Power distribution network line attribute parameter identification method, system, equipment and medium
By pre-constructing the feasible domain matrix of power operation parameters and the data correction model, the problems of accuracy and practicality of the identification results of distribution network line attribute parameters are solved, and accurate judgment and high-precision identification of abnormal data are achieved, which is applicable to complex data distribution.
Patent Information
- Application Number
- CN202511658882.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-13
- Publication Date
- 2026-02-24
AI Technical Summary
In existing technologies, the accuracy and practicality of the identification results of power distribution network line attribute parameters are affected by the lack and anomalies of real-time power operation parameter data, resulting in problems such as voltage monitoring distortion, line loss calculation deviation and protection device setting inaccuracy, which cannot meet the needs of refined operation of power distribution networks.
An outlier identification and imputation of real-time data is performed using a pre-constructed feasible domain matrix of power operation parameters. Feature vectors are obtained by combining Gaussian kernel function and nonparametric kernel density theory. Least square regression and random forest models are used for data correction. Finally, a neural network model containing physical constraint loss terms of power flow equations is constructed for parameter identification.
It improves the accuracy and practicality of power distribution network line attribute parameter identification, is applicable to complex non-normal data distribution, effectively identifies abnormal data, and improves parameter identification accuracy.
Smart Images
Figure CN121561702A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power distribution network management technology, and in particular to a method, system, device and medium for identifying power distribution network line attribute parameters. Background Technology
[0002] As a crucial link connecting the transmission network and the user side in the power system, the distribution network's line attribute parameters, mainly including resistance and reactance, serve as the foundational data for distribution network operation monitoring, dispatch control, protection, and planning design. However, with extended operation, factors such as conductor aging and oxidation, changes in ambient temperature and humidity, load fluctuations, and line icing or corrosion gradually cause deviations between actual line parameters and initial design / nameplate parameters. Continuing to use the initial parameters for distribution network operation management will lead to problems such as distorted voltage monitoring, inaccurate line loss calculations, and inaccurate protection device settings. In severe cases, this may cause safety hazards such as line overload, voltage exceeding limits, or even fault escalation, failing to meet the requirements for refined distribution network operation.
[0003] In existing technologies, line attribute parameter identification is generally based on real-time power operation parameter data of distribution network nodes. This involves using real-time power operation parameter data collected by the distribution network measurement system to construct mathematical models and inversely deduce line attribute parameters. However, due to factors such as short-term failures of distribution network-side measurement equipment, external interference, and transmission errors, the collected real-time power operation parameter data may be incomplete or anomaly-prone, thus affecting the accuracy of the distribution network line attribute parameter identification results. Furthermore, existing methods for identifying distribution network line attribute parameters often use mathematical models that construct loss functions based on "minimizing the error loss between the identified result and the true value." This singular optimization objective can lead to discrepancies in the physical laws underlying the identified distribution network line attribute parameter identification results, rendering them ineffective for practical application.
[0004] Therefore, improving the accuracy and practicality of the identification results of power distribution network line attribute parameters has become a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention
[0005] This invention provides a method, system, device, and medium for identifying the attribute parameters of power distribution network lines, in order to solve the technical problem of how to improve the accuracy and practicality of the identification results of the attribute parameters of power distribution network lines, and to achieve the effect of improving the accuracy and practicality of the identification results of the attribute parameters of power distribution network lines.
[0006] In a first aspect, the present invention provides a method for identifying the attribute parameters of a power distribution network line, the method comprising: A real-time dataset of power operation parameters of the target distribution network is obtained, and an outlier is identified in the real-time dataset of power operation parameters using a pre-constructed feasible domain matrix of power operation parameters. The outlier dataset is determined. The feasible domain matrix of power operation parameters is set to be obtained by correcting the upper and lower limits of the historical dataset of power operation parameters based on the feature vector of the historical dataset of power operation parameters of the target distribution network and the maximum permissible deviation of power operation parameters. The abnormal dataset is filled with abnormal data to obtain a real-time dataset for power operation parameter correction; The real-time dataset of power operation parameter correction is input into the pre-constructed distribution network line attribute parameter identification model to obtain the line attribute parameter identification result of the target distribution network. The first loss function of the distribution network line attribute parameter identification model includes at least a power flow equation physical constraint loss term. The power flow equation physical constraint loss term is set to reflect the deviation between the actual value of the line voltage amplitude difference and the estimated value of the line voltage amplitude difference obtained based on the line attribute parameter identification result.
[0007] Preferably, the step of acquiring a real-time dataset of power operation parameters of the target distribution network and using a pre-constructed feasible domain matrix of power operation parameters to identify outliers in the real-time dataset of power operation parameters and determine outlier datasets includes: A historical dataset of power operation parameters for several historical scheduling days of the target distribution network is obtained, and the Gaussian kernel function and nonparametric kernel density theory are used to process the historical dataset of power operation parameters to obtain the data probability density function of each historical data of power operation parameter in the historical dataset of power operation parameters at each time segment. Based on the data probability density function, the maximum probability density vector is obtained. According to the maximum probability density vector, a data similarity matching algorithm is used to obtain the weight of the historical dataset of power operation parameters corresponding to each historical scheduling day relative to the maximum probability density vector. The historical power operation parameter dataset is represented by a vector according to the weights to obtain the feature vector of the historical power operation parameter dataset. The upper and lower limits of each time segment of the historical power operation parameter dataset are then corrected according to the feature vector to obtain the initial feasible domain matrix of power operation parameters. Based on the maximum permissible deviation of the power operation parameters, the initial power operation parameter feasible region matrix is corrected twice to obtain the power operation parameter feasible region matrix. The feasible domain matrix of the power operation parameters is used to identify outliers in the real-time dataset of power operation parameters and determine the outlier dataset.
[0008] Preferably, the step of obtaining the weight of the historical dataset of power operation parameters corresponding to each historical scheduling day relative to the maximum probability density vector using a data similarity matching algorithm based on the maximum probability density vector includes: Calculate the Euclidean distance between the maximum probability density vector and the historical data of power operation parameters corresponding to each time segment of each historical scheduling day to obtain the intermediate variable corresponding to each historical scheduling day. The intermediate variables are normalized to obtain the weights of the historical dataset of power operation parameters corresponding to each historical scheduling day relative to the maximum probability density vector.
[0009] Preferably, the step of correcting the upper and lower limits of each time segment of the historical power operation parameter dataset based on the feature vector to obtain the initial feasible region matrix of power operation parameters includes: The maximum and minimum values of the historical dataset of power operation parameters at each time segment are identified to obtain the maximum and minimum values of the historical dataset of power operation parameters at each time segment. Based on the eigenvectors, a linear normalization algorithm is used to correct each maximum and minimum value once to obtain the initial feasible domain matrix of power operation parameters.
[0010] Preferably, the step of imputing abnormal data in the abnormal dataset to obtain a real-time dataset for power operation parameter correction includes: Based on a preset data length threshold, the abnormal dataset is divided into short-cycle abnormal data and long-cycle abnormal data. Select several adjacent time points of the short-period abnormal data and their corresponding neighborhood historical measurement data, and construct a regression model based on the neighborhood historical measurement data; The second loss function of the regression model is constructed using the least squares regression algorithm, and the parameters of the second loss function are solved. The regression model after parameter solving is then used to fill in the short-cycle abnormal data to obtain the initial power operation parameter correction real-time dataset. The initial real-time dataset for power operation parameter correction was filled in using a pre-trained random forest model to obtain the real-time dataset for power operation parameter correction.
[0011] Preferably, the method of using a pre-trained random forest model to fill in the initial real-time power operation parameter correction dataset to obtain a real-time power operation parameter correction dataset includes: The historical dataset of power operation parameters is resampled based on the self-sampling algorithm to obtain several subsets of historical measurement data; Training is performed on each subset of historical measurement data to obtain several decision trees. A random forest model is then constructed based on the decision trees. The time aspect of the long-period anomaly data is input into each of the random forest models, and the real-time dataset for power operation parameter correction is obtained based on the imputation values of the long-period anomaly data output by the random forest models.
[0012] Preferably, the step of inputting the real-time dataset of power operation parameter correction into a pre-built distribution network line attribute parameter identification model to obtain the line attribute parameter identification result of the target distribution network includes: Model the parameter identification error of the distribution network and construct the parameter identification error loss term; Model the active power, reactive power, voltage amplitude, and distribution network line attribute parameters in the real-time power operation parameter dataset to obtain the estimated value of the voltage amplitude difference between the two ends of the line. The deviation between the actual value of the line voltage amplitude difference and the estimated value of the line voltage amplitude difference corresponding to the nodes at both ends of the line is modeled in a regularized manner to obtain the physical constraint loss term of the power flow equation. Based on the parameter identification error loss term and the power flow equation physical constraint loss term, a first loss function is constructed. Based on the first loss function, the pre-constructed neural network model is trained to obtain the distribution network line attribute parameter identification model. The training dataset used for training has the same characteristics as the real-time dataset for power operation parameter correction. The real-time dataset of power operation parameter correction is input into the distribution network line attribute parameter identification model to obtain the line attribute parameter identification result of the target distribution network.
[0013] Secondly, the present invention also provides a distribution network line attribute parameter identification system to implement the above-mentioned distribution network line attribute parameter identification method. The system includes: an abnormal data identification module, an abnormal data correction module, and a line attribute parameter identification module. The abnormal data identification module is used to acquire the real-time data set of power operation parameters of the target distribution network, and to identify outliers in the real-time data set of power operation parameters using a pre-constructed feasible domain matrix of power operation parameters to determine the abnormal data set. The feasible domain matrix of power operation parameters is set to be obtained by correcting the upper and lower limits of the historical data set of power operation parameters based on the feature vector of the historical data set of power operation parameters of the target distribution network and the maximum permissible deviation of power operation parameters. The abnormal data correction module is used to fill in the abnormal data in the abnormal dataset to obtain a real-time dataset for power operation parameter correction. The line attribute parameter identification module is used to input the real-time dataset of power operation parameter correction into a pre-built distribution network line attribute parameter identification model to obtain the line attribute parameter identification result of the target distribution network. The first loss function of the distribution network line attribute parameter identification model includes at least a power flow equation physical constraint loss term. The power flow equation physical constraint loss term is set to reflect the deviation between the actual value of the line voltage amplitude difference and the estimated value of the line voltage amplitude difference obtained based on the line attribute parameter identification result.
[0014] Thirdly, the present invention also provides a computer device, the computer device including a memory, a processor and a transceiver, which are connected to each other via a bus; the memory is used to store a set of computer program instructions and data, and to transmit the stored data to the processor, the processor executing the computer program instructions stored in the memory to execute the above-described method for identifying the attribute parameters of power distribution network lines.
[0015] Fourthly, the present invention also provides a computer-readable storage medium storing a computer program that, when executed, implements the above-described method for identifying the attribute parameters of power distribution network lines.
[0016] This application provides a method, system, device, and medium for identifying the attribute parameters of power distribution network lines. Compared with the prior art, the beneficial effects of the embodiments of this application are as follows: The method for identifying power distribution network line attribute parameters disclosed in this application obtains a real-time dataset of power operation parameters of the target power distribution network, and uses a pre-constructed feasible domain matrix of power operation parameters to identify outliers in the real-time dataset, thus determining the outlier dataset. The feasible domain matrix of power operation parameters is set as the feature vector of the historical dataset of power operation parameters of the target power distribution network and the maximum permissible deviation of power operation parameters, and is obtained after correcting the upper and lower limits of the historical dataset of power operation parameters. The least squares regression algorithm and a pre-trained random forest model are used to impute outlier data in the outlier dataset in turn, thus obtaining a real-time dataset of corrected power operation parameters. The real-time dataset of corrected power operation parameters is input into a pre-constructed power distribution network line attribute parameter identification model to obtain the identification result of the line attribute parameters of the target power distribution network. The first loss function of the power distribution network line attribute parameter identification model includes at least a power flow equation physical constraint loss term, which is set to reflect the deviation between the actual value of the line voltage amplitude difference and the estimated value of the line voltage amplitude difference obtained based on the identification result of the line attribute parameters. The distribution network line attribute parameter identification method disclosed in this application uses Gaussian kernel function and nonparametric kernel density theory to model the historical dataset of power operation parameters to obtain the feature vector of the historical dataset. This feature vector is used to correct the upper and lower limits of each time segment of the historical dataset. Then, based on the maximum permissible deviation of the power operation parameters, the initial feasible region matrix of the power operation parameters is corrected a second time to obtain the feasible region matrix of the power operation parameters. This allows for accurate identification of abnormal data in the real-time dataset of power operation parameters. It does not require the assumption that the data follows a characteristic distribution and is applicable to complex and nonnormal data distributions. It can effectively identify abnormal data in the real-time dataset of power operation parameters, providing accurate basic data for the identification of distribution network line attribute parameters. The first loss function of the distribution network line attribute parameter identification model considers both parameter identification error and physical constraints of power flow equations, improving the identification accuracy of the obtained distribution network line attribute parameter identification model. Attached Figure Description
[0017] Figure 1 This is a schematic diagram of the steps of a method for identifying the attribute parameters of a power distribution network line according to a preferred embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of a power distribution network line attribute parameter identification system provided in a preferred embodiment of the present invention; Figure 3 This is an internal structural diagram of the computer device in an embodiment of the present invention; Figure label: 1-Abnormal data identification module, 2-Abnormal data correction module, 3-Line attribute parameter identification module. Detailed Implementation
[0018] The embodiments of the present invention are described in detail below with reference to the accompanying drawings. The embodiments are provided for illustrative purposes only and should not be construed as limiting the scope of the invention. The accompanying drawings are for reference and illustration only and do not constitute a limitation on the scope of protection of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of this invention.
[0019] In the description of this invention, it should be noted that, unless otherwise defined, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in this specification is for the purpose of describing specific embodiments only and is not intended to limit the invention. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.
[0020] Please see Figure 1 The diagram illustrates the steps of a method for identifying attribute parameters of a power distribution network line. In an embodiment of the present invention, a method for identifying attribute parameters of a power distribution network line is provided, the method comprising: S1. Obtain the real-time dataset of power operation parameters of the target distribution network, and use a pre-constructed feasible domain matrix of power operation parameters to identify outliers in the real-time dataset of power operation parameters, and determine the outlier dataset. The feasible domain matrix of power operation parameters is set to be obtained by correcting the upper and lower limits of the historical dataset of power operation parameters based on the feature vector of the historical dataset of power operation parameters of the target distribution network and the maximum permissible deviation of power operation parameters. In a preferred embodiment of this application, both the real-time dataset of power operation parameters and the historical dataset of power operation parameters are collected using existing distribution network measurement equipment. The historical dataset of power operation parameters is used to construct the feasible domain matrix of power operation parameters. Specifically, the full-cycle power operation parameter measurement data of the target distribution network for several historical scheduling days are obtained. Each historical scheduling day has 96 time segments with a time interval of 15 minutes. The power operation parameter measurement data of each time segment includes at least: line active power, line reactive power and voltage amplitude of the nodes at both ends of the line. The resulting historical dataset of power operation parameters is represented as follows: in, A vector representing historical data of power operation parameters. Indicates the first A vector composed of historical power operation parameter data at each time segment of a historical scheduling day. Indicates the first The first historical dispatch day Historical data of power operation parameters at various time points. This indicates the number of historical scheduling days.
[0021] The probability density function of historical power operation parameters at each time endpoint is calculated using Gaussian kernel function and nonparametric kernel density theory. The probability density function is expressed as follows: in, This indicates the historical data of power operation parameters. The probability density function of each time segment. Indicates the independent variable. This represents the window width parameter.
[0022] The window width parameter controls the smoothness of the probability density function curve. A larger d results in a "wider" kernel function, with more overlap in the contribution ranges of different samples, leading to a smoother final probability density function curve, but potentially losing local details of the data. Conversely, a smaller d results in a "narrower" kernel function, with a more concentrated contribution range for each sample, preserving more local fluctuations in the final probability density function curve, but potentially leading to distortion due to overfitting noise. In a preferred embodiment of this application, the window width parameter is expressed as follows: in, This represents the Gaussian kernel function expression.
[0023] Nonparametric kernel density theory does not require prior assumptions that data follows a specific distribution, such as a normal or exponential distribution. It learns density patterns directly from historical data, making it more flexible in adapting to the distribution patterns of real data. The Gaussian kernel function, chosen based on the normal distribution, has the advantages of good smoothness and simple mathematical properties, resulting in a smoother final probability density curve.
[0024] Based on the data probability density function, the maximum probability density value for each time segment is obtained, and the maximum probability density vector is derived from the maximum probability density value. The maximum probability density vector is represented as follows: in, This represents a vector composed of the maximum probability density values at each time segment. Indicates the first The maximum probability density value of a time segment, that is, the maximum value corresponding to the data probability density function.
[0025] Calculate the Euclidean distance between the maximum probability density vector and the historical power operation parameter data corresponding to each time segment of each historical scheduling day to obtain the intermediate variable corresponding to each historical scheduling day; normalize the intermediate variable to obtain the weight of the historical power operation parameter dataset corresponding to each historical scheduling day relative to the maximum probability density vector, and the weight is expressed as follows: in, Indicates the first The weights of the historical dataset of power operation parameters corresponding to each historical scheduling day relative to the maximum probability density vector. Indicates the first Intermediate variables corresponding to each historical scheduling day This represents an adjustable parameter between 0 and 1.
[0026] Furthermore, based on the weights, the historical data of power operation parameters in the historical power operation parameter dataset are adjusted to obtain the feature vector of the historical power operation parameter dataset. The feature vector is represented as follows: in, The feature vector representing the historical dataset of power operation parameters.
[0027] In a preferred embodiment of this application, the upper and lower limits of each time segment of the historical power operation parameter dataset are corrected based on the feature vector. The upper and lower limits of each time segment of the historical power operation parameter dataset are the maximum and minimum values of each time segment of the historical power operation parameter dataset. Therefore, the upper and lower limits of the historical power operation parameter dataset are expressed as follows: in, This represents a vector consisting of the maximum values of historical power operation parameters at each time point in the historical power operation parameter dataset. Indicates the first The maximum value of historical data of power operation parameters at a given time point. This represents a vector composed of the minimum values of historical power operation parameters at each time point in the historical power operation parameter dataset. Indicates the first The minimum value of historical power operation parameters for a given time period.
[0028] Based on the eigenvectors, a linear normalization algorithm is used to correct each maximum and minimum value, resulting in the upper and lower limits of the initial measurement data for each time segment, expressed as follows: in, express The One portion, Indicates the first The upper limit of the initial measurement data after one correction at each time section. Indicates the first The lower limit of the initial measurement data after one correction at each time section.
[0029] Furthermore, based on the maximum permissible deviation of power operation parameters, the upper and lower limits of the initial measurement data are corrected twice, respectively, to obtain the upper and lower limits of the measurement data for each time segment, as shown below: in, Indicates the first The upper limit of measurement data after secondary correction at each time section. Indicates the first The lower limit of measurement data after secondary correction for each time section. This indicates the maximum permissible deviation of power operation parameters.
[0030] Based on the upper and lower limits of the measured data corresponding to the second correction at each time section, a feasible region matrix of power operation parameters is formed, which is expressed as: in, Represents the feasible region matrix of power operation parameters. This indicates the upper limit of the measurement data after secondary correction at the first time section. This indicates the upper limit of the measurement data after secondary correction at the 96th time section. Indicates the first The upper limit of measurement data after secondary correction at each time section. This represents the lower limit of the measurement data after secondary correction at the first time section. Indicates the first The lower limit of measurement data after secondary correction for each time section. This indicates the lower limit of the measurement data after the second correction at the 96th time section.
[0031] The feasible domain matrix of power operation parameters is used to identify outliers in the real-time power operation parameter dataset and determine the outlier dataset. For each time segment of the real-time power operation parameter data, if it is within the range of the feasible domain matrix of power operation parameters, it is considered normal data; if it is not within the range of the feasible domain matrix of power operation parameters, it is considered outlier data. In this way, the outlier dataset of the real-time power operation parameter dataset is obtained.
[0032] In a preferred embodiment of this application, a Gaussian kernel function and nonparametric kernel density theory are used to model the historical dataset of power operation parameters to obtain the feature vector of the historical dataset. This feature vector is then used to correct the upper and lower limits of each time segment of the historical dataset. Finally, the initial feasible region matrix of power operation parameters is corrected a second time based on the maximum permissible deviation of the power operation parameters to obtain the feasible region matrix of power operation parameters. This allows for accurate identification of abnormal data in the real-time dataset of power operation parameters without assuming that the data follows a characteristic distribution. It is suitable for complex and nonnormal data distributions and can effectively identify abnormal data in the real-time dataset of power operation parameters, providing accurate basic data for the identification of distribution network line attribute parameters.
[0033] S2. Abnormal data imputation is performed on the abnormal dataset to obtain a real-time dataset for power operation parameter correction. In a preferred embodiment of this application, the identified abnormal dataset is deleted. Based on a preset data length threshold, the abnormal dataset is divided into short-cycle abnormal data and long-cycle abnormal data. The data length threshold is 5. Abnormal data intervals less than or equal to the data length threshold are classified as short-cycle abnormal data, and abnormal data intervals greater than the data length threshold are classified as long-cycle abnormal data. For short-cycle abnormal data, a least squares regression algorithm is used for imputation. Specifically, several adjacent time endpoints of the short-cycle abnormal data are selected from the corresponding neighborhood historical measurement data. Based on the neighborhood historical measurement data, a regression model is constructed. The regression model is expressed as: in, An index representing a time segment of historical measurement data in the neighborhood. This represents the index of the data point corresponding to each time segment in the historical measurement data of the neighborhood. Indicates the first in the neighborhood historical measurement data Measurement data at each time section, These represent the parameters of the regression model. The first historical measurement data in the neighborhood represents the... The first time section Measurement values of each data point.
[0034] The second loss function of the regression model is constructed using the least squares regression algorithm. The parameters of the second loss function are then solved by minimizing this second loss function. The equation for solving the second loss function is as follows: in, This represents the second loss function. This represents a matrix composed of each measurement value from the historical measurement data of the neighborhood. This represents a column vector containing the parameters of the regression model. This represents a column vector consisting of measurement data from each time segment in the historical measurement data of the neighborhood.
[0035] Solving the equation for the second loss function yields the parameters of the regression model. The regression model with the solved parameters is then used to impute short-period outlier data. The imputed data is represented as follows: in, Indicates the first Fill-in data for each time segment Indicates the first In the time section and the first The filling measurement value corresponding to each data point.
[0036] Based on the imputation data, an initial real-time dataset for power operation parameter correction is obtained. Then, a pre-trained random forest model is used to impute the initial real-time dataset for power operation parameter correction, resulting in a new real-time dataset for power operation parameter correction. Specifically, a bootstrap sampling algorithm is used to resample the historical dataset for power operation parameters, resulting in several subsets of historical measurement data. Each subset of historical measurement data is trained to obtain a corresponding decision tree. The time aspect of long-period abnormal data is input into each decision tree, and the average value of the output value of each decision tree is used as the imputation value for the long-period abnormal data, thus obtaining the real-time dataset for power operation parameter correction.
[0037] In a preferred embodiment of this application, least squares regression and random forest algorithms are combined to achieve adaptive missing data imputation. The appropriate imputation model can be dynamically selected according to the data missing characteristics at different time scales to ensure the accuracy and computational efficiency of data imputation.
[0038] S3. Input the real-time dataset of power operation parameter correction into the pre-constructed distribution network line attribute parameter identification model to obtain the line attribute parameter identification result of the target distribution network. The first loss function of the distribution network line attribute parameter identification model includes at least a power flow equation physical constraint loss term. The power flow equation physical constraint loss term is set to reflect the deviation between the actual value of the line voltage amplitude difference and the estimated value of the line voltage amplitude difference obtained based on the line attribute parameter identification result. In a preferred embodiment of this application, a neural network is used to construct the distribution network line attribute parameter identification model. Specifically, the neural network model is constructed, including an input layer, an output layer, and several hidden layers. The data propagation process in the network is as follows: in, Indicates the first The final output result obtained after forward propagation of a real-time dataset of power operation parameter corrections , and All represent weight parameters. , and These represent the bias parameters, Represents a non-linear activation function. Indicates the first Real-time dataset for power operation parameter correction It is the Hadamard product.
[0039] Neural network models typically construct loss functions based on the error between predicted and actual values, resulting in a singular optimization objective. This can lead to deviations in the physical laws governing the identification of distribution network line attribute parameters, rendering them ineffective for practical application. In this application, the distribution network parameter identification error is modeled, and a parameter identification error loss term is constructed. The parameter identification error loss term is expressed as follows: in, This represents the parameter identification error loss term. Indicates the first The real values of the distribution network line attribute parameters corresponding to the real-time dataset of power operation parameter correction are given. This indicates the number of real-time measurement correction data in the real-time dataset for power operation parameter correction.
[0040] Furthermore, the active power, reactive power, and attribute parameters of the distribution network lines to be identified are modeled in the real-time power operation parameter dataset to obtain the estimated value of the line voltage amplitude difference at the two ends of the line. The deviation between the actual value and the estimated value of the line voltage amplitude difference at the two ends of the line is then regularized to obtain the physical constraint loss term of the power flow equation. The physical constraint loss term of the power flow equation is expressed as: in, This represents the physical constraint loss term in the power flow equations. Indicates the line index of the target distribution network. Indicates the number of lines. Indicates the first The voltage amplitude difference between the two nodes at both ends of the line, Indicates the first The active power of the line, Indicates the first The reactive power of the line, Indicates the first The average voltage amplitude at both ends of the line. This indicates the first parameter in the attribute parameters of the distribution network line to be identified. The resistance value of the line, This indicates the first parameter in the attribute parameters of the distribution network line to be identified. The reactance value of the line.
[0041] Based on the parameter identification error loss term and the power flow equation physical constraint loss term, a first loss function is constructed. The expression of the first loss function is: in, and This represents the weights used to normalize the parameter identification error loss term and the power flow equation physical constraint loss term; their values are obtained by fitting the training dataset.
[0042] Based on the first loss function, the pre-constructed neural network model is trained to obtain the distribution network line attribute parameter identification model. The training dataset used for training has the same characteristics as the real-time dataset for power operation parameter correction.
[0043] In a preferred embodiment of this application, the first loss function of the distribution network line attribute parameter identification model simultaneously considers the parameter identification error and the physical constraints of the power flow equation, thereby improving the identification accuracy of the obtained distribution network line attribute parameter identification model.
[0044] In a preferred embodiment of the present invention, a real-time dataset of power operation parameters of the target distribution network is obtained, and outlier identification is performed on the real-time dataset of power operation parameters using a pre-constructed feasible domain matrix of power operation parameters to determine the outlier dataset. The feasible domain matrix of power operation parameters is set as the feature vector of the historical dataset of power operation parameters of the target distribution network and the maximum permissible deviation of power operation parameters, and the upper and lower limits of the historical dataset of power operation parameters are corrected. The least squares regression algorithm and a pre-trained random forest model are used to sequentially impute the outlier dataset to obtain a real-time dataset of corrected power operation parameters. The real-time dataset of corrected power operation parameters is input into a pre-constructed distribution network line attribute parameter identification model to obtain the line attribute parameter identification result of the target distribution network. The first loss function of the distribution network line attribute parameter identification model includes at least a power flow equation physical constraint loss term, which is set to reflect the deviation between the actual value of the line voltage amplitude difference and the estimated value of the line voltage amplitude difference obtained based on the line attribute parameter identification result. The distribution network line attribute parameter identification method disclosed in this application uses Gaussian kernel function and nonparametric kernel density theory to model the historical dataset of power operation parameters to obtain the feature vector of the historical dataset. This feature vector is used to correct the upper and lower limits of each time segment of the historical dataset. Then, based on the maximum permissible deviation of the power operation parameters, the initial feasible region matrix of the power operation parameters is corrected a second time to obtain the feasible region matrix of the power operation parameters. This allows for accurate identification of abnormal data in the real-time dataset of power operation parameters. It does not require the assumption that the data follows a characteristic distribution and is applicable to complex and nonnormal data distributions. It can effectively identify abnormal data in the real-time dataset of power operation parameters, providing accurate basic data for the identification of distribution network line attribute parameters. The first loss function of the distribution network line attribute parameter identification model considers both parameter identification error and physical constraints of power flow equations, improving the identification accuracy of the obtained distribution network line attribute parameter identification model.
[0045] Accordingly, such as Figure 2 The schematic diagram of the distribution network line attribute parameter identification system shown is based on a distribution network line attribute parameter identification method. This embodiment of the invention also provides a distribution network line attribute parameter identification system to implement the distribution network line attribute parameter identification method disclosed in this embodiment of the invention, including: an abnormal data identification module 1, an abnormal data correction module 2, and a line attribute parameter identification module 3. The abnormal data identification module 1 is used to acquire the real-time data set of power operation parameters of the target distribution network, and to identify outliers in the real-time data set of power operation parameters using a pre-constructed feasible domain matrix of power operation parameters to determine the abnormal data set. The feasible domain matrix of power operation parameters is set to be obtained by correcting the upper and lower limits of the historical data set of power operation parameters based on the feature vector of the historical data set of power operation parameters of the target distribution network and the maximum permissible deviation of power operation parameters. The abnormal data correction module 2 is used to fill in the abnormal data in the abnormal dataset to obtain a real-time dataset for power operation parameter correction. The line attribute parameter identification module 3 is used to input the real-time dataset of power operation parameter correction into the pre-constructed distribution network line attribute parameter identification model to obtain the line attribute parameter identification result of the target distribution network. The first loss function of the distribution network line attribute parameter identification model includes at least a power flow equation physical constraint loss term. The power flow equation physical constraint loss term is set to reflect the deviation between the actual value of the line voltage amplitude difference and the estimated value of the line voltage amplitude difference obtained based on the line attribute parameter identification result.
[0046] For specific limitations regarding the distribution network line attribute parameter identification system, please refer to the above-described limitations regarding the distribution network line attribute parameter identification method, which will not be repeated here. Those skilled in the art will recognize that the various modules and steps described in conjunction with the embodiments disclosed in this invention can be implemented in hardware, software, or a combination of both. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this invention.
[0047] like Figure 3 The diagram shows the internal structure of a computer device. An embodiment of the present invention provides a computer device including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the steps described in the embodiment of the power distribution network line attribute parameter identification method, for example... Figure 1 Steps S1 to S3 as described above.
[0048] Those skilled in the art will understand that the illustrations Figure 3 This is merely an example of a computer device and does not constitute a limitation on the computer device. It may include more or fewer components than shown, or combine certain components, or different components. For example, the computer device may also include input / output devices, network access devices, buses, etc.
[0049] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the computer device, connecting various parts of the computer device via various interfaces and lines.
[0050] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the computer device by running or executing the computer programs and / or modules stored in the memory and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.
[0051] If the modules integrated into the computer device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.
[0052] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.
[0053] Accordingly, embodiments of the present invention provide a computer-readable storage medium, the computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to perform the steps described in the embodiments of the power distribution network line attribute parameter identification method, for example... Figure 1 Steps S1 to S3 as described above.
[0054] In summary, the present application provides a method, system, device, and medium for identifying the attribute parameters of distribution network lines, addressing the technical problem of improving the accuracy and practicality of the identification results. The method includes: acquiring a real-time dataset of power operation parameters of the target distribution network; identifying outliers in the real-time dataset using a pre-constructed feasible domain matrix of power operation parameters; determining outlier datasets; and setting the feasible domain matrix of power operation parameters to a feature vector based on the historical dataset of power operation parameters of the target distribution network and the maximum permissible deviation of the power operation parameters, while also considering the upper and lower bounds of the historical dataset of power operation parameters. After correction, the following steps are taken: using the least squares regression algorithm and a pre-trained random forest model, abnormal data is imputed in the abnormal dataset to obtain a real-time dataset for power operation parameter correction; the real-time dataset for power operation parameter correction is input into a pre-built distribution network line attribute parameter identification model to obtain the line attribute parameter identification results of the target distribution network. The first loss function of the distribution network line attribute parameter identification model includes at least a power flow equation physical constraint loss term, which is set to reflect the deviation between the actual value of the line voltage amplitude difference and the estimated value of the line voltage amplitude difference obtained based on the line attribute parameter identification results. The distribution network line attribute parameter identification method disclosed in this application uses Gaussian kernel function and nonparametric kernel density theory to model the historical dataset of power operation parameters to obtain the feature vector of the historical dataset. This feature vector is used to correct the upper and lower limits of each time segment of the historical dataset. Then, based on the maximum permissible deviation of the power operation parameters, the initial feasible region matrix of the power operation parameters is corrected a second time to obtain the feasible region matrix of the power operation parameters. This allows for accurate identification of abnormal data in the real-time dataset of power operation parameters. It does not require the assumption that the data follows a characteristic distribution and is applicable to complex and non-normal data distributions. It can effectively identify abnormal data in the real-time dataset of power operation parameters, providing accurate basic data for the identification of distribution network line attribute parameters. The first loss function of the distribution network line attribute parameter identification model considers both parameter identification error and physical constraints of power flow equations, improving the identification accuracy of the distribution network line attribute parameter identification model, thereby improving the accuracy and practicality of the identification results.
[0055] The various embodiments in this specification are described in a progressive manner. For directly identical or similar parts of the embodiments, refer to each other. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. It should be noted that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.
[0056] The embodiments described above are merely preferred embodiments of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various improvements and substitutions without departing from the technical principles of this application, and these improvements and substitutions should also be considered within the scope of protection of this application. Therefore, the scope of protection of this patent application should be determined by the scope of the claims.
Claims
1. A method for identifying attribute parameters of power distribution network lines, characterized in that, The method includes: A real-time dataset of power operation parameters of the target distribution network is obtained, and an outlier is identified in the real-time dataset of power operation parameters using a pre-constructed feasible domain matrix of power operation parameters. The outlier dataset is determined. The feasible domain matrix of power operation parameters is set to be obtained by correcting the upper and lower limits of the historical dataset of power operation parameters based on the feature vector of the historical dataset of power operation parameters of the target distribution network and the maximum permissible deviation of power operation parameters. The abnormal dataset is filled with abnormal data to obtain a real-time dataset for power operation parameter correction; The real-time dataset of power operation parameter correction is input into the pre-constructed distribution network line attribute parameter identification model to obtain the line attribute parameter identification result of the target distribution network. The first loss function of the distribution network line attribute parameter identification model includes at least a power flow equation physical constraint loss term. The power flow equation physical constraint loss term is set to reflect the deviation between the actual value of the line voltage amplitude difference and the estimated value of the line voltage amplitude difference obtained based on the line attribute parameter identification result.
2. The method for identifying the attribute parameters of a power distribution network line as described in claim 1, characterized in that, The process of acquiring a real-time dataset of power operation parameters of the target distribution network and identifying outliers in the real-time dataset using a pre-constructed feasible domain matrix of power operation parameters to determine outlier datasets includes: The historical data set of power operation parameters for several historical scheduling days of the target distribution network is obtained, and the historical data set of power operation parameters is processed by Gaussian kernel function and nonparametric kernel density theory to obtain the data probability density function of each historical data of power operation parameter in the historical data set of power operation parameters at each time segment. Based on the data probability density function, the maximum probability density vector is obtained. According to the maximum probability density vector, a data similarity matching algorithm is used to obtain the weight of the historical dataset of power operation parameters corresponding to each historical scheduling day relative to the maximum probability density vector. The historical power operation parameter dataset is represented by a vector according to the weights to obtain the feature vector of the historical power operation parameter dataset. The upper and lower limits of each time segment of the historical power operation parameter dataset are then corrected according to the feature vector to obtain the initial feasible domain matrix of power operation parameters. Based on the maximum permissible deviation of the power operation parameters, the initial power operation parameter feasible region matrix is corrected twice to obtain the power operation parameter feasible region matrix. The feasible domain matrix of the power operation parameters is used to identify outliers in the real-time dataset of power operation parameters and determine the outlier dataset.
3. The method for identifying the attribute parameters of a power distribution network line as described in claim 2, characterized in that, The step of obtaining the weight of the historical dataset of power operation parameters corresponding to each historical scheduling day relative to the maximum probability density vector using a data similarity matching algorithm, based on the maximum probability density vector, includes: Calculate the Euclidean distance between the maximum probability density vector and the historical data of power operation parameters corresponding to each time segment of each historical scheduling day to obtain the intermediate variable corresponding to each historical scheduling day. The intermediate variables are normalized to obtain the weights of the historical dataset of power operation parameters corresponding to each historical scheduling day relative to the maximum probability density vector.
4. The method for identifying the attribute parameters of a power distribution network line as described in claim 2, characterized in that, The step of correcting the upper and lower limits of each time segment of the historical dataset of power operation parameters based on the feature vector to obtain the initial feasible region matrix of power operation parameters includes: The maximum and minimum values of the historical dataset of power operation parameters at each time segment are identified to obtain the maximum and minimum values of the historical dataset of power operation parameters at each time segment. Based on the eigenvectors, a linear normalization algorithm is used to correct each maximum and minimum value once to obtain the initial feasible domain matrix of power operation parameters.
5. The method for identifying the attribute parameters of a power distribution network line as described in claim 1, characterized in that, The process of imputing abnormal data in the abnormal dataset to obtain a real-time dataset for power operation parameter correction includes: Based on a preset data length threshold, the abnormal dataset is divided into short-cycle abnormal data and long-cycle abnormal data. Select several adjacent time points of the short-period abnormal data and their corresponding neighborhood historical measurement data, and construct a regression model based on the neighborhood historical measurement data; The second loss function of the regression model is constructed using the least squares regression algorithm, and the parameters of the second loss function are solved. The regression model after parameter solving is then used to fill in the short-cycle abnormal data to obtain the initial power operation parameter correction real-time dataset. The initial real-time dataset for power operation parameter correction was filled in using a pre-trained random forest model to obtain the real-time dataset for power operation parameter correction.
6. The method for identifying the attribute parameters of a power distribution network line as described in claim 5, characterized in that, The initial real-time power operation parameter correction dataset is filled in using a pre-trained random forest model, resulting in a real-time power operation parameter correction dataset, including: The historical dataset of power operation parameters is resampled based on the self-sampling algorithm to obtain several subsets of historical measurement data; Training is performed on each subset of historical measurement data to obtain several decision trees. A random forest model is then constructed based on the decision trees. The time aspect of the long-period anomaly data is input into each of the random forest models, and the real-time dataset for power operation parameter correction is obtained based on the imputation values of the long-period anomaly data output by the random forest models.
7. The method for identifying the attribute parameters of a power distribution network line as described in claim 1, characterized in that, The step of inputting the real-time dataset of power operation parameter correction into a pre-built distribution network line attribute parameter identification model to obtain the line attribute parameter identification result of the target distribution network includes: Model the parameter identification error of the distribution network and construct the parameter identification error loss term; Model the active power, reactive power, voltage amplitude, and distribution network line attribute parameters in the real-time power operation parameter dataset to obtain the estimated value of the voltage amplitude difference between the two ends of the line. The deviation between the actual value of the line voltage amplitude difference and the estimated value of the line voltage amplitude difference corresponding to the nodes at both ends of the line is modeled in a regularized manner to obtain the physical constraint loss term of the power flow equation. Based on the parameter identification error loss term and the power flow equation physical constraint loss term, a first loss function is constructed. Based on the first loss function, the pre-constructed neural network model is trained to obtain the distribution network line attribute parameter identification model. The training dataset used for training has the same characteristics as the real-time dataset for power operation parameter correction. The real-time dataset of power operation parameter correction is input into the distribution network line attribute parameter identification model to obtain the line attribute parameter identification result of the target distribution network.
8. A distribution network line attribute parameter identification system, used to implement the distribution network line attribute parameter identification method according to any one of claims 1-7, characterized in that, The system includes: an abnormal data identification module, an abnormal data correction module, and a line attribute parameter identification module; The abnormal data identification module is used to acquire the real-time data set of power operation parameters of the target distribution network, and to identify outliers in the real-time data set of power operation parameters using a pre-constructed feasible domain matrix of power operation parameters to determine the abnormal data set. The feasible domain matrix of power operation parameters is set to be obtained by correcting the upper and lower limits of the historical data set of power operation parameters based on the feature vector of the historical data set of power operation parameters of the target distribution network and the maximum permissible deviation of power operation parameters. The abnormal data correction module is used to fill in the abnormal data in the abnormal dataset to obtain a real-time dataset for power operation parameter correction. The line attribute parameter identification module is used to input the real-time dataset of power operation parameter correction into a pre-built distribution network line attribute parameter identification model to obtain the line attribute parameter identification result of the target distribution network. The first loss function of the distribution network line attribute parameter identification model includes at least a power flow equation physical constraint loss term. The power flow equation physical constraint loss term is set to reflect the deviation between the actual value of the line voltage amplitude difference and the estimated value of the line voltage amplitude difference obtained based on the line attribute parameter identification result.
9. A computer device, characterized in that: The computer device includes a memory, a processor, and a transceiver, which are connected to each other via a bus; the memory is used to store a set of computer program instructions and data, and to transmit the stored data to the processor, and the processor executes the computer program instructions stored in the memory to perform the distribution network line attribute parameter identification method as described in any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program that, when executed, implements the distribution network line attribute parameter identification method as described in any one of claims 1 to 7.