Distribution network metering current transformer measurement state evaluation method based on improved isolated forest

Abstract

The invention relates to an improved isolated forest-based current transformer measurement state evaluation method for distribution network metering. The method comprises the following specific steps of constructing a current model with n branch lines under a single bus; constructing a multiple linear regression equation based on the Kirchhoff's current law, and establishing a relation equation between the ratio error of the current transformer and each regression coefficient; the change rule of each regression coefficient under different current transformer measurement abnormal conditions is analyzed; and calculating the confidence coefficient of the multiple linear regression equation obtained by calculation, counting the change statistics of each regression coefficient, and positioning the abnormal current transformer through an improved isolated forest algorithm. According to the invention, the improved isolated forest algorithm is provided to realize the identification of the abnormal regression coefficient, and the positioning of the abnormal mutual inductor is realized.

Description

Technical Field

[0001] The present application relates to the field of current transformer measurement status evaluation, and in particular to a method for evaluating the measurement status of a current transformer for distribution network metering based on an improved isolation forest. Background Art

[0002] As the country continues to pay attention to the construction of intelligent power grids, investment in smart grids is also increasing. The boundaries between primary and secondary equipment in the traditional sense are gradually blurred, and the intelligence level of various power equipment is constantly improving. The current transformer accurately provides the current signal of the primary line to the detection equipment on the secondary circuit. It is a key equipment to ensure the stability, reliability, fairness and justice of the power system. It is the most widely used and widely used secondary power equipment. According to statistics, as of the end of 2021, the number of transformers for measurement in my country exceeded 100 million sets, and it is increasing by 5% each year. Faced with the huge and rapidly growing transformer calibration requirements in my country, traditional manual calibration consumes a lot of manpower and material resources, and is inefficient, resulting in the current huge cost but still a large number of equipment overdue for inspection.

[0003] Under my country's new power system reforms, the scale of electricity trading is steadily increasing, and a fair and equitable electricity trading environment is crucial. The key to electricity metering lies in the accuracy and stability of the current transformer secondary circuit signal, but existing technologies cannot fully assess the transformer's measurement status in real time. Therefore, the development of accurate and efficient current transformer measurement status verification methods to ensure fair and equitable distribution network electricity trading is an urgent need in the current power system reforms. Summary of the Invention

[0004] The purpose of the embodiment of the present application is to provide a method for evaluating the measurement status of current transformers for distribution network metering based on an improved isolation forest. Based on Kirchhoff's current law, a multivariate linear regression equation for the relationship between busbar and branch line currents is constructed. By analyzing the changes in the regression coefficients of each item in the multivariate linear regression equation, the normal measurement status and abnormal measurement status of the current transformer are distinguished, thereby realizing the status judgment of the high-voltage current transformer in the distribution network.

[0005] To achieve the above objectives, the present invention provides a method for evaluating the measurement status of a current transformer for distribution network metering based on an improved isolation forest, comprising the following specific steps:

[0006] Construct a current model with n branches under a single bus;

[0007] Based on Kirchhoff's current law, a multivariate linear regression equation is constructed to establish the relationship between the current transformer ratio difference and each regression coefficient;

[0008] Analyze the changing patterns of various regression coefficients under different abnormal current transformer measurement conditions;

[0009] The confidence of the multivariate linear regression equation is calculated, and the statistical changes of each regression coefficient are counted. Then, the abnormal current transformer is located using the improved isolation forest algorithm.

[0010] After constructing the current model of n branches under a single bus, the relationship between the primary current of the line and the secondary current of the current transformer is derived according to the rated transformation ratio, ratio difference and angle difference of the current transformer in the line.

[0011] The true value of the current in the primary circuit is recorded as:

[0012]

[0013] Where, represents the true value of the current in the primary side circuit, I represents the amplitude of the true value of the current in the primary side circuit, θ represents the phase of the true value of the current in the primary side circuit,

[0014] Assume that the primary current on the line is The true value of the secondary current measurement of the corresponding current transformer on the line is The ratio difference of the current transformer is k, K is the rated transformation ratio, then the ratio difference expression is

[0015]

[0016] The primary side current value of the line can be expressed as follows through formula (1):

[0017]

[0018] Similarly, the expression of the primary side current value of the total busbar can be obtained as follows:

[0019]

[0020] According to Kirchhoff's current law, we have:

[0021]

[0022] Where, represents the primary side current value of the kth branch, Indicates the total bus current value, K S and k S Respectively represent the rated transformation ratio and ratio difference of the total bus current transformer,

[0023] According to (2), (3) and (4), the relationship between the primary current and the transformer secondary current is:

[0024]

[0025] Based on Kirchhoff's current law, a multivariate linear regression equation is constructed to establish the relationship between the current transformer ratio difference and each regression coefficient. Specifically,

[0026] Assuming the difference between the regression value and the actual value is b, the multivariate linear expression of the sum current and the partial current based on Kirchhoff's law as the constraint condition is:

[0027]

[0028]

[0029] Let the variables in each term in formula (7) be have to

[0030]

[0031] Multiple linear regression analysis was performed between the sum current and the current of each branch line

[0032]

[0033] Where, represents the true value of the secondary current of the sum current of the kth sample, and βj represents the regression coefficient of the jth branch current; Indicates the true value of the secondary side current measurement of the j-th branch at the k-th sample.

[0034] According to the relationship between the regression coefficient β and the current transformer ratio k, set The multivariate linear equation is shown in formula (11), and the weight fluctuation matrix H is constructed through the regression coefficients to realize the diagnosis of abnormal transformers.

[0035]

[0036]

[0037] The analysis of the changing rules of various regression coefficients under different current transformer measurement abnormalities is as follows: when a branch current transformer has abnormal measurement data, the regression coefficient of the corresponding item will be too high or too low, and the regression coefficients corresponding to other normal branches will fluctuate around the rated transformation ratio, and the floating range is proportional to the error amount of the faulty branch; if the busbar current transformer produces an abnormal error, the regression coefficients corresponding to all branches will have a certain change, and the upper and lower floating range is large; if the busbar and a branch produce abnormal errors at the same time, the fluctuation of the abnormal branch will be masked by the busbar fluctuation, resulting in the inability to identify the abnormal branch; if all branch transformers are abnormal at the same time, the corresponding regression coefficients will be reduced accordingly, and the smaller the ratio difference range, the smaller the distinction between abnormal branches.

[0038] The confidence of the calculated multivariate linear regression equation is calculated, and the statistical changes of each regression coefficient are counted. Then, the abnormal current transformer is located by the improved isolation forest algorithm. Specifically,

[0039] (1) First, determine the fluctuation of the regression coefficient under normal conditions. Since the changes of the regression coefficients in the corresponding multivariate linear regression equations are different when errors occur in the busbar and branch current transformers, an adaptive fluctuation coefficient is proposed. The calculation formula is:

[0040]

[0041] Where βi represents the regression coefficient of the i-th term, β represents the multiple linear regression coefficient under ideal conditions, α represents the regression coefficient fluctuation threshold under normal conditions, and ε represents the fluctuation coefficient of each regression line coefficient;

[0042] (2) The fluctuation coefficient ε is input into the isolation forest algorithm as a feature, and the standard coefficient α is added as the isolation forest benchmark value;

[0043] (3) Establishing the fitness function of the isolated forest ridge algorithm

[0044]

[0045] Where, They represent the average precision and difference of the input data set, γ and η represent the weights of precision and difference respectively;

[0046] (4) Using each regression coefficient as input to identify abnormal points, determine the abnormal current transformer.

[0047] Compared with the prior art, the present invention has the following advantages:

[0048] Based on Kirchhoff's current law, the present invention constructs a multiple linear regression equation for the relationship between busbar and branch line currents. By analyzing the changes in the regression coefficients of each item in the multiple linear regression equation, the normal measurement state and abnormal measurement state of the current transformer are distinguished, thereby realizing the state judgment of the high-voltage current transformer in the distribution network. An improved isolation forest algorithm is proposed to realize the identification of abnormal regression coefficients and the positioning of abnormal transformers. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without creative work.

[0050] Figure 1 A flow chart of the method of this application;

[0051] Figure 2 This is a current model diagram with n branches under a single bus;

[0052] Figure 3 is the relationship diagram of the current i of each branch and the current I;

[0053] Figure 4 Schematic diagram of the improved isolation forest algorithm. DETAILED DESCRIPTION

[0054] The technical solutions in the embodiments of the present application will be described below in conjunction with the accompanying drawings. It should be noted that similar reference numerals and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings.

[0055] The terms "comprises," "comprising," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not preclude the presence of additional identical elements in the process, method, article, or apparatus that includes the element.

[0056] The terms "first," "second," etc. are only used to distinguish one entity or operation from another entity or operation, and are not to be understood as indicating or implying relative importance, nor are they to be understood as requiring or implying any actual relationship or order between these entities or operations.

[0057] like Figure 1 As shown, a method for evaluating the measurement status of a current transformer for distribution network metering based on an improved isolation forest method includes the following specific steps:

[0058] S1. Construct a current model with n branches under a single busbar;

[0059] S2. Construct a multivariate linear regression equation based on Kirchhoff's current law and establish the relationship between the current transformer ratio error and each regression coefficient;

[0060] S3. Analyze the changing patterns of various regression coefficients under different abnormal current transformer measurement conditions;

[0061] S4. Calculate the confidence of the calculated multivariate linear regression equation, calculate the statistical changes of each regression coefficient, and then locate the abnormal current transformer using the improved isolation forest algorithm.

[0062] Construct a current model with n branches under a single busbar. Figure 2 As shown,

[0063] 1.1 Derivation of CT secondary current on branch lines

[0064] The true value of the current in the primary circuit is recorded as:

[0065]

[0066] Where, represents the true value of the current in the primary side line, I represents the amplitude of the true value of the current in the primary side line, and θ represents the phase of the true value of the current in the primary side line.

[0067] Assume that the primary current on the line is The true value of the secondary current measurement of the corresponding current transformer on the line is The ratio difference of the current transformer is k, K is the rated transformation ratio, then the ratio difference expression is

[0068]

[0069] The primary side current value of the line can be expressed as follows through formula (1):

[0070]

[0071] 1.2 Derivation of CT secondary current on bus line

[0072] Similarly, the expression of the primary side current value of the total busbar can be obtained as follows:

[0073]

[0074] According to Kirchhoff's current law, we have:

[0075]

[0076] Where, represents the primary side current value of the kth branch, IS represents the total bus current value, Ks and kS represent the rated transformation ratio and ratio difference of the total bus current transformer respectively.

[0077] According to (2), (3) and (4), the relationship between the sum current and the branch current can be obtained as follows:

[0078]

[0079] Based on Kirchhoff's current law, a multivariate linear regression equation is constructed to establish the relationship between the current transformer ratio difference and each regression coefficient.

[0080] Assuming that the sum current value and the branch current value are accurate, the relationship between the branch current i and the sum current I is as follows: Figure 3 shown

[0081] Assuming the difference between the regression value and the actual value is b, the multivariate linear expression of the sum current and the partial current based on Kirchhoff's law as the constraint condition is:

[0082]

[0083]

[0084] Let the variables in each term in formula (7) be have to

[0085]

[0086] Multiple linear regression analysis was performed between the sum current and the current of each branch line

[0087]

[0088] Where, represents the true value of the secondary current of the sum current of the kth sample, and βj represents the regression coefficient of the jth branch current; Indicates the true value of the secondary side current measurement of the j-th branch at the k-th sample.

[0089] Determine whether β0 is within the specified threshold. If β0 is not within the threshold range and there are abnormal current transformers, the system is considered unstable and further analysis is required to determine the abnormal state of each branch current transformer. If β0 is within the threshold range, the system is considered normal.

[0090] According to the relationship between the regression coefficient β and the current transformer ratio k, set The multivariate linear equation is shown in Formula 11, and the weight fluctuation matrix H is constructed through the regression coefficients to realize the diagnosis of abnormal transformers.

[0091]

[0092]

[0093] Analyze the changing rules of various regression coefficients under different abnormal current transformer measurement conditions:

[0094] When a branch current transformer measures abnormal data, the corresponding regression coefficient will be biased high or low. The corresponding regression coefficients of other normal branches will fluctuate around the rated transformation ratio, with the fluctuation range being proportional to the error of the faulty branch. If a busbar current transformer produces an abnormal error, the corresponding regression coefficients of all branches will also change to a certain extent, with a large fluctuation range. If the busbar and a branch simultaneously produce abnormal errors, the fluctuation of the abnormal branch will be masked by the busbar fluctuation, making it impossible to identify the abnormal branch. If all branch current transformers are abnormal at the same time, the corresponding regression coefficients will be reduced accordingly. The smaller the ratio difference range, the lower the ability to distinguish abnormal branches.

[0095] When there is no error in the busbar current, if there is an error in any branch current, its corresponding regression coefficient will be significantly different from the regression coefficients of other normal branches, which can effectively identify the abnormal branch.

[0096] In the case of errors in the bus current, all regression coefficients will fluctuate greatly. If there are measurement anomalies in the branch current transformer at the same time, accurate identification will be impossible.

[0097] The confidence of the multivariate linear regression equation is calculated, and the statistical changes of each regression coefficient are counted. Then, the abnormal current transformer is located by the improved isolation forest algorithm.

[0098] Isolation Forest is a tree-based anomaly detection algorithm with high detection capabilities applicable to multiple fields. The Isolation Forest algorithm primarily uses the length of the path from an evaluation point to its isolated state as an evaluation metric. The closer the distance to the isolated state, the greater the probability that the data point is judged as an outlier. Because the algorithm does not require density and probability calculations, it is relatively fast. However, the traditional Isolation Forest algorithm has limitations in certain specific areas. For example, in distribution network electricity metering, current fluctuations on busbars and branches can cause different paths for data points. Failure to consider the relationship between current fluctuations on different lines and data points can lead to reduced algorithm recognition accuracy.

[0099] Improved Isolation Forest Algorithm Figure 4 As shown in the figure, by introducing the standard judgment point, since the rated transformation ratio of the transformer used for distribution network metering is determined at the factory and will not change, in order to ensure that the transformer can operate under normal conditions, it is necessary to calculate whether the rated transformation ratio of the transformer is within a certain range, rather than generating a large error according to the change of the current in the line. By introducing the range interval and variation amplitude under normal random error conditions, a benchmark model is constructed, which effectively suppresses the change of the data point path caused by the current in the line. The change fluctuation and the benchmark difference are used as the characteristics of the isolation forest to improve the isolation forest algorithm.

[0100] (1) First, determine the fluctuation of the regression coefficient under normal conditions. Since the changes of the regression coefficients in the corresponding multivariate linear regression equations are different when errors occur in the busbar and branch current transformers, an adaptive fluctuation coefficient is proposed. The calculation formula is:

[0101]

[0102] Where βi represents the i-th regression coefficient, β represents the multiple linear regression coefficient under ideal conditions, α represents the regression coefficient fluctuation threshold under normal conditions, and ε represents the fluctuation coefficient of each regression line coefficient.

[0103] The fluctuation coefficient ε is input into the isolation forest algorithm as a feature, and the standard coefficient α is added as the isolation forest benchmark value;

[0104] Establishing the fitness function of the Isolated Forest Ridge algorithm

[0105]

[0106] Where, They represent the average accuracy and difference of the input data set, γ and η represent the weights of accuracy and difference, respectively.

[0107] (4) Using each regression coefficient as input to identify abnormal points, determine the abnormal current transformer.

[0108] The above description is merely an embodiment of the present application and is not intended to limit the scope of protection of the present application. For those skilled in the art, various modifications and variations of the present application are possible. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A method for evaluating the measurement status of a current transformer for distribution network metering based on improved isolation forest, characterized in that: The specific steps include: Construct a current model with n branches under a single bus; Based on Kirchhoff's current law, a multivariate linear regression equation is constructed to establish the relationship equation between the current transformer ratio difference and each regression coefficient. Analyze the changing rules of various regression coefficients under different abnormal current transformer measurement conditions; The obtained multivariate linear regression equation is calculated, the change statistics of each regression coefficient are counted, and then the abnormal current transformer is located using the improved isolation forest algorithm.

2. The method for evaluating the measurement status of a current transformer for distribution network metering based on improved isolation forest according to claim 1, characterized in that: After constructing the current model of n branches under a single bus, the relationship between the primary current of the line and the secondary current of the current transformer is derived according to the rated transformation ratio, ratio difference and angle difference of the current transformer in the line.

3. The method for evaluating the measurement status of current transformer for distribution network metering based on improved isolation forest according to claim 2, characterized in that: The true value of the current in the primary line is recorded as: In the formula, represents the true value of the current in the primary side line, I represents the amplitude of the true value of the current in the primary side line, θ represents the phase of the true value of the current in the primary side line, Assume that the primary current value on the line is The true value of the secondary current measurement of the corresponding current transformer on the line is The ratio difference of the current transformer is k, K is the rated transformation ratio, then the ratio difference expression is The primary current value of the line can be expressed as follows through formula (1): Similarly, the expression of the primary side current value of the total busbar can be obtained as follows: According to Kirchhoff's current law, we have: In the formula, represents the primary current value of the kth branch line, Indicates the primary current value of the total busbar, K S and k S They represent the rated transformation ratio and ratio difference of the total bus current transformer respectively. According to (2), (3) and (4), the relationship between the primary current and the transformer secondary current is: Where Kn represents the rated transformation ratio of the transformer of the nth branch line, and kn represents the ratio difference of the transformer of the nth branch line.

4. The method for evaluating the measurement status of current transformers for distribution network metering based on improved isolation forest according to claim 2, characterized in that: Based on Kirchhoff's current law, a multivariate linear regression equation is constructed to establish the relationship between the current transformer ratio difference and each regression coefficient. Specifically, Assuming the difference between the regression value and the actual value is b, the multivariate linear expression of the sum current and the partial current based on Kirchhoff's law as a constraint condition is: Let the variables in each term in formula (7) be have to Multiple linear regression analysis was performed between the current and the current of each branch line In the formula, represents the true value of the secondary current of the sum current of the kth sample, βj represents the regression coefficient of the jth branch current; Represents the true value of the secondary current measurement of the j-th branch line at the k-th sample.

5. The method for evaluating the measurement status of current transformers for distribution network metering based on improved isolation forest according to claim 4, characterized in that: According to the relationship between the regression coefficient β and the current transformer ratio k, set The multivariate linear equation is shown in formula (11), and the weight fluctuation matrix H is constructed through various regression coefficients to realize the diagnosis of abnormal transformers.

6. The method for evaluating the measurement status of current transformers for distribution network metering based on improved isolation forest according to claim 1, characterized in that: The analysis of the changing rules of various regression coefficients under different current transformer measurement abnormalities is as follows: when a branch current transformer has abnormal measurement data, the regression coefficient of the corresponding item will be too high or too low, and the corresponding regression coefficients of other normal branches will fluctuate around the rated transformation ratio, and the floating range is proportional to the error amount of the faulty branch; if the bus current transformer has an abnormal error, the corresponding regression coefficients of all branches will have a certain change, and the floating range is large; if the bus and a branch have abnormal errors at the same time, the fluctuation of the abnormal branch will be masked in the bus fluctuation, resulting in the inability to identify the abnormal branch; if all branch transformers are abnormal at the same time, the corresponding regression coefficients will be reduced accordingly, and the smaller the ratio difference range, the smaller the distinction between abnormal branches.

7. The method for evaluating the measurement status of current transformers for distribution network metering based on improved isolation forest according to claim 4, characterized in that: The confidence of the calculated multivariate linear regression equation is calculated, and the statistical changes of each regression coefficient are counted. Then, the abnormal current transformer is located by the improved isolation forest algorithm. Specifically, (1) First, determine the fluctuation of the regression coefficient under normal conditions. When the busbar and branch current transformers produce errors, the corresponding regression coefficients in the multivariate linear regression equation change differently. Therefore, an adaptive fluctuation coefficient is proposed. The calculation formula is: In the formula, βi represents the regression coefficient of the i-th item, β represents the multivariate linear regression coefficient under ideal conditions, α represents the regression coefficient fluctuation threshold under normal conditions, and ε represents the fluctuation coefficient of each regression line coefficient; (2) The fluctuation coefficient ε is input into the isolation forest algorithm as a feature, and the standard coefficient α is added as the isolation forest benchmark value; (3) Establish the fitness function of the isolated forest ridge algorithm In the formula, They represent the average precision and difference of the input data set, respectively, and γ and η represent the weights of precision and difference, respectively; (4) Taking each regression coefficient as input to identify abnormal points, determine the abnormal current transformer.

Images