations are calculated as input data for the clustering stage. The optimal number of clusters k is determined by combining the sum of squares within the cluster (WCSS) and the elbow method. The final result is as follows. Figure 2 and Figure 3 As shown in Table 1, the WCSS values ​​for different cluster numbers k, and the k values ​​corresponding to the elbows are given.

[0138] Table 1: WCSS values ​​for different cluster numbers k, and the k value corresponding to the elbow point.

[0139]

[0140]

[0141] Depend on Figure 2As shown in Table 1, the reduction in WCSS decreases significantly when k=4, and this holds true for both clustering methods. Therefore, choosing the optimal number of clusters k=4 is appropriate. Based on determining the optimal number of clusters k=4, and using the proposed bi-objective optimization model, the clustering process was performed using k-means++ and hierarchical clustering algorithms. The results are shown in Table 2.

[0142] k-means++ Hierarchical clustering Profile coefficient 0.6517 0.7662 WCSS 0.3738 0.3476 Comprehensive indicators 0.6390 0.7093

[0143] In Table 2, when the number of clusters k=4, the comprehensive index value of the hierarchical clustering method is relatively large. Therefore, the combination of hierarchical clustering algorithm and k=4 is selected for clustering.

[0144] After dividing the data-rich transformer areas into several clusters, the feature vectors of the transformer areas are calculated, and the feature regions of each cluster are plotted, as shown in the figure. Figure 4 As shown in Table 3, the classification accuracy of KNN, decision tree, random forest, AdaBoost, and MSVM is compared respectively.

[0145] Classification methods accuracy KNN 0.8889 Decision Tree 0.8222 Random Forest 0.8667 AdaBoost 0.77.63 MSVM 0.8744

[0146] Table 3 shows that KNN and MSVM achieve higher classification accuracy than kAdaBoost and decision trees. KNN achieves the highest accuracy at 87.89%, followed by MSVM at 87.44%, and decision trees at 82.22%. Therefore, the KNN method with the highest accuracy is selected for the classification process, and the confusion matrix of the KNN classification results is plotted as follows. Figure 5 As shown in the figure, the confusion matrix illustrates the classification accuracy for each class in detail.

[0147] The additional losses due to imbalance in each cluster are also directly related to line resistance, which in turn is directly related to resistivity and line length. In my country, most low-voltage distribution lines use copper conductors with a resistivity of 1.68 × 10⁻⁶. -8 Based on the obtained information about the power grid structure of a high-tech zone in a certain city, the line lengths in the power distribution area range from 100m to 10000m. Therefore, the line resistance varies between 0.01Ω and 4Ω, leading to the confidence range of the unbalanced additional losses for each cluster, as shown in Ω·m. Figure 6 As shown.

[0148] The five-fold cross-validation method was used to verify the confidence range of the monthly imbalance-related additional loss. The verification results showed that 12% of the data-rich networks belonging to cluster 0 exceeded the confidence range for cluster 0, while 9% and 6% of the data-rich networks belonging to clusters 1 and 2, respectively, exceeded the confidence range. Cluster 3 has only 10 data-rich networks and is not suitable for data deletion based on confidence range. Therefore, the confidence range of imbalance-related additional loss for cluster 3 is the maximum range for this cluster. To verify the accuracy of the classification and imbalance-related additional loss estimation, several data-scarce areas were randomly selected for error verification. The verification results are shown in Table 4. The last column of the table defines the evaluation index for imbalance-related additional loss estimation.

[0149]

[0150] Among them, E real E represents the actual imbalance loss in the transformer area. mean E represents the average of the confidence range of the unbalanced additional loss. min This represents the lower limit of the confidence range for the unbalanced additional loss. When Eva < 1, it indicates that the actual unbalanced additional loss is within the confidence range; when Eva > 1, it indicates that the actual unbalanced additional loss exceeds the confidence range. In Table 4, experimental areas in groups 1, 2, and 3 were classified into incorrect clusters, resulting in actual unbalanced additional loss values ​​outside the confidence range; experimental areas in groups 4, 5, and 6 were classified into correct clusters, and their actual unbalanced additional loss values ​​were within the confidence range.

[0151] Table 4: Verification results.

[0152]

[0153] As can be seen from the above implementation results, this invention obtains detailed time-series current data by selecting some typical transformer substations and combining it with the average phase current data of the remaining transformer substations. It uses statistical methods such as Chebyshev's inequality to evaluate and improve the accuracy and confidence of the unbalanced additional loss range. This not only significantly reduces the cost and workload of data acquisition and processing, but also successfully achieves a comprehensive assessment of the total unbalanced additional loss of all transformer substations, including neutral and phase lines. This provides power supply companies with an economical, effective, accurate and reliable line loss assessment solution.

[0154] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

[0155] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of this application can be implemented using various computer languages.

[0156] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0157] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0158] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0159] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0160] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for statistical determination of unbalanced additional losses in a regional network, characterized by, Comprise: The extracted data-rich substation loss current is clustered by using a clustering algorithm, and a clustering evaluation index is used to minimize the weighted items of the silhouette coefficient and the cluster cluster sum; The feature vectors of the data-scarce substation are extracted, and the data-scarce substation is classified into the clusters of the clustering; Considering that the unbalanced additional loss estimation has outliers, a distance estimation method is used to realize the statistics of the unbalanced additional loss of the regional substation; The loss current of the data-rich substation is extracted, comprising: When the system has a three-phase imbalance problem, the line loss is obtained according to the relationship between the current sequence component and the phasor: wherein, are the positive, negative, and zero sequence currents of the three-phase currents when the system is unbalanced, respectively, R is the line resistance, is the neutral current; Further, according to the characteristic that the positive sequence current of the unbalanced system is equal to the current of the symmetrical system, the three-phase unbalanced additional loss is obtained: wherein, are the three phase currents, S is the three phase apparent power, U is the line voltage; a phase line loss current is obtained based on the three-phase unbalance additional loss and the neutral line loss current ; The extracted data-rich substation loss current is clustered by using a clustering algorithm, comprising: The phase and neutral loss currents are clustered into k clusters using an improved k-means algorithm and a hierarchical clustering algorithm and neutral loss currents into k clusters; The improved k-means clustering selects k initial clustering cluster centers, calculates the distance of each data point to each cluster center, and assigns it to the nearest cluster. For each cluster, the average distance between all data points in the cluster is calculated as the new cluster center. The assignment of data points and the recalculation of cluster centers are repeated until the cluster center no longer changes or the iteration converges. The hierarchical clustering algorithm takes each data point as a cluster cluster, and generates higher-level clusters by merging clusters with the minimum similarity to each other. The distance estimation method is used to realize the statistics of the unbalanced additional loss of the regional substation, comprising: The data-rich station area is divided into k clusters by clustering, and the maximum range of unbalanced additional loss of each cluster is obtained wherein, and respectively represent the minimum value and the maximum value of the unbalanced additional energy loss; The Chebyshev inequality is used to describe the unbalanced energy loss distribution of the cluster cluster, which is expressed as: wherein, X represents the unbalanced energy loss of the cell in the cluster cluster, represents the average of the unbalanced energy loss, represents the standard deviation of the unbalanced energy loss; For clustering a cluster m, the average unbalanced energy loss needs to be calculated, and the difference between the unbalanced energy loss of each data-rich substation and the average unbalanced energy loss is calculated. After deleting the data-rich substations with a difference greater than a first value, the range of the unbalanced energy loss of the remaining substations constitutes the unbalanced energy loss confidence range of the cluster m.

2. The method for statistical analysis of additional losses due to regional transformer imbalance as described in claim 1, characterized in that, The clustering evaluation index is used to minimize the weighted items of the silhouette coefficient and the cluster cluster sum, comprising: By taking the double-objective optimization model as the clustering evaluation index for minimizing the weighted items of the silhouette coefficient and the cluster cluster sum, the calculation of the double-objective optimization model is: wherein, C is a weighting factor, s(k) is the profile coefficient, and w(k) is the cluster cluster sum of squares.

3. The method for statistical analysis of additional losses due to regional transformer imbalance as described in claim 2, characterized in that, The feature vectors of the data-scarce substation are extracted, comprising: Defining a virtual loss current and a virtual three-phase unbalance As a feature vector, the formula is expressed as: wherein, respectively represent the average currents of the three phases A, B, C.

4. The method of claim 3, wherein the step of determining the unbalance additional loss of the regional cell comprises the steps of: determining the unbalance additional loss of the regional cell based on the unbalance additional loss of the regional cell and the unbalance additional loss of the local cell. The data-scarce substation is classified into the clusters of the clustering by using the K-nearest neighbor algorithm and the MSVM algorithm, comprising: For a sample to be classified, find the K nearest samples in the feature space, and vote according to the categories of the K samples. Finally, the sample to be classified is divided into the category with the most votes. The MSVM algorithm converts the multi-classification problem into multiple binary classification problems; The feature vectors and clustering labels are input into the classification model as training data. The classification accuracy of the two models is compared, and the classification model with higher accuracy is selected for data-scarce substation classification. The feature vectors of the data-scarce substation are input to realize the classification process of the substation.

5. A system for calculating the unbalance additional loss of a regional station area, which applies the method for calculating the unbalance additional loss of a regional station area according to any one of claims 1 to 4, characterized in that, Comprise: A clustering module is configured to cluster the extracted loss current of the data-rich substation by using a clustering algorithm, and to minimize the weighted items of the silhouette coefficient and the cluster cluster sum through a clustering evaluation index. A classification module is configured to classify the data-scarce area into the clusters by extracting a feature vector of the data-scarce area. A loss statistics module is configured to consider outliers in the unbalanced additional loss estimation and to realize the statistics of the unbalanced additional loss of the regional area by using a distance estimation method. 6.An electronic device, comprising: a memory and a processor; the memory is configured to store computer executable instructions, and the processor is configured to execute the computer executable instructions, and the computer executable instructions, when executed by the processor, realize the steps of the method for the statistics of the unbalanced additional loss of the regional area according to any one of claims 1 to 4. 7.A computer readable storage medium storing computer executable instructions, and the computer executable instructions, when executed by a processor, realize the steps of the method for the statistics of the unbalanced additional loss of the regional area according to any one of claims 1 to 4.

Images