Current transformer misalignment prediction method, device and equipment and storage medium

Abstract

The invention discloses a current transformer misalignment prediction method, device and equipment and a storage medium, and relates to the technical field of electric power measurement misalignment identification, and the method comprises the steps: obtaining the current data of a current transformer on each line measurement node in a power distribution network, and determining a current phasor corresponding to each current transformer through the current data; based on the current phasors corresponding to the current transformers and the Kirchhoff's current law, current unbalance analysis is carried out on the current transformers, and contribution values of the current transformers to the total current unbalance degree are determined according to obtained analysis results; and performing model training on an initial adaptive random forest model based on the current data and the contribution value corresponding to each current transformer, and triggering a corresponding current transformer misalignment prediction operation by using a determined target adaptive random forest model. According to the invention, the misaligned current transformer can be accurately predicted under the condition of adapting to the current data change.

Description

Technical Field

[0001] The present invention relates to the technical field of power metering misalignment identification, and in particular to a current transformer misalignment prediction method, device, equipment and storage medium. Background Art

[0002] Although current transformers are designed and manufactured with high precision in mind, they can still experience measurement errors in actual operation due to a variety of factors. These errors can cause protective devices to malfunction or fail to operate, seriously impacting the safe and stable operation of the power grid. Therefore, accurately predicting current transformer inaccuracies in advance is crucial.

[0003] However, the current prediction scheme analyzes the measurement data of three current transformers at the same measurement point in the distribution network, performs principal component analysis on the measurement data of the current transformer under the constraints of electrical physical correlation, and maps the error into However, due to the complex situation of distribution network users and large fluctuations in line current, the corresponding The statistics are different, which will lead to errors in the prediction results. Summary of the Invention

[0004] In view of this, the present invention aims to provide a current transformer misalignment prediction method, apparatus, device, and storage medium that can accurately predict misaligned current transformers while adapting to changes in current data, thereby reducing the incidence of power grid accidents caused by current transformer misalignment and improving the safety of power system operation. The specific solution is as follows:

[0005] In a first aspect, the present application provides a current transformer misalignment prediction method, comprising:

[0006] By acquiring current data of current transformers on each line measurement node in the distribution network, the current phasors corresponding to each current transformer are determined using the current data;

[0007] performing a current imbalance analysis on each current transformer based on the current phasor corresponding to each current transformer and Kirchhoff's current law, and determining a contribution value of each current transformer to a total current imbalance using the obtained analysis result;

[0008] An initial adaptive random forest model is trained based on the current data and the contribution values ​​respectively corresponding to each current transformer, and a corresponding current transformer misalignment prediction operation is triggered using the determined target adaptive random forest model.

[0009] Optionally, after obtaining the current data of the current transformer on each line measurement node in the distribution network, the method further includes:

[0010] Determine a corresponding first mean and a first standard deviation based on the current data corresponding to each of the current transformers; the current data includes a current amplitude and a current phase;

[0011] Determining a standard score corresponding to each of the current transformers using the first mean value and the first standard deviation;

[0012] respectively determining whether each of the standard scores is greater than a first preset threshold value to obtain a first determination result corresponding to each of the current transformers;

[0013] Based on each of the first judgment results, it is determined whether the corresponding current data is an abnormal value, and when the abnormal value exists, a corresponding abnormal value removal operation is triggered, so as to perform a current phasor conversion operation after completing the abnormal value removal operation.

[0014] Optionally, the performing current imbalance analysis on each current transformer based on the current phasor corresponding to each current transformer and Kirchhoff's current law includes:

[0015] Performing vector regression on the current of each line measurement node based on the current phasor corresponding to each current transformer, Kirchhoff's current law, and the rated transformation ratio of the current transformer corresponding to each line in the distribution network to obtain a vector regression equation corresponding to each line measurement node;

[0016] The current unbalance degree of the current transformer at each of the line measurement nodes is determined by accumulating the equation coefficients in each of the vector regression equations.

[0017] Optionally, performing vector regression on the current of each line measurement node based on the current phasor corresponding to each current transformer, Kirchhoff's current law, and the rated transformation ratio of the current transformer corresponding to each line in the distribution network includes:

[0018] Determine a current incoming line from each of the lines in the distribution network based on a non-repetitive screening principle, and determine a target primary-side current amplitude and a target secondary-side current amplitude corresponding to the current incoming line based on the current data of the current transformer corresponding to the current incoming line;

[0019] Determining a target primary side current phasor of the current incoming line based on the target primary side current amplitude, the current phasor corresponding to the current incoming line, the primary side current amplitudes corresponding to the remaining lines in each of the lines except the current incoming line, Kirchhoff's current law, and the initial current phases corresponding to the measurement nodes of each line;

[0020] Determining a target secondary-side current phasor of the current incoming line by using the target-side current phasor, the target secondary-side current amplitude, the rated transformation ratio and current transformer phase error of the current transformer corresponding to each of the lines, and the secondary-side current amplitudes corresponding to the remaining lines in each of the lines except the current incoming line;

[0021] Performing a vector regression fitting calculation based on the target primary-side current phasor and the target secondary-side current phasor of the current incoming line to obtain a vector regression equation corresponding to the line measurement node on the current incoming line;

[0022] Jump back to the step of determining the current incoming line from each of the lines in the distribution network based on the non-repetitive screening principle until the vector regression equations corresponding to the measurement nodes of each line are determined.

[0023] Optionally, determining the contribution value of each current transformer to the total current imbalance by using the obtained analysis result includes:

[0024] For any of the current transformers, determining a corresponding second standard deviation and a second mean based on an analysis result determined within a measurement time period;

[0025] A contribution value of the current transformer to the total current imbalance is determined based on a ratio between the second standard deviation and the second mean.

[0026] Optionally, the performing model training on the initial adaptive random forest model based on the current data and the contribution value respectively corresponding to each of the current transformers includes:

[0027] Constructing a data set based on the current data and the contribution value respectively corresponding to each current transformer, and marking the data in the data set with a current transformer state to determine a corresponding training set and a test set;

[0028] Constructing an initial adaptive random forest model, and training the initial adaptive random forest model using the training set to obtain the trained initial adaptive random forest model;

[0029] The trained initial adaptive random forest model is tested based on the test set to determine a target adaptive random forest model after the test passes.

[0030] Optionally, constructing an initial adaptive random forest model and training the initial adaptive random forest model using the training set includes:

[0031] By creating a set of initial decision trees to obtain an initial adaptive random forest model;

[0032] Extracting corresponding training subsets from the training set through each of the initial decision trees for training to obtain corresponding trained decision trees;

[0033] During the training process, the training set is segmented and the dynamic change of any two batches of data before and after the segmentation result is calculated, so as to detect the data change by determining whether the dynamic change is greater than a second preset threshold;

[0034] When a data change is detected, each of the trained decision trees is evaluated using the changed current batch data and a preset performance indicator to obtain an evaluation result;

[0035] Based on the evaluation results, a target decision tree with the worst performance is determined from each of the trained decision trees, and the target decision tree is trained using the current batch data to obtain the trained initial adaptive random forest model.

[0036] In a second aspect, the present application provides a current transformer misalignment prediction device, comprising:

[0037] a data acquisition module, configured to acquire current data of current transformers on each line measurement node in the distribution network, and determine current phasors corresponding to each current transformer using the current data;

[0038] an unbalance analysis module, configured to perform a current unbalance analysis on each current transformer based on the current phasor corresponding to each current transformer and Kirchhoff's current law, and determine a contribution value of each current transformer to a total current unbalance using the obtained analysis result;

[0039] The misalignment prediction module is used to perform model training on an initial adaptive random forest model based on the current data and the contribution value corresponding to each of the current transformers, and to trigger a corresponding current transformer misalignment prediction operation using a determined target adaptive random forest model.

[0040] In a third aspect, the present application provides an electronic device, comprising:

[0041] Memory, used to store computer programs;

[0042] A processor is used to execute the computer program to implement the steps of the above-mentioned current transformer misalignment prediction method.

[0043] In a fourth aspect, the present application provides a computer-readable storage medium for storing a computer program, wherein the computer program, when executed by a processor, implements the steps of the aforementioned current transformer misalignment prediction method.

[0044] As can be seen, in this application, current data from current transformers at each line measurement node in the distribution network is obtained, and the current phasors corresponding to each current transformer are determined using the current data. Current imbalance analysis is performed on each current transformer based on the current phasors corresponding to each current transformer and Kirchhoff's current law, and the contribution of each current transformer to the total current imbalance is determined using the obtained analysis results. An initial adaptive random forest model is trained based on the current data and contribution values ​​corresponding to each current transformer, and a target adaptive random forest model is used to trigger a corresponding current transformer misalignment prediction operation. Specifically, the application first obtains current data from current transformers at each line measurement node in the distribution network and converts them into current phasors. The current phasors are then used to analyze and quantify the contribution values ​​of each current transformer to the total current imbalance. The initial adaptive random forest model is then trained based on the current data and contribution values ​​of each current transformer, and the target adaptive random forest model obtained after training is used to predict current transformer misalignment. In this way, the misaligned current transformer can be accurately predicted while adapting to changes in current data, thereby reducing the incidence of power grid accidents caused by misaligned current transformers and improving the safety of power system operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0046] Figure 1 A flow chart of a current transformer misalignment prediction method provided in this application;

[0047] Figure 2 A flow chart of a specific current transformer misalignment prediction method provided in this application;

[0048] Figure 3 A schematic diagram of the structure of a current transformer misalignment prediction device provided in this application;

[0049] Figure 4 This is a structural diagram of an electronic device provided in this application. DETAILED DESCRIPTION

[0050] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0051] The current prediction scheme analyzes the measurement data of three current transformers at the same measurement point in the distribution network, performs principal component analysis on the measurement data of the current transformer under the constraints of electrical physical correlation, and maps the error into However, due to the complex situation of distribution network users and large fluctuations in line current, the corresponding The different statistics will lead to errors in the prediction results. Therefore, the present application provides a current transformer misalignment prediction solution that can accurately predict the misalignment of the current transformer while adapting to the changes in current data.

[0052] See also Figure 1 As shown, an embodiment of the present invention discloses a current transformer misalignment prediction method, comprising:

[0053] Step S11 : acquiring current data of current transformers on each line measurement node in the distribution network, and determining current phasors corresponding to each current transformer using the current data.

[0054] Specific, combined Figure 2 As shown, in this embodiment, to achieve high-precision measurement and real-time monitoring of current transformers, high-precision current transformers and other measuring equipment are installed at all line measurement nodes (including busbar nodes and all corresponding branch nodes) in the distribution network. These measuring equipment collects and acquires current data from the current transformers at each line measurement node in the distribution network in real time, ensuring that the current data at all line measurement nodes is synchronized in time. The collected current data is then converted into complex phasors to improve the convenience and efficiency of subsequent vector operations and imbalance calculations. Specifically, the collected current data is used to obtain corresponding current amplitude and phase information, which is then converted into complex current phasors to facilitate subsequent phasor analysis.

[0055] ;

[0056] Where, represents the current amplitude, Indicates the current phase, Indicates the true value of current, is the angle symbol.

[0057] Due to the performance of the measuring equipment and the influence of environmental factors, it is inevitable that some abnormal values ​​will appear during the data collection process. In addition, the calculation of the unbalance contribution of each current transformer depends on the accuracy of the data. Therefore, the abnormal data needs to be processed. That is, first, based on the current data corresponding to each current transformer, a corresponding first mean value and a first standard deviation are determined; the current data includes current amplitude and current phase; the first mean value and the first standard deviation are used to determine the standard score corresponding to each current transformer; each standard score is judged to be greater than a first preset threshold value to obtain a first judgment result corresponding to each current transformer; based on each first judgment result, it is determined whether the corresponding current data is an abnormal value, and when an abnormal value exists, a corresponding abnormal value removal operation is triggered, so that a current phasor conversion operation is performed after the abnormal value removal operation is completed.

[0058] Furthermore, regarding the specific abnormal data processing process, first, the mean value of the current data collected from each current transformer is calculated. and standard deviation , get the first mean and the first standard deviation. Next, calculate the value of each current data , that is, the standard score.

[0059] ;

[0060] Where, Indicates the first data point (i.e. current data of each current transformer), Indicates the The standard score of each data point. Then, a suitable threshold is selected to determine whether the data point is an outlier. The commonly used threshold is or , that is, if , the data point is considered an outlier and the corresponding outlier removal operation needs to be triggered. After removing all outliers, completing filtering and denoising, and eliminating interference and noise in the collected current data, the current amplitude and phase information are used to determine the current phasor to ensure data accuracy and reliability.

[0061] Step S12: performing current imbalance analysis on each current transformer based on the current phasor corresponding to each current transformer and Kirchhoff's current law, and determining a contribution value of each current transformer to the total current imbalance using the obtained analysis result.

[0062] Specifically, this embodiment considers that unbalanced current may be caused by current transformer measurement errors, load imbalance, or other factors. By analyzing the current imbalance, we can preliminarily determine which nodes may have current transformer misalignment issues. Contribution analysis is then used to quantify the contribution of each current transformer to the total current imbalance. The contribution of each transformer is calculated to quantify the current transformer misalignment status.

[0063] It should be understood that with respect to current imbalance analysis, vector regression is first performed on the current at each line measurement node based on the current phasors corresponding to each current transformer, Kirchhoff's current law, and the rated transformation ratios of the current transformers corresponding to each line in the distribution network, thereby obtaining vector regression equations corresponding to each line measurement node. The current imbalance of the current transformer at each line measurement node is then determined by accumulating the coefficients in the vector regression equations. That is, in this embodiment, each line is first traversed and calculated sequentially as an incoming line to obtain the dispersion of the ratio differences of the branch line transformers after removing the ratio differences of the incoming line transformers. The current imbalance of the current transformers at each line measurement node is then determined by accumulating the coefficients.

[0064] Furthermore, in this embodiment, regarding current vector regression, the current incoming line is determined from each of the lines in the distribution network based on the principle of non-repetitive screening, and the target primary-side current amplitude and target secondary-side current amplitude corresponding to the current incoming line are determined according to the current data of the current transformer corresponding to the current incoming line; the target primary-side current phasor of the current incoming line is determined based on the target primary-side current amplitude, the current phasor corresponding to the current incoming line, the primary-side current amplitude corresponding to the remaining lines in each of the lines except the current incoming line, Kirchhoff's current law, and the initial current phase corresponding to the measurement node of each line; the target primary-side current phasor of the current incoming line is determined using the target primary-side current phasor , the target secondary current amplitude, the rated transformation ratio of the current transformer and the current transformer phase error corresponding to each of the lines, and the secondary current amplitude corresponding to the remaining lines in each of the lines except the current line determine the target secondary current phasor of the current incoming line; based on the target primary current phasor and the target secondary current phasor of the current incoming line, perform vector regression fitting calculation to obtain the vector regression equation corresponding to the line measurement node on the current incoming line; jump back to the step of determining the current incoming line from each of the lines in the distribution network based on the principle of non-repetitive screening until the vector regression equation corresponding to each of the line measurement nodes is determined. Regarding the specific calculation process, the true value of the current of the primary line is recorded as:

[0065] (1);

[0066] Where, Indicates the true value of the current in the primary side circuit; is the plural symbol; Indicates the amplitude of the true value of the current in the primary side circuit, that is, the primary side current amplitude; Indicates the phase of the true value of the current in the primary line.

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

[0068] (2);

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

[0070] (3);

[0071] The primary current phasor representation of the line measurement node on the bus under the KCL law (Kirchhoff current law) is as follows:

[0072] (4);

[0073] Where, Indicates the primary side current amplitude of the busbar; Indicates branch line The primary side current amplitude; Indicates the initial phase of the line measurement node on the bus; Indicates branch line The initial phase of the current at the line measurement node on the .

[0074] Considering the conversion of the mutual inductor error to the secondary side result:

[0075] (5);

[0076] Where, Indicates the secondary side current amplitude of the bus current transformer; Indicates branch line The secondary current amplitude of the current transformer; Indicates the Phase error of current transformer at each line measurement node; Indicates the rated transformation ratio of the busbar current transformer; Indicates branch line The rated transformation ratio of the current transformer; Indicates the ratio difference of busbar measurement current transformer, express The current transformer ratio difference of each branch line is measured. Since the phase error is small, this embodiment ignores the phase error and calculates the vector regression fitting calculation formula, that is, the vector regression equation:

[0077] (6);

[0078] (7);

[0079] Where, Represents the bias term of the multiple linear regression equation; Respectively represent branches The regression coefficient corresponding to each mutual inductor; Indicates the angular velocity of the current phase change; Indicates the time point of current data collection. By performing ergodic vector regression on the current of each line measurement node and then accumulating the obtained ergodic regression equation coefficients, the measurement imbalance of the current transformer of each line measurement node is obtained. ( ), the calculation formula is:

[0080] (8);

[0081] After analyzing the above formula, we can see that by traversing each line as an incoming line and performing the calculation, we can obtain the dispersion of the branch line transformer ratio difference, excluding the incoming line transformer ratio difference. The constraint condition is that the number of transformers with metering anomalies in the entire line is less than half of the total number of transformers. The smaller the dispersion coefficient between the coefficients, the larger the ratio difference of the incoming line transformer. Therefore, the inverse of the dispersion coefficient can be used as the imbalance contribution of the incoming line transformer to the overall transformer ratio difference. The calculation formula is:

[0082] (9);

[0083] Where, Indicates the corresponding current transformer unbalance obtained by traversing and calculating the measurement time period The standard deviation of Indicates the corresponding current transformer unbalance obtained during the traversal calculation measurement period The average value of It means the The contribution value of the current transformer of each line measurement node to the total current imbalance. That is, the contribution value of each current transformer to the total current imbalance is determined by using the obtained analysis result, including: for any current transformer, determining the corresponding second standard deviation based on the analysis result determined within the measurement time period and the second mean ; Determine the contribution value of the current transformer to the total current imbalance based on the ratio between the second standard deviation and the second mean.

[0084] Furthermore, according to the above formula, the current transformer ratio difference with the current node as the parent node has the greatest impact on the contribution value obtained by each traversal calculation, so The larger the value is, the greater the current transformer ratio difference of the parent node of the current vector regression is, which may be an inaccurate transformer, thus realizing the quantitative representation of the inaccurate current transformer.

[0085] Step S13: performing model training on an initial adaptive random forest model based on the current data and the contribution value corresponding to each of the current transformers, and triggering a corresponding current transformer misalignment prediction operation using the determined target adaptive random forest model.

[0086] Specifically, in this embodiment, the model structure and parameters of the traditional random forest remain fixed after training. This means that the model cannot adapt to changes in data distribution, especially in a data stream environment where the data distribution changes over time. In other words, the traditional random forest cannot detect and adapt to such changes, which may lead to a decline in model performance. However, the adaptive random forest can better adapt to changes in the data stream by dynamically adjusting the model, maintaining the accuracy and adaptability of the model. Therefore, this embodiment introduces a dynamically adjusted ensemble learning method to dynamically update and optimize the random forest model in the data volume. That is, this embodiment first needs to construct an optimized random forest model, namely the initial adaptive random forest model.

[0087] It should be understood that, considering the periodicity of the electricity consumption habits of distribution network users, the corresponding current data will change periodically, and the corresponding recognition model needs to update the model parameters in real time according to the changes in the current data, so as to improve the robustness of the recognition model. First, a set of initial decision trees are created. These trees can be obtained by training with subsets randomly extracted from the training data. That is, by creating a set of initial decision trees, an initial adaptive random forest model is obtained. Through each of the initial decision trees, a corresponding training subset is extracted from the training set for training to obtain a corresponding trained decision tree. Afterwards, the data stream is segmented and cut, and the statistics of each data segment are calculated. When the statistics of the two sub-windows before and after exceed a certain threshold, it is considered that the data stream has undergone concept drift. That is, during the training process, the training set is segmented and cut, and the dynamic change of any two batches of data before and after the segmentation result is calculated to detect the data change by judging whether the dynamic change is greater than the second preset threshold. If drift is subsequently detected, the performance of each tree is evaluated using the current mini-batch data. Based on the evaluation results, trees with poor performance are selected and retrained using the current mini-batch data. The corresponding decision tree structures are reset or updated, allowing the decision trees comprising the random forest to update their parameters in real time based on the dynamic changes in the data stream. In other words, when a data change is detected, each trained decision tree is evaluated using the changed current batch data and preset performance metrics to obtain an evaluation result. Based on the evaluation result, a target decision tree with the worst performance is determined from each trained decision tree, and the target decision tree is trained using the current batch data to obtain the trained initial adaptive random forest model.

[0088] Furthermore, the specific implementation steps for building the initial adaptive random forest model are as follows:

[0089] 1) Create an initial random forest.

[0090] Assume there is an initial training dataset ,in, For the The misalignment characteristic of the current transformer of each line measurement node, is the corresponding label. Let the number of trees in the random forest be .

[0091] 2) Data stream segmentation.

[0092] If the data flow ,have data points, and the length of each small batch of data is , the data flow Divide Small batches .

[0093] 3) Detect dynamic changes in data flow

[0094] Assume the current mini-batch is ( ), the previous subwindow is , the dynamic judgment threshold is , calculate the dynamic change of the two batches of data before and after, if , it is considered that a dynamic change has occurred. The specific calculation formula is as follows:

[0095] (10);

[0096] (11);

[0097] Where, Represents the mean of the current small batch data, represents the mean of the previous mini-batch data, Indicates the dynamic change of the current small batch data.

[0098] 4) Evaluate the performance of each decision tree in the random forest using accuracy Evaluate the performance of each tree. The specific steps are: first use the current mini-batch Evaluate the performance of each decision tree, for each mini-batch , calculate the accuracy of each decision tree . Select the decision tree with the lowest accuracy and use the current mini-batch Retrain the decision tree model to achieve adaptive updates to the decision trees in the random forest and enhance classification accuracy.

[0099] (12);

[0100] Where, is the true positive class, that is, the number of samples that are actually positive and predicted as positive by the random forest model; True negative examples, that is, the number of samples that are actually negative and predicted as negative by the model; False positives are the number of samples that are actually negative but predicted as positive by the model; is the number of false negative examples, that is, the number of samples that are actually positive but predicted as negative by the model.

[0101] In this embodiment, the above steps can be used to obtain an optimized random forest model, namely the initial adaptive random forest model. That is, this embodiment proposes an improved random forest algorithm that dynamically adjusts the decision tree parameters in the random forest based on the constantly changing characteristics of the current data. This algorithm can adapt to changes in the data stream in real time and ensure the stable operation of the recognition system. Features can then be extracted from the current data, including current amplitude, phase, and unbalance contribution. Labels are generated based on the known current transformer status (normal or misaligned), and the data set is divided into a training set and a test set. The initial adaptive random forest model is trained using the training set data. The trained target adaptive random forest model is then used to obtain the current transformer contribution value (i.e., unbalance contribution), current amplitude, and phase based on the real-time collected current data for classification, thereby predicting misaligned current transformers.

[0102] In summary, this embodiment calculates the contribution of each current transformer to the total current imbalance, quantifies the error impact of each transformer, and then classifies the imbalance of each current transformer using the optimized random forest algorithm to identify misaligned current transformers.

[0103] In addition, in order to verify the effectiveness of the prediction method proposed in this embodiment, in a specific implementation method, the default fixed error of the test is selected to be less than It is a normal transformer, greater than or equal to The transformers with abnormal errors are marked according to the rule, and then identified using the algorithm proposed in this patent. The identification labels are compared with the actual labels to analyze the identification results, as shown in Table 1 below.

[0104] Table 1

[0105]

[0106] As can be seen, in this application, current data from current transformers at each line measurement node in the distribution network is obtained, and the current phasors corresponding to each current transformer are determined using the current data. Current imbalance analysis is performed on each current transformer based on the current phasors corresponding to each current transformer and Kirchhoff's current law, and the contribution of each current transformer to the total current imbalance is determined using the obtained analysis results. An initial adaptive random forest model is trained based on the current data and contribution values ​​corresponding to each current transformer, and a target adaptive random forest model is used to trigger a corresponding current transformer misalignment prediction operation. Specifically, the application first obtains current data from current transformers at each line measurement node in the distribution network and converts them into current phasors. The current phasors are then used to analyze and quantify the contribution values ​​of each current transformer to the total current imbalance. The initial adaptive random forest model is then trained based on the current data and contribution values ​​of each current transformer, and the target adaptive random forest model obtained after training is used to predict current transformer misalignment. In this way, the misaligned current transformer can be accurately predicted while adapting to changes in current data, thereby reducing the incidence of power grid accidents caused by misaligned current transformers and improving the safety of power system operation.

[0107] See also Figure 3 As shown, the embodiment of the present application also discloses a current transformer misalignment prediction device, including:

[0108] A data acquisition module 11 is configured to acquire current data of current transformers on each line measurement node in the distribution network, and determine current phasors corresponding to each current transformer using the current data;

[0109] an unbalance analysis module 12, configured to perform a current unbalance analysis on each current transformer based on the current phasor corresponding to each current transformer and Kirchhoff's current law, and determine a contribution value of each current transformer to a total current unbalance using the obtained analysis result;

[0110] The misalignment prediction module 13 is used to perform model training on an initial adaptive random forest model based on the current data and the contribution value corresponding to each of the current transformers, and trigger a corresponding current transformer misalignment prediction operation using a determined target adaptive random forest model.

[0111] It can be seen that the present application first obtains the current data of the current transformer on each line measurement node in the distribution network and converts it into a current phasor. Then, the current phasor is used to analyze and quantify the contribution value of each current transformer to the total current imbalance. Afterwards, the initial adaptive random forest model is trained based on the current data and contribution value of each current transformer, and the target adaptive random forest model obtained after training is used to predict the misalignment of the current transformer. In this way, the misaligned current transformer can be accurately predicted under the condition of adapting to the changes in current data, thereby reducing the incidence of power grid accidents caused by misalignment of the current transformer and improving the safety of power system operation.

[0112] In some specific embodiments, the current transformer misalignment prediction device may further include:

[0113] a first data processing module, configured to determine a corresponding first mean value and a first standard deviation based on the current data corresponding to each of the current transformers; the current data including current amplitude and current phase;

[0114] a score determination module, configured to determine a standard score corresponding to each of the current transformers using the first mean value and the first standard deviation;

[0115] a first judgment module, configured to respectively judge whether each of the standard scores is greater than a first preset threshold value, so as to obtain a first judgment result corresponding to each of the current transformers;

[0116] An outlier removal module is used to determine whether the corresponding current data is an outlier based on each of the first judgment results, and trigger a corresponding outlier removal operation when the outlier exists, so as to perform a current phasor conversion operation after completing the outlier removal operation.

[0117] In some specific embodiments, the imbalance analysis module 12 may specifically include:

[0118] a current vector regression submodule, configured to perform vector regression on the current of each line measurement node based on the current phasor corresponding to each current transformer, Kirchhoff's current law, and the rated transformation ratio of the current transformer corresponding to each line in the distribution network, so as to obtain a vector regression equation corresponding to each line measurement node;

[0119] The coefficient accumulation submodule is used to determine the current imbalance of the current transformer at each of the line measurement nodes by accumulating the equation coefficients in each of the vector regression equations.

[0120] In some specific embodiments, the current vector regression submodule may specifically include:

[0121] an incoming line selection unit, configured to determine a current incoming line from each of the lines in the distribution network based on a non-repetitive screening principle, and determine a target primary-side current amplitude and a target secondary-side current amplitude corresponding to the current incoming line based on the current data of the current transformer corresponding to the current incoming line;

[0122] a primary-side current phasor determination unit, configured to determine a target primary-side current phasor of a current incoming line based on the target primary-side current amplitude, the current phasor corresponding to the current incoming line, the primary-side current amplitudes corresponding to the remaining lines in each of the lines except the current incoming line, Kirchhoff's current law, and the initial current phase corresponding to each of the line measurement nodes;

[0123] a secondary-side current phasor determination unit, configured to determine a target secondary-side current phasor of a current incoming line by using the target secondary-side current phasor, the target secondary-side current amplitude, the rated transformation ratio and the current transformer phase error corresponding to each of the lines, and the secondary-side current amplitudes corresponding to the remaining lines in each of the lines except the current incoming line;

[0124] a vector regression equation determination unit, configured to perform a vector regression fitting calculation based on a target primary-side current phasor and a target secondary-side current phasor of a current incoming line, so as to obtain a vector regression equation corresponding to the line measurement node on the current incoming line;

[0125] The step jump unit is used to jump back to the step of determining the current incoming line from each of the lines in the distribution network based on the non-repetitive screening principle until the vector regression equations corresponding to the measurement nodes of each line are determined.

[0126] In some specific embodiments, the imbalance analysis module 12 may specifically include:

[0127] a second data processing unit, configured to determine, for any of the current transformers, a corresponding second standard deviation and a second mean value based on an analysis result determined within a measurement time period;

[0128] A contribution value determining unit is configured to determine a contribution value of the current transformer to the total current imbalance based on a ratio between the second standard deviation and the second mean value.

[0129] In some specific embodiments, the misalignment prediction module 13 may specifically include:

[0130] a data set construction submodule, configured to construct a data set based on the current data and the contribution value corresponding to each current transformer, and mark the data in the data set with a current transformer state to determine a corresponding training set and a test set;

[0131] An adaptive model training submodule, configured to construct an initial adaptive random forest model and train the initial adaptive random forest model using the training set to obtain the trained initial adaptive random forest model;

[0132] The target model determination unit is used to test the trained initial adaptive random forest model based on the test set, so as to determine the target adaptive random forest model after the test passes.

[0133] In some specific embodiments, the adaptive model training submodule may specifically include:

[0134] An initial model building unit, used to obtain an initial adaptive random forest model by creating a set of initial decision trees;

[0135] A decision tree training unit, configured to extract corresponding training subsets from the training set through each of the initial decision trees for training, so as to obtain corresponding trained decision trees;

[0136] a data change detection unit, configured to segment the training set during training and calculate a dynamic change between any two batches of data before and after the segmentation result, so as to detect data changes by determining whether the dynamic change is greater than a second preset threshold;

[0137] A decision tree evaluation unit is used to evaluate each of the trained decision trees using the changed current batch data and preset performance indicators when a data change is detected to obtain an evaluation result;

[0138] A training completion unit is used to determine a target decision tree with the worst performance from each of the trained decision trees based on the evaluation result, and train the target decision tree using the current batch data to obtain the trained initial adaptive random forest model.

[0139] Furthermore, the embodiment of the present application also discloses an electronic device, Figure 4 This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content in the diagram should not be considered as any limitation to the scope of application of the present application.

[0140] Figure 4This is a schematic diagram of the structure of an electronic device 20 provided in an embodiment of the present application. The electronic device 20 may specifically include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 is used to store a computer program, which is loaded and executed by the processor 21 to implement the relevant steps of the current transformer misalignment prediction method disclosed in any of the aforementioned embodiments. Furthermore, the electronic device 20 in this embodiment may specifically be an electronic computer.

[0141] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and the external device. The communication protocol it follows is any communication protocol that can be applied to the technical solution of this application and is not specifically limited here; the input and output interface 25 is used to obtain external input data or output data to the outside world. Its specific interface type can be selected according to specific application needs and is not specifically limited here.

[0142] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or CD, etc. The resources stored thereon can include an operating system 221, a computer program 222, etc., and the storage method can be temporary storage or permanent storage.

[0143] The operating system 221 is used to manage and control the hardware devices and computer program 222 on the electronic device 20, and can be Windows Server, Netware, Unix, Linux, etc. In addition to including a computer program capable of implementing the current transformer misalignment prediction method performed by the electronic device 20 disclosed in any of the aforementioned embodiments, the computer program 222 can further include a computer program capable of implementing other specific tasks.

[0144] Furthermore, the present application also discloses a computer-readable storage medium for storing a computer program; wherein, when executed by a processor, the computer program implements the aforementioned current transformer misalignment prediction method. The specific steps of this method can be referred to the corresponding contents disclosed in the aforementioned embodiments and will not be repeated here.

[0145] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from the other embodiments. Reference can be made to the descriptions of the identical or similar parts between the various embodiments. For the devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple, and the relevant parts can be referred to the descriptions of the methods.

[0146] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the components and steps of each example according to their functions. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0147] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in random access memory (RAM), internal memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0148] Finally, it should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or device comprising the element.

[0149] The above is a detailed introduction to the technical solution provided by the present application. Specific examples are used herein to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea. At the same time, for those skilled in the art, according to the ideas of the present application, there may be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A current transformer misalignment prediction method, characterized in that: include: By acquiring current data of current transformers on each line measurement node in the distribution network, the current phasors corresponding to each current transformer are determined using the current data; performing a current imbalance analysis on each current transformer based on the current phasor corresponding to each current transformer and Kirchhoff's current law, and determining a contribution value of each current transformer to a total current imbalance using the obtained analysis result; An initial adaptive random forest model is trained based on the current data and the contribution values ​​respectively corresponding to each current transformer, and a corresponding current transformer misalignment prediction operation is triggered using the determined target adaptive random forest model.

2. The current transformer misalignment prediction method according to claim 1, characterized in that: After obtaining the current data of the current transformer on each line measurement node in the distribution network, the method further includes: Determine a corresponding first mean and a first standard deviation based on the current data corresponding to each of the current transformers; the current data includes a current amplitude and a current phase; Determining a standard score corresponding to each of the current transformers using the first mean value and the first standard deviation; respectively determining whether each of the standard scores is greater than a first preset threshold value to obtain a first determination result corresponding to each of the current transformers; Based on each of the first judgment results, it is determined whether the corresponding current data is an abnormal value, and when the abnormal value exists, a corresponding abnormal value removal operation is triggered, so as to perform a current phasor conversion operation after completing the abnormal value removal operation.

3. The current transformer misalignment prediction method according to claim 1, characterized in that: The performing current imbalance analysis on each current transformer based on the current phasor corresponding to each current transformer and Kirchhoff's current law includes: Performing vector regression on the current of each line measurement node based on the current phasor corresponding to each current transformer, Kirchhoff's current law, and the rated transformation ratio of the current transformer corresponding to each line in the distribution network to obtain a vector regression equation corresponding to each line measurement node; The current unbalance degree of the current transformer at each of the line measurement nodes is determined by accumulating the equation coefficients in each of the vector regression equations.

4. The current transformer misalignment prediction method according to claim 3, characterized in that: The performing vector regression on the current of each line measurement node based on the current phasor corresponding to each current transformer, Kirchhoff's current law, and the rated transformation ratio of the current transformer corresponding to each line in the distribution network includes: Determine a current incoming line from each of the lines in the distribution network based on a non-repetitive screening principle, and determine a target primary-side current amplitude and a target secondary-side current amplitude corresponding to the current incoming line based on the current data of the current transformer corresponding to the current incoming line; Determining a target primary side current phasor of the current incoming line based on the target primary side current amplitude, the current phasor corresponding to the current incoming line, the primary side current amplitudes corresponding to the remaining lines in each of the lines except the current incoming line, Kirchhoff's current law, and the initial current phases corresponding to the measurement nodes of each line; Determining a target secondary-side current phasor of the current incoming line by using the target-side current phasor, the target secondary-side current amplitude, the rated transformation ratio and current transformer phase error of the current transformer corresponding to each of the lines, and the secondary-side current amplitudes corresponding to the remaining lines in each of the lines except the current incoming line; Performing a vector regression fitting calculation based on the target primary-side current phasor and the target secondary-side current phasor of the current incoming line to obtain a vector regression equation corresponding to the line measurement node on the current incoming line; Jump back to the step of determining the current incoming line from each of the lines in the distribution network based on the non-repetitive screening principle until the vector regression equations corresponding to the measurement nodes of each line are determined.

5. The current transformer misalignment prediction method according to claim 3, characterized in that: Determining the contribution value of each current transformer to the total current imbalance by using the obtained analysis result includes: For any of the current transformers, determining a corresponding second standard deviation and a second mean based on an analysis result determined within a measurement time period; A contribution value of the current transformer to the total current imbalance is determined based on a ratio between the second standard deviation and the second mean.

6. The current transformer misalignment prediction method according to any one of claims 1 to 5, characterized in that: The performing model training on the initial adaptive random forest model based on the current data and the contribution value respectively corresponding to each of the current transformers includes: Constructing a data set based on the current data and the contribution value respectively corresponding to each current transformer, and marking the data in the data set with a current transformer state to determine a corresponding training set and a test set; Constructing an initial adaptive random forest model, and training the initial adaptive random forest model using the training set to obtain the trained initial adaptive random forest model; The trained initial adaptive random forest model is tested based on the test set to determine a target adaptive random forest model after the test passes.

7. The current transformer misalignment prediction method according to claim 6, characterized in that: The constructing of the initial adaptive random forest model and training the initial adaptive random forest model using the training set includes: By creating a set of initial decision trees to obtain an initial adaptive random forest model; Extracting corresponding training subsets from the training set through each of the initial decision trees for training to obtain corresponding trained decision trees; During the training process, the training set is segmented and the dynamic change of any two batches of data before and after the segmentation result is calculated, so as to detect the data change by determining whether the dynamic change is greater than a second preset threshold; When a data change is detected, each of the trained decision trees is evaluated using the changed current batch data and a preset performance indicator to obtain an evaluation result; Based on the evaluation results, a target decision tree with the worst performance is determined from each of the trained decision trees, and the target decision tree is trained using the current batch data to obtain the trained initial adaptive random forest model.

8. A current transformer misalignment prediction device, characterized in that: include: a data acquisition module, configured to acquire current data of current transformers on each line measurement node in the distribution network, and determine current phasors corresponding to each current transformer using the current data; an unbalance analysis module, configured to perform a current unbalance analysis on each current transformer based on the current phasor corresponding to each current transformer and Kirchhoff's current law, and determine a contribution value of each current transformer to a total current unbalance using the obtained analysis result; The misalignment prediction module is used to perform model training on an initial adaptive random forest model based on the current data and the contribution value corresponding to each of the current transformers, and to trigger a corresponding current transformer misalignment prediction operation using a determined target adaptive random forest model.

9. An electronic device, characterized in that: include: Memory, used to store computer programs; A processor, configured to execute the computer program to implement the current transformer misalignment prediction method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that Used to store a computer program, which, when executed by a processor, implements the current transformer misalignment prediction method according to any one of claims 1 to 7.

Images