Power distribution network fault intelligent detection method and device based on current dip analysis

Abstract

The application discloses a power distribution network fault intelligent detection method and device based on current sudden drop analysis, and the method comprises the following steps: acquiring the electric parameter data set of the power distribution network to be processed, and performing data processing on the electric parameter data set to obtain the processed electric parameter data set; training the fault detection model to be trained according to the processed electric parameter data set to obtain the trained fault detection model, and judging whether the trained fault detection model converges or not, if yes, the trained fault detection model is determined as the target fault detection model. It can be seen that the target fault detection model for analyzing the current sudden drop condition can be trained by implementing the application, so as to detect whether the power distribution network has faults through current sudden drop analysis, and even if the telemetry information cannot be acquired, the faults existing in the power distribution network can also be detected in time, and then the power distribution network can be maintained in time, so that the reliability, accuracy and effectiveness of the power distribution network fault detection work and maintenance work are improved.

Description

Intelligent Fault Detection Method and Device for Distribution Network Based on Current Drop Analysis Technical Field

[0001] This invention relates to the field of power equipment testing technology, and in particular to an intelligent fault detection method and device for distribution networks based on current descent analysis. Background Technology

[0002] With the increasing scale of power systems, the safe and stable operation of distribution networks is becoming increasingly important for ensuring the power supply capacity of the power system. When a fault occurs in the distribution network, it is necessary to detect the fault in a timely manner so that timely repairs can be carried out, thereby ensuring the safe and stable operation of the distribution network.

[0003] Currently, fault detection in distribution networks typically involves acquiring and analyzing real-time telemetry information to detect faults. However, practical experience has shown that existing distribution networks lack telemetry information on low-voltage side switches, preventing them from proactively sensing switch status and hindering timely detection of tripped switches, thus impacting user experience. Therefore, providing a novel method for analyzing distribution network switch status to promptly detect faults is crucial. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method and device for intelligent detection of distribution network faults based on current drop analysis. Even if telemetry information cannot be obtained, faults in the distribution network can be detected in a timely manner, and the distribution network can be repaired in a timely manner, thereby improving the reliability, accuracy and effectiveness of distribution network fault detection and repair work.

[0005] To address the aforementioned technical problems, the first aspect of this invention discloses an intelligent fault detection method for distribution networks based on current descent analysis, the method comprising:

[0006] A dataset of electrical parameters of the distribution network to be processed is obtained, and data processing operations are performed on the dataset to obtain a processed dataset of electrical parameters; the data processing operations include data augmentation, data feature extraction, and data equalization.

[0007] Based on the processed electrical parameter dataset, the preset fault detection model to be trained is trained to obtain the trained fault detection model, and it is determined whether the trained fault detection model has converged.

[0008] When it is determined that the trained fault detection model has converged, the trained fault detection model is determined as the target fault detection model. The target fault detection model is used to analyze the current drop of the distribution network to be determined based on the electrical parameter data of the distribution network to be determined, so as to detect whether there is a fault in the switching state of the distribution network to be determined based on the current drop.

[0009] As an optional implementation, in the first aspect of the present invention, each subset of electrical parameter data in the electrical parameter dataset is a subset containing electrical parameter data of multiple phase types;

[0010] The step of performing data processing operations on the electrical parameter dataset of the distribution network to obtain a processed electrical parameter dataset includes:

[0011] Based on all the phase-type electrical parameter data contained in each electrical parameter data subset, perform a data augmentation operation on each electrical parameter data subset to obtain an augmented electrical parameter data set corresponding to each electrical parameter data subset;

[0012] Perform a data feature extraction operation on the enhanced electrical parameter data set corresponding to each subset of electrical parameter data to obtain the electrical parameter feature data set corresponding to each subset of electrical parameter data.

[0013] Perform a data equalization operation on the electrical parameter feature data set corresponding to each subset of electrical parameter data to obtain the equalized feature data set corresponding to each subset of electrical parameter data, and determine the equalized feature data set corresponding to all subsets of electrical parameter data as the processed electrical parameter dataset.

[0014] As an optional implementation, in the first aspect of the present invention, the step of performing a data augmentation operation on each of the electrical parameter data subsets based on all the phase-type electrical parameter data contained in each of the electrical parameter data subsets to obtain an augmented electrical parameter data set corresponding to each of the electrical parameter data subsets includes:

[0015] For each subset of electrical parameter data, the decimation weight corresponding to each phase type of electrical parameter data is determined based on all the phase type of electrical parameter data contained in the subset of electrical parameter data;

[0016] Based on the shunting weight corresponding to the electrical parameter data of each phase type, determine the shunting ratio corresponding to the electrical parameter data of each phase type;

[0017] Based on the shunting ratio corresponding to the electrical parameter data of each phase type, multiple data shunting operations are performed in parallel on all electrical parameter data of the phase types contained in the electrical parameter data subset to obtain all shunted electrical parameter data corresponding to the electrical parameter data subset, which serves as the enhanced electrical parameter data set corresponding to the electrical parameter data subset; each data shunting operation corresponds to one shunted electrical parameter data, and each shunted electrical parameter data contains all the shunted electrical parameter data of the phase types.

[0018] As an optional implementation, in the first aspect of the present invention, the enhanced electrical parameter data set corresponding to each subset of electrical parameter data includes an enhanced current parameter data set and an enhanced load parameter data set.

[0019] The step of performing data feature extraction on the enhanced electrical parameter data set corresponding to each subset of electrical parameter data to obtain the electrical parameter feature data set corresponding to each subset of electrical parameter data includes:

[0020] For each subset of electrical parameter data, a data feature extraction operation is performed on the enhanced current parameter data set and the enhanced load parameter data set corresponding to the subset of electrical parameter data to obtain a first feature data set corresponding to the enhanced current parameter data set and a second feature data set corresponding to the enhanced load parameter data set, which are used as the electrical parameter feature data set corresponding to the subset of electrical parameter data. The first feature data set includes at least one of the following: current drop amplitude parameter, three-phase current imbalance parameter, current drop depth, current amplitude ratio, and peak-valley difference parameter. The second feature data set includes a load rate parameter and / or a load rate difference parameter.

[0021] As an optional implementation, in the first aspect of the present invention, performing a data equalization operation on the electrical parameter feature data set corresponding to each subset of electrical parameter data to obtain an equalized feature data set corresponding to each subset of electrical parameter data includes:

[0022] For each subset of electrical parameter data, based on the pre-determined tag type of each electrical parameter feature data corresponding to the subset of electrical parameter data, all target electrical parameter feature data are determined from all electrical parameter feature data, and all neighboring target electrical parameter feature data corresponding to each target electrical parameter feature data are determined; the tag type of each electrical parameter feature data is associated with the current drop situation of the distribution network.

[0023] For each target electrical parameter feature data in each subset of electrical parameter data, determine the neighboring data sampling ratio corresponding to the target electrical parameter feature data, and sample all the neighboring target electrical parameter feature data corresponding to the target electrical parameter feature data according to the neighboring data sampling ratio, so as to update all the neighboring target electrical parameter feature data corresponding to the target electrical parameter feature data.

[0024] For each target electrical parameter feature data in each subset of electrical parameter data, construct electrical parameter feature data between the target electrical parameter feature data and each updated neighboring target electrical parameter feature data is calculated based on all the updated neighboring target electrical parameter feature data corresponding to the target electrical parameter feature data.

[0025] For each subset of electrical parameter data, the subset of electrical parameter data is updated according to all the constructed electrical parameter feature data corresponding to the obtained subset of electrical parameter data;

[0026] The corresponding electrical parameter feature data are as follows:

[0027] x new = x + rand(0,1) * (x` - x),

[0028] x is the target electrical parameter feature data, and x` is any of the updated target electrical parameter feature data corresponding to the target electrical parameter feature data.

[0029] As an optional implementation, in the first aspect of the present invention, the step of determining all target electrical parameter feature data from all the electrical parameter feature data according to the tag type to which each electrical parameter feature data corresponding to the predetermined subset of electrical parameter data belongs, and determining all neighboring target electrical parameter feature data corresponding to each target electrical parameter feature data, includes:

[0030] Based on the label type to which each electrical parameter feature data belongs to the predetermined subset of electrical parameter data, determine all target electrical parameter feature data under the target label type whose data volume is less than or equal to a preset data volume threshold from all electrical parameter feature data;

[0031] For each target electrical parameter feature data, calculate the Euclidean distance parameter between the target electrical parameter feature data and all other target electrical parameter feature data, and based on the Euclidean distance parameter between the target electrical parameter feature data and all other target electrical parameter feature data, determine all other target electrical parameter feature data whose Euclidean distance parameter is less than or equal to a preset parameter threshold from all other target electrical parameter feature data, and use them as all neighboring target electrical parameter feature data corresponding to the target electrical parameter feature data.

[0032] As an optional implementation, in the first aspect of the invention, the three-phase current imbalance parameter is calculated using the following feature extraction formula:

[0033]

[0034] ε i Let I be the three-phase current imbalance parameter. ai I bi I ci All of these are current parameter data of the corresponding phase type extracted from the enhanced current parameter data, I mi It is the average value of all current parameter data of the phase type extracted from the enhanced current parameter data.

[0035] A second aspect of this invention discloses an intelligent fault detection device for distribution networks based on current descent analysis, the device comprising:

[0036] The acquisition module is used to acquire the electrical parameter dataset of the distribution network to be processed;

[0037] The processing module is used to perform data processing operations on the electrical parameter dataset of the distribution network to obtain a processed electrical parameter dataset; the data processing operations include data augmentation operations, data feature extraction operations, and data equalization operations;

[0038] The training module is used to train the preset fault detection model to be trained based on the processed electrical parameter dataset, so as to obtain the trained fault detection model.

[0039] The judgment module is used to determine whether the trained fault detection model has converged.

[0040] The determination module is used to determine the trained fault detection model as the target fault detection model when the judgment module determines that the trained fault detection model has converged; the target fault detection model is used to analyze the current drop situation of the distribution network to be judged based on the electrical parameter data of the distribution network to be judged, so as to detect whether there is a fault in the switching state of the distribution network to be judged based on the current drop situation.

[0041] As an optional implementation, in the second aspect of the present invention, each subset of electrical parameter data in the electrical parameter dataset is a subset containing electrical parameter data of multiple phase types;

[0042] Specifically, the processing module performs data processing operations on the electrical parameter dataset of the distribution network to obtain the processed electrical parameter dataset in the following manner:

[0043] Based on all the phase-type electrical parameter data contained in each electrical parameter data subset, perform a data augmentation operation on each electrical parameter data subset to obtain an augmented electrical parameter data set corresponding to each electrical parameter data subset;

[0044] Perform a data feature extraction operation on the enhanced electrical parameter data set corresponding to each subset of electrical parameter data to obtain the electrical parameter feature data set corresponding to each subset of electrical parameter data.

[0045] Perform a data equalization operation on the electrical parameter feature data set corresponding to each subset of electrical parameter data to obtain the equalized feature data set corresponding to each subset of electrical parameter data, and determine the equalized feature data set corresponding to all subsets of electrical parameter data as the processed electrical parameter dataset.

[0046] As an optional implementation, in the second aspect of the present invention, the processing module performs a data augmentation operation on each electrical parameter data subset based on all the phase-type electrical parameter data contained in each electrical parameter data subset, and obtains the augmented electrical parameter data set corresponding to each electrical parameter data subset in the following specific manner:

[0047] For each subset of electrical parameter data, the decimation weight corresponding to each phase type of electrical parameter data is determined based on all the phase type of electrical parameter data contained in the subset of electrical parameter data;

[0048] Based on the shunting weight corresponding to the electrical parameter data of each phase type, determine the shunting ratio corresponding to the electrical parameter data of each phase type;

[0049] Based on the shunting ratio corresponding to the electrical parameter data of each phase type, multiple data shunting operations are performed in parallel on all electrical parameter data of the phase types contained in the electrical parameter data subset to obtain all shunted electrical parameter data corresponding to the electrical parameter data subset, which serves as the enhanced electrical parameter data set corresponding to the electrical parameter data subset; each data shunting operation corresponds to one shunted electrical parameter data, and each shunted electrical parameter data contains all the shunted electrical parameter data of the phase types.

[0050] As an optional implementation, in the second aspect of the present invention, the enhanced electrical parameter data set corresponding to each subset of electrical parameter data includes an enhanced current parameter data set and an enhanced load parameter data set.

[0051] Specifically, the processing module performs data feature extraction on the enhanced electrical parameter data set corresponding to each subset of electrical parameter data to obtain the electrical parameter feature data set corresponding to each subset of electrical parameter data in the following manner:

[0052] For each subset of electrical parameter data, a data feature extraction operation is performed on the enhanced current parameter data set and the enhanced load parameter data set corresponding to the subset of electrical parameter data to obtain a first feature data set corresponding to the enhanced current parameter data set and a second feature data set corresponding to the enhanced load parameter data set, which are used as the electrical parameter feature data set corresponding to the subset of electrical parameter data. The first feature data set includes at least one of the following: current drop amplitude parameter, three-phase current imbalance parameter, current drop depth, current amplitude ratio, and peak-valley difference parameter. The second feature data set includes a load rate parameter and / or a load rate difference parameter.

[0053] As an optional implementation, in the second aspect of the present invention, the processing module performs a data equalization operation on the electrical parameter feature data set corresponding to each subset of electrical parameter data to obtain the equalized feature data set corresponding to each subset of electrical parameter data in the following specific manner:

[0054] For each subset of electrical parameter data, based on the pre-determined tag type of each electrical parameter feature data corresponding to the subset of electrical parameter data, all target electrical parameter feature data are determined from all electrical parameter feature data, and all neighboring target electrical parameter feature data corresponding to each target electrical parameter feature data are determined; the tag type of each electrical parameter feature data is associated with the current drop situation of the distribution network.

[0055] For each target electrical parameter feature data in each subset of electrical parameter data, determine the neighboring data sampling ratio corresponding to the target electrical parameter feature data, and sample all the neighboring target electrical parameter feature data corresponding to the target electrical parameter feature data according to the neighboring data sampling ratio, so as to update all the neighboring target electrical parameter feature data corresponding to the target electrical parameter feature data.

[0056] For each target electrical parameter feature data in each subset of electrical parameter data, construct electrical parameter feature data between the target electrical parameter feature data and each updated neighboring target electrical parameter feature data is calculated based on all the updated neighboring target electrical parameter feature data corresponding to the target electrical parameter feature data.

[0057] For each subset of electrical parameter data, the subset of electrical parameter data is updated according to all the constructed electrical parameter feature data corresponding to the obtained subset of electrical parameter data;

[0058] The corresponding electrical parameter feature data are as follows:

[0059] x new = x + rand(0,1) * (x` - x),

[0060] x is the target electrical parameter feature data, and x` is any of the updated target electrical parameter feature data corresponding to the target electrical parameter feature data.

[0061] As an optional implementation, in the second aspect of the present invention, the processing module determines all target electrical parameter feature data from all the electrical parameter feature data according to the tag type to which each electrical parameter feature data belongs to the predetermined subset of electrical parameter data, and determines all neighboring target electrical parameter feature data corresponding to each target electrical parameter feature data in the following specific manner:

[0062] Based on the label type to which each electrical parameter feature data belongs to the predetermined subset of electrical parameter data, determine all target electrical parameter feature data under the target label type whose data volume is less than or equal to a preset data volume threshold from all electrical parameter feature data;

[0063] For each target electrical parameter feature data, calculate the Euclidean distance parameter between the target electrical parameter feature data and all other target electrical parameter feature data, and based on the Euclidean distance parameter between the target electrical parameter feature data and all other target electrical parameter feature data, determine all other target electrical parameter feature data whose Euclidean distance parameter is less than or equal to a preset parameter threshold from all other target electrical parameter feature data, and use them as all neighboring target electrical parameter feature data corresponding to the target electrical parameter feature data.

[0064] As an optional implementation, in a second aspect of the invention, the three-phase current imbalance parameter is calculated using the following feature extraction formula:

[0065]

[0066] ε i Let I be the three-phase current imbalance parameter. ai I bi I ci All of these are current parameter data of the corresponding phase type extracted from the enhanced current parameter data, I mi It is the average value of all current parameter data of the phase type extracted from the enhanced current parameter data.

[0067] A third aspect of this invention discloses another intelligent fault detection device for distribution networks based on current descent analysis, the device comprising:

[0068] Memory containing executable program code;

[0069] A processor coupled to the memory;

[0070] The processor calls the executable program code stored in the memory to execute the intelligent fault detection method for distribution networks based on current descent analysis disclosed in the first aspect of the present invention.

[0071] The fourth aspect of the present invention discloses a computer storage medium storing computer instructions, which, when invoked, are used to execute the intelligent fault detection method for distribution networks based on current descent analysis disclosed in the first aspect of the present invention.

[0072] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0073] In this embodiment of the invention, a dataset of electrical parameters of the distribution network to be processed is obtained, and the dataset is processed to obtain a processed dataset. Based on the processed dataset, a fault detection model is trained to obtain a trained fault detection model. The convergence of the trained model is then determined; if convergence is achieved, the trained model is identified as the target fault detection model. Therefore, implementing this invention can train a target fault detection model for analyzing sudden current drops, enabling the detection of faults in the distribution network through current drop analysis. Even without telemetry information, faults in the distribution network can be detected promptly, facilitating timely maintenance. This improves the reliability, accuracy, and effectiveness of fault detection and maintenance, ensuring the safe and stable operation of the power system and protecting users' electricity experience. Attached Figure Description

[0074] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0075] Figure 1 is a flowchart illustrating an intelligent fault detection method for distribution networks based on current descent analysis disclosed in an embodiment of the present invention.

[0076] Figure 2 is a flowchart illustrating another intelligent fault detection method for distribution networks based on current descent analysis disclosed in an embodiment of the present invention.

[0077] Figure 3 is a schematic diagram of the structure of a power distribution network fault intelligent detection device based on current descent analysis disclosed in an embodiment of the present invention;

[0078] Figure 4 is a schematic diagram of another intelligent fault detection device for distribution networks based on current descent analysis disclosed in an embodiment of the present invention. Detailed Implementation

[0079] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0080] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, apparatus, product, or end that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or ends.

[0081] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0082] This invention discloses an intelligent fault detection method and device for distribution networks based on current descent analysis. Even without telemetry information, it can promptly detect faults in the distribution network, enabling timely maintenance and thus improving the reliability, accuracy, and effectiveness of fault detection and maintenance. Detailed descriptions follow.

[0083] Example 1

[0084] Please refer to Figure 1, which is a flowchart illustrating an intelligent distribution network fault detection method based on current descent analysis disclosed in an embodiment of the present invention. The intelligent distribution network fault detection method based on current descent analysis described in Figure 1 can be applied to fault detection in distribution networks, as well as to fault detection in other power equipment, such as transmission lines, transformers, and contactors. This embodiment of the present invention does not limit its application. Optionally, this method can be implemented by a power grid fault detection system, which can be integrated into a power grid fault detection device, or it can be a local server or cloud server used to process the power grid fault detection process. This embodiment of the present invention does not limit its application. As shown in Figure 1, the intelligent distribution network fault detection method based on current descent analysis may include the following operations:

[0085] 101. Obtain the electrical parameter dataset of the distribution network to be processed, and perform data processing operations on the electrical parameter dataset of the distribution network to obtain the processed electrical parameter dataset.

[0086] In this embodiment of the invention, optionally, the data processing operations include data augmentation, data feature extraction, and data equalization. Data augmentation involves increasing the amount of data in the electrical parameter dataset used for training; data feature extraction involves calculating the feature parameters of the electrical parameter dataset; and data equalization involves equalizing a small number of highly variable electrical parameter data to reduce their difference from other electrical parameter data. Further optionally, the electrical parameter dataset may include a current parameter dataset, a load parameter dataset, and a load-specific parameter dataset. Specifically, each subset of electrical parameter data in the dataset contains electrical parameter data of multiple phase types, such as each subset containing electrical parameter data of phase A, phase B, and phase C.

[0087] 102. Based on the processed electrical parameter dataset, train the preset fault detection model to be trained to obtain the trained fault detection model, and determine whether the trained fault detection model has converged.

[0088] In this embodiment of the invention, each processed electrical parameter data in the processed electrical parameter dataset has a corresponding preset label, which indicates the type of current drop in the corresponding processed electrical parameter data. Specifically, a preset fault detection model to be trained is trained, that is, the predicted current drop and the type of predicted current drop in the corresponding processed electrical parameter data can be obtained through the fault detection model to be trained. Then, based on the predicted current drop, the type of predicted current drop, the predetermined current drop, and the label, the loss function corresponding to the trained fault detection model is calculated. The convergence of the trained fault detection model is determined based on the magnitude of the loss function. The fault detection module to be trained can use one or more decision tree models.

[0089] 103. When it is determined that the fault detection model after training has converged, the fault detection model after training is determined as the target fault detection model.

[0090] In this embodiment of the invention, the target fault detection model is used to analyze the current drop of the distribution network to be judged based on the electrical parameter data of the distribution network to be judged, so as to detect whether there is a fault in the switching state of the distribution network to be judged based on the current drop. For example, when it is found that the current of the three-phase AC power has dropped significantly, it can be determined that there is a fault in the switching state of the distribution network to be judged.

[0091] It is evident that implementing the embodiments of the present invention can train a target fault detection model for analyzing current drop situations, thereby detecting whether there are faults in the distribution network through current drop analysis. In this way, even if telemetry information cannot be obtained, faults in the distribution network can be detected in a timely manner. This not only improves the efficiency of distribution network fault detection, thereby improving the timeliness of distribution network detection and maintenance, but also enhances the reliability, accuracy, and effectiveness of distribution network fault detection and maintenance work, so as to ensure the safe and stable operation of the power system and that users' electricity experience is not affected.

[0092] Example 2

[0093] Please refer to Figure 2, which is a flowchart illustrating an intelligent distribution network fault detection method based on current descent analysis disclosed in an embodiment of the present invention. The intelligent distribution network fault detection method based on current descent analysis described in Figure 2 can be applied to fault detection in distribution networks, as well as to fault detection in other power equipment, such as transmission lines, transformers, and contactors. This embodiment of the present invention does not limit its application. Optionally, this method can be implemented by a power grid fault detection system, which can be integrated into a power grid fault detection device, or it can be a local server or cloud server used to process the power grid fault detection process. This embodiment of the present invention does not limit its application. As shown in Figure 2, the intelligent distribution network fault detection method based on current descent analysis may include the following operations:

[0094] 201. Obtain the electrical parameter dataset of the distribution network to be processed.

[0095] 202. Based on the electrical parameter data of all phase types contained in each electrical parameter data subset, perform data augmentation operation on each electrical parameter data subset to obtain the augmented electrical parameter data set corresponding to each electrical parameter data subset.

[0096] In this embodiment of the invention, the electrical parameter data subsets M and N are augmented with electrical parameter data of phase types A / B / C contained in the electrical parameter data subset M and electrical parameter data of phase types A / B / C contained in the electrical parameter data subset N to obtain more training samples.

[0097] 203. Perform data feature extraction operation on the enhanced electrical parameter data set corresponding to each subset of electrical parameter data to obtain the electrical parameter feature data set corresponding to each subset of electrical parameter data.

[0098] In this embodiment of the invention, the enhanced electrical parameter data set corresponding to each electrical parameter data subset is used to calculate the electrical parameter feature data set corresponding to each electrical parameter data subset. For example, the current parameter feature data set (such as current drop amplitude, three-phase current compensation balance, etc.), load feature data set (such as load rate, load rate difference, etc.), and load angle feature data set corresponding to each electrical parameter data subset are calculated.

[0099] 204. Perform data equalization on the electrical parameter feature data set corresponding to each electrical parameter data subset to obtain the equalized feature data set corresponding to each electrical parameter data subset, and determine the equalized feature data set corresponding to all electrical parameter data subsets as the processed electrical parameter dataset.

[0100] In this embodiment of the invention, the target electrical parameter feature data set of the minority class is subjected to equalization processing to reduce the extreme class imbalance between normal current data samples and abnormal samples at sudden drops, effectively balancing the proportion of normal and abnormal samples. Specifically, Euclidean distance calculation can be used to construct new samples for the target electrical parameter feature data set of the minority class to update the corresponding electrical parameter data subset, thereby achieving overall equalization of the electrical parameter dataset.

[0101] 205. Based on the processed electrical parameter dataset, train the preset fault detection model to be trained to obtain the trained fault detection model, and determine whether the trained fault detection model has converged.

[0102] 206. When it is determined that the fault detection model after training has converged, the fault detection model after training is determined as the target fault detection model.

[0103] In this embodiment of the invention, for other descriptions of steps 201, 205 and 206, please refer to the detailed description of steps 101-103 in Embodiment 1. This embodiment of the invention will not repeat them.

[0104] As can be seen, implementing the embodiments of the present invention can realize the processing operation of the electrical parameter dataset by enhancing, extracting features and equalizing the corresponding subset of electrical parameter data, thus realizing an intelligent processing method for the electrical parameter dataset. This is beneficial to improving the reliability and accuracy of the processing operation of the electrical parameter dataset, and in turn, it is beneficial to improve the reliability and accuracy of the subsequent training operation of the fault detection model to be trained using the processed electrical parameter dataset, thereby facilitating the accurate training of the target fault detection model for analyzing current drop situations.

[0105] In an optional embodiment, step 202 above, which involves performing a data augmentation operation on each electrical parameter data subset based on all phase-type electrical parameter data contained in each subset, to obtain an augmented electrical parameter data set corresponding to each electrical parameter data subset, includes:

[0106] For each subset of electrical parameter data, the decimation weight corresponding to each phase type of electrical parameter data is determined based on all phase type electrical parameter data contained in the subset of electrical parameter data.

[0107] Based on the shunting weight corresponding to the electrical parameter data of each phase type, determine the shunting ratio corresponding to the electrical parameter data of each phase type;

[0108] Based on the decimation ratio corresponding to each phase type of electrical parameter data, multiple data decimation operations are performed in parallel on all phase types of electrical parameter data contained in the electrical parameter data subset to obtain all decimated electrical parameter data corresponding to the electrical parameter data subset, which serves as the enhanced electrical parameter data set corresponding to the electrical parameter data subset.

[0109] In this optional embodiment, specifically, each data merging operation corresponds to a merged electrical parameter data set, and each merged electrical parameter data set contains electrical parameter data of all phase types that were merged. For example, if multiple data merging operations are performed on the A / B / C phase type electrical parameter data contained in the electrical parameter data subset M, and in each data merging process, since the A / B / C phase type electrical parameter data are of equal importance, one electrical parameter data set can be randomly extracted from each of the A / B / C phase type electrical parameter data sets and merged into a new electrical parameter data set, thereby realizing the data enhancement (i.e., amplification) operation on the corresponding electrical parameter data subset. Optionally, the enhanced electrical parameter data set corresponding to each electrical parameter data subset includes an enhanced current parameter data set and an enhanced load parameter data set, and may also include an enhanced load parameter data set.

[0110] As can be seen, this optional embodiment can perform merging operations on the corresponding electrical parameter data subsets according to the determined merging ratio, so as to achieve the enhancement effect of the electrical parameter dataset. In this way, the reliability and accuracy of the merging operation on the electrical parameter data subsets can be improved, thereby improving the effectiveness of the enhancement of the electrical parameter dataset, and thus significantly increasing the amount of reliable, accurate and effective electrical parameter dataset.

[0111] In another optional embodiment, step 203 above involves performing a data feature extraction operation on the enhanced electrical parameter data set corresponding to each subset of electrical parameter data to obtain an electrical parameter feature data set corresponding to each subset of electrical parameter data, including:

[0112] For each subset of electrical parameter data, data feature extraction operations are performed on the enhanced current parameter data set and the enhanced load parameter data set corresponding to the subset of electrical parameter data to obtain the first feature data set corresponding to the enhanced current parameter data set and the second feature data set corresponding to the enhanced load parameter data set, which are used as the electrical parameter feature data set corresponding to the subset of electrical parameter data.

[0113] In this optional embodiment, the first feature data set may optionally include at least one of the following: current descent amplitude parameter, three-phase current imbalance parameter, current drop depth, current amplitude ratio, and peak-to-valley difference parameter; and the second feature data set may optionally include a load rate parameter and / or a load rate difference parameter. Further optionally, data feature extraction operations can be performed on the enhanced load parameter data set corresponding to the subset of electrical parameter data to obtain a load angle feature data set corresponding to the enhanced current parameter data set. Specifically, the current drop amplitude parameter represents the ratio between the current change value at time i and the daily average current; the three-phase current imbalance parameter represents the deviation of the three-phase current amplitude; the current drop depth parameter reflects the degree of drop in current at time i compared to the current at time i-1; the current amplitude ratio represents the ratio of current at time i to rated current; the peak-valley difference parameter represents the ratio of the maximum difference between the current i at the current time and the effective current values ​​of the previous n times to the rated current; and the load factor parameter represents the ratio of the actual load borne by the transformer to its rated capacity, reflecting the load level of the transformer at the current time; the load factor difference parameter represents the difference in load factor between time i and time i-1, reflecting the change in load factor status; and the load angle characteristic data represents the angle between the current load value and the load value at the previous time.

[0114] The three-phase current imbalance parameter is calculated using the following feature extraction formula:

[0115]

[0116] Specifically, ε i I is the parameter for three-phase current imbalance. ai I bi I ci All of these are current parameter data of the corresponding phase type extracted from the enhanced current parameter data, I mi This is the average value of the current parameter data of all phase types extracted from the enhanced current parameter data.

[0117] The load angle feature data is calculated using the following feature extraction formula:

[0118]

[0119] Specifically, P n Let n be the load value at time n. Let θ be the average load value over the previous n time steps, and θ be the characteristic data of the angle between the loads.

[0120] As can be seen, this optional embodiment can selectively extract feature data from the enhanced current parameter data set and the corresponding enhanced load parameter data set corresponding to the subset of electrical parameter data to obtain the corresponding electrical parameter feature data set. This helps to improve the reliability and accuracy of the feature data extraction operation of the subset of electrical parameter data, and thus helps to improve the reliability and accuracy of the obtained electrical parameter feature data set. In turn, it helps to improve the reliability and accuracy of the subsequent fault detection model to be trained in analyzing the current drop situation through the electrical parameter feature data set.

[0121] In another optional embodiment, step 204 above, which involves performing a data equalization operation on the electrical parameter feature data set corresponding to each subset of electrical parameter data to obtain an equalized feature data set corresponding to each subset of electrical parameter data, includes:

[0122] For each subset of electrical parameter data, based on the label type of each electrical parameter feature data corresponding to the pre-determined subset of electrical parameter data, all target electrical parameter feature data are determined from all electrical parameter feature data, and all neighboring target electrical parameter feature data corresponding to each target electrical parameter feature data are determined.

[0123] For each target electrical parameter feature data in each subset of electrical parameter data, determine the sampling ratio of the neighboring data corresponding to the target electrical parameter feature data, and sample all neighboring target electrical parameter feature data corresponding to the target electrical parameter feature data according to the neighboring data sampling ratio, so as to update all neighboring target electrical parameter feature data corresponding to the target electrical parameter feature data.

[0124] For each target electrical parameter feature data in each subset of electrical parameter data, based on all updated neighboring target electrical parameter feature data corresponding to the target electrical parameter feature data, construct electrical parameter feature data between the target electrical parameter feature data and each updated neighboring target electrical parameter feature data;

[0125] For each subset of electrical parameter data, update the electrical parameter data subset based on all constructed electrical parameter feature data corresponding to the obtained electrical parameter data subset.

[0126] In this optional embodiment, specifically, the tag type of each electrical parameter feature data is associated with the current sag of the distribution network. For example, when all neighboring target electrical parameter feature data corresponding to any target electrical parameter feature data are determined, random sampling can be performed from all neighboring target electrical parameter feature data. Based on all the sampled neighboring target electrical parameter feature data, the constructed electrical parameter feature data between the target electrical parameter feature data and each sampled neighboring target electrical parameter feature data is calculated. Then, based on all the constructed electrical parameter feature data corresponding to any obtained subset of electrical parameter data, the subset of electrical parameter data is updated to achieve equalization of all electrical parameter data subsets, thereby achieving equalization of the overall electrical parameter dataset.

[0127] The corresponding electrical parameter feature data are as follows:

[0128] x new = x + rand(0,1) * (x` - x).

[0129] Specifically, x represents the target electrical parameter feature data, and x' represents any updated target electrical parameter feature data corresponding to the target electrical parameter feature data.

[0130] As can be seen, this optional embodiment can achieve data equalization of the electrical parameter data subset based on the calculated electrical parameter feature data. In this way, while ensuring the reliability and accuracy of the calculation of the electrical parameter feature data, it is beneficial to improve the reliability and accuracy of the data equalization operation of the electrical parameter data subset, which in turn is beneficial to improve the reliability and accuracy of the subsequent training of the fault detection model to be trained. This is conducive to accurately obtaining the target fault detection model that can be used to analyze the current drop situation and realize timely fault detection of the distribution network.

[0131] In another optional embodiment, the steps described above, including determining all target electrical parameter feature data from all electrical parameter feature data based on the label type of each electrical parameter feature data corresponding to a predetermined subset of electrical parameter data, and determining all neighboring target electrical parameter feature data corresponding to each target electrical parameter feature data, include:

[0132] Based on the label type to which each electrical parameter feature data belongs to the predetermined subset of electrical parameter data, determine all target electrical parameter feature data under the target label type whose data volume is less than or equal to the preset data volume threshold from all electrical parameter feature data;

[0133] For each target electrical parameter feature data, calculate the Euclidean distance parameter between the target electrical parameter feature data and all other target electrical parameter feature data. Based on the Euclidean distance parameter between the target electrical parameter feature data and all other target electrical parameter feature data, determine all other target electrical parameter feature data whose Euclidean distance parameter is less than or equal to a preset parameter threshold from all other target electrical parameter feature data, and use them as all neighboring target electrical parameter feature data corresponding to the target electrical parameter feature data.

[0134] In this optional embodiment, for any subset of electrical parameter data, target electrical parameter feature data belonging to the minority class is first determined from the subset of electrical parameter data. Then, by calculating the Euclidean distance parameter between it and other target electrical parameter feature data belonging to the minority class, all its corresponding neighboring target electrical parameter feature data are determined. This constructs the electrical parameter feature data between it and all its corresponding neighboring target electrical parameter feature data, thereby achieving the equalization of the subset of electrical parameter data.

[0135] As can be seen, this optional embodiment can determine all neighboring target electrical parameter feature data corresponding to the target electrical parameter feature data based on the calculated Euclidean distance parameter between the target electrical parameter feature data and all other target electrical parameter feature data. This can improve the reliability and accuracy of the obtained neighboring target electrical parameter feature data corresponding to the target electrical parameter feature data, thereby improving the reliability and accuracy of the calculation of the constructed electrical parameter feature data, which is conducive to achieving a precise equalization effect on the subset of electrical parameter data.

[0136] Example 3

[0137] Please refer to Figure 3, which is a schematic diagram of a distribution network fault intelligent detection device based on current descent analysis disclosed in an embodiment of the present invention. As shown in Figure 3, the distribution network fault intelligent detection device based on current descent analysis may include:

[0138] The acquisition module 301 is used to acquire the electrical parameter dataset of the distribution network to be processed;

[0139] Processing module 302 is used to perform data processing operations on the electrical parameter dataset of the distribution network to obtain the processed electrical parameter dataset;

[0140] Training module 303 is used to train a preset fault detection model based on the processed electrical parameter dataset to obtain a trained fault detection model.

[0141] Decision module 304 is used to determine whether the fault detection model has converged after training.

[0142] The determination module 305 is used to determine the trained fault detection model as the target fault detection model when the judgment module 304 determines that the trained fault detection model has converged.

[0143] In this embodiment of the invention, the data processing operations include data augmentation, data feature extraction, and data equalization; the target fault detection model is used to analyze the current drop of the distribution network to be judged based on the electrical parameter data of the distribution network to be judged, so as to detect whether there is a fault in the switching state of the distribution network to be judged based on the current drop; each subset of electrical parameter data in the electrical parameter dataset is a subset containing electrical parameter data of multiple phase types.

[0144] As can be seen, the intelligent distribution network fault detection device based on current descent analysis described in Figure 3 can train a target fault detection model for analyzing current descent situations. This model can detect whether there are faults in the distribution network through current descent analysis. In this way, even if telemetry information cannot be obtained, faults in the distribution network can be detected in a timely manner. This not only improves the efficiency of distribution network fault detection, thereby improving the timeliness of distribution network detection and maintenance, but also enhances the reliability, accuracy, and effectiveness of distribution network fault detection and maintenance work, so as to ensure the safe and stable operation of the power system and that users' electricity experience is not affected.

[0145] In an optional embodiment, the processing module 302 performs data processing operations on the electrical parameter dataset of the distribution network to obtain the processed electrical parameter dataset in the following specific manner:

[0146] Based on the electrical parameter data of all phase types contained in each electrical parameter data subset, perform data augmentation operation on each electrical parameter data subset to obtain the augmented electrical parameter data set corresponding to each electrical parameter data subset;

[0147] Perform data feature extraction on the enhanced electrical parameter data set corresponding to each subset of electrical parameter data to obtain the electrical parameter feature data set corresponding to each subset of electrical parameter data.

[0148] Perform data equalization on the electrical parameter feature data set corresponding to each electrical parameter data subset to obtain the equalized feature data set corresponding to each electrical parameter data subset, and determine the equalized feature data set corresponding to all electrical parameter data subsets as the processed electrical parameter dataset.

[0149] As can be seen, the intelligent fault detection device for distribution networks based on current descent analysis described in Figure 3 can process the electrical parameter dataset by enhancing, extracting features, and equalizing the corresponding subset of electrical parameter data. This achieves an intelligent processing method for the electrical parameter dataset, which helps improve the reliability and accuracy of the processing operation of the electrical parameter dataset. In turn, it helps improve the reliability and accuracy of the subsequent training operation of the fault detection model to be trained using the processed electrical parameter dataset, thereby facilitating the accurate training of the target fault detection model for analyzing current descent.

[0150] In another optional embodiment, the processing module 302 performs a data augmentation operation on each electrical parameter data subset based on all phase type electrical parameter data contained in each electrical parameter data subset, and obtains the augmented electrical parameter data set corresponding to each electrical parameter data subset in the following specific manner:

[0151] For each subset of electrical parameter data, the decimation weight corresponding to each phase type of electrical parameter data is determined based on all phase type electrical parameter data contained in the subset of electrical parameter data.

[0152] Based on the shunting weight corresponding to the electrical parameter data of each phase type, determine the shunting ratio corresponding to the electrical parameter data of each phase type;

[0153] Based on the decimation ratio corresponding to each phase type of electrical parameter data, multiple data decimation operations are performed in parallel on all phase types of electrical parameter data contained in the electrical parameter data subset to obtain all decimated electrical parameter data corresponding to the electrical parameter data subset, which serves as the enhanced electrical parameter data set corresponding to the electrical parameter data subset.

[0154] In this optional embodiment, each data shunting operation corresponds to a shunted electrical parameter data, and each shunted electrical parameter data contains electrical parameter data of all phase types that have been shunted; the enhanced electrical parameter data set corresponding to each subset of electrical parameter data includes an enhanced current parameter data set and an enhanced load parameter data set.

[0155] As can be seen, the intelligent fault detection device for distribution networks based on current descent analysis described in Figure 3 can perform merging operations on the corresponding subset of electrical parameter data according to the determined merging ratio, so as to enhance the electrical parameter dataset. This can improve the reliability and accuracy of the merging operation on the subset of electrical parameter data, thereby improving the effectiveness of the enhancement of the electrical parameter dataset, and thus significantly increasing the amount of reliable, accurate and effective electrical parameter dataset.

[0156] In another optional embodiment, the processing module 302 performs a data feature extraction operation on the enhanced electrical parameter data set corresponding to each subset of electrical parameter data to obtain the electrical parameter feature data set corresponding to each subset of electrical parameter data in the following specific manner:

[0157] For each subset of electrical parameter data, data feature extraction operations are performed on the enhanced current parameter data set and the enhanced load parameter data set corresponding to the subset of electrical parameter data to obtain the first feature data set corresponding to the enhanced current parameter data set and the second feature data set corresponding to the enhanced load parameter data set, which are used as the electrical parameter feature data set corresponding to the subset of electrical parameter data.

[0158] In this optional embodiment, the first feature data set includes at least one of the following: current descent amplitude parameter, three-phase current imbalance parameter, current drop depth, current amplitude ratio, and peak-valley difference parameter; the second feature data set includes load rate parameter and / or load rate difference parameter.

[0159] The three-phase current unbalance parameter is calculated using the following feature extraction formula:

[0160]

[0161] ε i I is the parameter for three-phase current imbalance. ai I bi I ci All of these are current parameter data of the corresponding phase type extracted from the enhanced current parameter data, I mi This is the average value of the current parameter data of all phase types extracted from the enhanced current parameter data.

[0162] As can be seen, the intelligent fault detection device for distribution networks based on current descent analysis described in Figure 3 can selectively extract feature data from the enhanced current parameter data set and the corresponding enhanced load parameter data set corresponding to the subset of electrical parameter data, thereby obtaining the corresponding electrical parameter feature data set. This helps improve the reliability and accuracy of the feature data extraction operation of the subset of electrical parameter data, which in turn helps improve the reliability and accuracy of the obtained electrical parameter feature data set. This, in turn, helps improve the reliability and accuracy of the subsequent fault detection model to be trained in analyzing the current descent situation through the electrical parameter feature data set.

[0163] In another optional embodiment, the processing module 302 performs a data equalization operation on the electrical parameter feature data set corresponding to each subset of electrical parameter data to obtain the equalized feature data set corresponding to each subset of electrical parameter data in the following specific manner:

[0164] For each subset of electrical parameter data, based on the label type of each electrical parameter feature data corresponding to the pre-determined subset of electrical parameter data, all target electrical parameter feature data are determined from all electrical parameter feature data, and all neighboring target electrical parameter feature data corresponding to each target electrical parameter feature data are determined.

[0165] For each target electrical parameter feature data in each subset of electrical parameter data, determine the sampling ratio of the neighboring data corresponding to the target electrical parameter feature data, and sample all neighboring target electrical parameter feature data corresponding to the target electrical parameter feature data according to the neighboring data sampling ratio, so as to update all neighboring target electrical parameter feature data corresponding to the target electrical parameter feature data.

[0166] For each target electrical parameter feature data in each subset of electrical parameter data, based on all updated neighboring target electrical parameter feature data corresponding to the target electrical parameter feature data, construct electrical parameter feature data between the target electrical parameter feature data and each updated neighboring target electrical parameter feature data;

[0167] For each subset of electrical parameter data, update the electrical parameter data subset based on all constructed electrical parameter feature data corresponding to the obtained electrical parameter data subset.

[0168] In this optional embodiment, the tag type to which each electrical parameter feature data belongs is associated with a current sag in the distribution network;

[0169] The corresponding electrical parameter feature data are as follows:

[0170] x new = x + rand(0,1) * (x` - x),

[0171] x represents the target electrical parameter feature data, and x' represents any updated target electrical parameter feature data corresponding to the target electrical parameter feature data.

[0172] As can be seen, the intelligent distribution network fault detection device based on current descent analysis described in Figure 3 can achieve data equalization of a subset of electrical parameter data based on the calculated electrical parameter feature data. This ensures the reliability and accuracy of the calculation of the electrical parameter feature data, while improving the reliability and accuracy of the data equalization operation of the electrical parameter data subset. This, in turn, improves the reliability and accuracy of the subsequent training of the fault detection model to be trained, thus facilitating the accurate acquisition of a target fault detection model that can be used to analyze current descent situations and enabling timely fault detection in the distribution network.

[0173] In another optional embodiment, the processing module 302 determines all target electrical parameter feature data from all electrical parameter feature data according to the tag type to which each electrical parameter feature data belongs in the predetermined subset of electrical parameter data, and determines all neighboring target electrical parameter feature data corresponding to each target electrical parameter feature data in the following specific way:

[0174] Based on the label type to which each electrical parameter feature data belongs to the predetermined subset of electrical parameter data, determine all target electrical parameter feature data under the target label type whose data volume is less than or equal to the preset data volume threshold from all electrical parameter feature data;

[0175] For each target electrical parameter feature data, calculate the Euclidean distance parameter between the target electrical parameter feature data and all other target electrical parameter feature data. Based on the Euclidean distance parameter between the target electrical parameter feature data and all other target electrical parameter feature data, determine all other target electrical parameter feature data whose Euclidean distance parameter is less than or equal to a preset parameter threshold from all other target electrical parameter feature data, and use them as all neighboring target electrical parameter feature data corresponding to the target electrical parameter feature data.

[0176] As can be seen, the intelligent fault detection device for distribution networks based on current descent analysis described in Figure 3 can determine all neighboring target electrical parameter feature data corresponding to the target electrical parameter feature data based on the Euclidean distance parameter between the calculated target electrical parameter feature data and all other target electrical parameter feature data. This can improve the reliability and accuracy of the obtained target electrical parameter feature data corresponding to all neighboring target electrical parameter feature data, thereby improving the reliability and accuracy of the calculation of the constructed electrical parameter feature data, which is conducive to achieving the effect of accurate equalization of the electrical parameter data subset.

[0177] Example 4

[0178] Please refer to Figure 4, which is a schematic diagram of another intelligent distribution network fault detection device based on current descent analysis disclosed in an embodiment of the present invention. As shown in Figure 4, the intelligent distribution network fault detection device based on current descent analysis may include:

[0179] Memory 401 storing executable program code;

[0180] Processor 402 coupled to memory 401;

[0181] The processor 402 calls the executable program code stored in the memory 401 to execute the steps in the intelligent fault detection method for distribution networks based on current descent analysis described in Embodiment 1 or Embodiment 2 of the present invention.

[0182] Example 5

[0183] This invention discloses a computer storage medium storing computer instructions. When these computer instructions are invoked, they are used to execute the steps in the intelligent fault detection method for distribution networks based on current descent analysis described in Embodiment 1 or Embodiment 2 of this invention.

[0184] Example 6

[0185] This invention discloses a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to cause a computer to perform the steps in the intelligent fault detection method for distribution networks based on current descent analysis described in Embodiment 1 or Embodiment 2.

[0186] The device embodiments described above are merely illustrative. The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0187] Through the detailed description of the above embodiments, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-Erasable Programmable Read-Only Memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0188] Finally, it should be noted that the intelligent detection method and device for distribution network faults based on current descent analysis disclosed in the embodiments of the present invention are merely preferred embodiments of the present invention and are only used to illustrate the technical solutions of the present invention, not to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for intelligent fault detection in distribution networks based on current descent analysis, characterized in that, The method includes: acquiring a power parameter dataset of the distribution network to be processed, and performing data processing operations on the power parameter dataset to obtain a processed power parameter dataset; the data processing operations include data augmentation, data feature extraction, and data equalization, wherein each subset of power parameter data in the power parameter dataset is a subset containing power parameter data of multiple phase types; training a preset fault detection model to be trained based on the processed power parameter dataset to obtain a trained fault detection model, and determining whether the trained fault detection model has converged; when it is determined that the trained fault detection model has converged, the trained fault detection model is determined as the target fault detection model; the target fault detection... The model is used to analyze the current drop situation of the distribution network to be judged based on the electrical parameter data of the distribution network to be judged, so as to detect whether there is a fault in the switching state of the distribution network to be judged based on the current drop situation; wherein, the step of performing data processing operation on the electrical parameter dataset of the distribution network to obtain a processed electrical parameter dataset includes: performing a data augmentation operation on each electrical parameter data subset based on all the phase type electrical parameter data contained in each electrical parameter data subset to obtain an augmented electrical parameter data set corresponding to each electrical parameter data subset; and performing a data feature extraction operation on the augmented electrical parameter data set corresponding to each electrical parameter data subset to obtain electrical parameter features corresponding to each electrical parameter data subset. Data set; for each subset of electrical parameter data, based on the predetermined label type of each electrical parameter feature data corresponding to the subset of electrical parameter data, all target electrical parameter feature data are determined from all electrical parameter feature data, and all neighboring target electrical parameter feature data corresponding to each target electrical parameter feature data are determined; the label type of each electrical parameter feature data is associated with the current drop situation of the distribution network; for each target electrical parameter feature data in each subset of electrical parameter data, the neighboring data sampling ratio corresponding to the target electrical parameter feature data is determined, and based on the neighboring data sampling ratio, all neighboring target electrical parameter feature data corresponding to the target electrical parameter feature data are processed. The target electrical parameter feature data is sampled to update all neighboring target electrical parameter feature data corresponding to the target electrical parameter feature data; for each target electrical parameter feature data in each subset of electrical parameter data, based on all updated neighboring target electrical parameter feature data corresponding to the target electrical parameter feature data, the constructed electrical parameter feature data between the target electrical parameter feature data and each updated neighboring target electrical parameter feature data is calculated; for each subset of electrical parameter data, the subset of electrical parameter data is updated based on all the constructed electrical parameter feature data corresponding to the obtained subset of electrical parameter data; the equalized feature data set corresponding to all subsets of electrical parameter data is determined as the processed electrical parameter dataset;The corresponding electrical parameter feature data is: x; new = x + rand(0,1)*(x`- x), where x is the target electrical parameter feature data, x` is any updated target electrical parameter feature data corresponding to the target electrical parameter feature data, and x new The electrical parameter feature data is constructed between the target electrical parameter feature data and the corresponding updated neighboring target electrical parameter feature data.

2. The intelligent fault detection method for distribution networks based on current descent analysis according to claim 1, characterized in that, The step of performing data augmentation on each electrical parameter data subset based on all phase-type electrical parameter data contained in each electrical parameter data subset to obtain an augmented electrical parameter data set corresponding to each electrical parameter data subset includes: for each electrical parameter data subset, determining the simplification weight corresponding to each phase-type electrical parameter data based on all phase-type electrical parameter data contained in the electrical parameter data subset; determining the simplification ratio corresponding to each phase-type electrical parameter data based on the simplification weight corresponding to each phase-type electrical parameter data; performing multiple data simplification operations in parallel on all phase-type electrical parameter data contained in the electrical parameter data subset based on the simplification ratio corresponding to each phase-type electrical parameter data to obtain all simplified electrical parameter data corresponding to the electrical parameter data subset, which serves as the augmented electrical parameter data set corresponding to the electrical parameter data subset; each data simplification operation corresponds to one simplified electrical parameter data, and each simplified electrical parameter data contains all phase-type electrical parameter data that has been simplified.

3. The intelligent fault detection method for distribution networks based on current descent analysis according to claim 1, characterized in that, Each subset of electrical parameter data corresponds to an enhanced set of electrical parameter data, which includes an enhanced set of current parameter data and an enhanced set of load parameter data. The step of performing data feature extraction on each subset of electrical parameter data to obtain a set of electrical parameter feature data includes: for each subset of electrical parameter data, performing data feature extraction on the enhanced set of current parameter data and the corresponding enhanced set of load parameter data to obtain a first set of feature data corresponding to the enhanced set of current parameter data and a second set of feature data corresponding to the enhanced set of load parameter data, which are used as the set of electrical parameter feature data corresponding to the subset of electrical parameter data. The first set of feature data includes at least one of current descent amplitude parameter, three-phase current imbalance parameter, current drop depth, current amplitude ratio, and peak-valley difference parameter; the second set of feature data includes a load rate parameter and / or a load rate difference parameter.

4. The intelligent fault detection method for distribution networks based on current descent analysis according to any one of claims 1-3, characterized in that, The step of determining all target electrical parameter feature data from all electrical parameter feature data according to the label type to which each electrical parameter feature data belongs to the predetermined subset of electrical parameter data, and determining all neighboring target electrical parameter feature data corresponding to each target electrical parameter feature data, includes: determining all target electrical parameter feature data under a target label type whose data volume is less than or equal to a preset data volume threshold, based on the label type to which each electrical parameter feature data belongs to the predetermined subset of electrical parameter data; for each target electrical parameter feature data, calculating the Euclidean distance parameter between the target electrical parameter feature data and all other target electrical parameter feature data, and determining all other target electrical parameter feature data whose Euclidean distance parameter is less than or equal to a preset parameter threshold from all other target electrical parameter feature data, as all neighboring target electrical parameter feature data corresponding to the target electrical parameter feature data.

5. The intelligent fault detection method for distribution networks based on current descent analysis according to claim 3, characterized in that, The three-phase current unbalance parameter is calculated using the following feature extraction formula: , ε i Let I be the three-phase current imbalance parameter. ai I bi I ci All of these are current parameter data of the corresponding phase type extracted from the enhanced current parameter data, I mi It is the average value of all current parameter data of the phase type extracted from the enhanced current parameter data.

6. A smart fault detection device for distribution networks based on current descent analysis, characterized in that, The apparatus is used to execute the intelligent fault detection method for distribution networks based on current descent analysis as described in any one of claims 1-5, and the apparatus comprises: an acquisition module for acquiring a dataset of electrical parameters of the distribution network to be processed; a processing module for performing data processing operations on the dataset of electrical parameters of the distribution network to obtain a processed dataset of electrical parameters; the data processing operations include data augmentation operations, data feature extraction operations, and data equalization operations; a training module for training a preset fault detection model to be trained based on the processed dataset of electrical parameters to obtain a trained fault detection model; a judgment module for judging whether the trained fault detection model has converged; and a determination module for determining the trained fault detection model as a target fault detection model when the judgment module judges that the trained fault detection model has converged; the target fault detection model is used to analyze the current descent of the distribution network to be judged based on the electrical parameter data of the distribution network to be judged, so as to detect whether there is a fault in the switching state of the distribution network to be judged based on the current descent.

7. A smart fault detection device for distribution networks based on current descent analysis, characterized in that, The device includes: a memory storing executable program code; a processor coupled to the memory; the processor calls the executable program code stored in the memory to execute the intelligent fault detection method for distribution networks based on current descent analysis as described in any one of claims 1-5.

8. A computer storage medium, characterized in that, The computer storage medium stores computer instructions, which, when invoked, are used to execute the intelligent fault detection method for distribution networks based on current descent analysis as described in any one of claims 1-5.

Images