Low-voltage distribution network topology identification method and device based on voltage similarity evaluation

Abstract

The invention discloses a low-voltage distribution network topology identification method based on voltage similarity evaluation. The method comprises the following steps: acquiring oscillogram of transformer area side voltage and user side voltage; removing abnormal points on the oscillogram; performing interpolation processing on the abnormal points to supplement a complete oscillogram; and segmenting the complemented complete oscillogram, gradually comparing the segmented oscillogram through Hausdorff distance calculation, and stopping subsequent identification when the household transformer relation is identified in the process of comparing the plurality of segments of oscillogram. Through the comprehensive application of sectional comparison, abnormal data point elimination and an intelligent optimization algorithm, the recognition precision of the phase relation between the transformer area and the user side under the complex working condition is remarkably improved, the recognition efficiency is improved, through the application of a successive comparison strategy, a self-adaptive segmentation method and a deep learning algorithm, unnecessary calculation is greatly reduced, and the recognition efficiency is improved. The overall recognition efficiency is improved, and the real-time requirement is met.

Description

Technical Field

[0001] The present invention belongs to the technical field of power system automation and smart grid, and specifically relates to a method and device for identifying low-voltage distribution network topology based on voltage similarity evaluation. Background Art

[0002] In the daily operation and maintenance of low-voltage distribution networks, accurately identifying the phase relationship between the substation and the user side is of great significance for ensuring power supply quality, optimizing the grid structure, and improving fault response speed. Traditional identification methods mostly rely on manual field tests or complex mathematical model analysis, which is not only time-consuming and labor-intensive, but also easily affected by environmental factors, and it is difficult to meet the efficient and intelligent management needs of modern power grids. In recent years, methods based on voltage waveform similarity evaluation have attracted widespread attention due to their intuitive and fast characteristics. Among them, the Hausdorff distance algorithm has been applied in this field because of its ability to efficiently compare waveform similarity. However, there are still two major problems in practical applications: First, when the load fluctuation on the user side is large and the voltage difference is significant, it is difficult to accurately distinguish the substations by simply relying on voltage similarity; second, abnormal points in the data will seriously affect the accuracy and stability of recognition. Summary of the invention

[0003] The object of the present invention is to provide a method and device for identifying low-voltage distribution network topology based on voltage similarity evaluation, so as to solve the problems raised in the above background technology.

[0004] To achieve the above object, the present invention provides the following technical solution: a method for identifying low-voltage distribution network topology based on voltage similarity evaluation, comprising the following steps:

[0005] 1) Obtain the waveform diagram of the voltage on the station side and the voltage on the user side;

[0006] 2) Remove abnormal points on the waveform;

[0007] 3) Perform interpolation processing on abnormal points to supplement the complete waveform diagram;

[0008] 4) The completed waveform is segmented, and the segmented waveforms are compared one by one by calculating the Hausdorff distance. When a household change relationship is identified in the process of comparing several segments of the waveforms, the subsequent identification is stopped.

[0009] Preferably, in the step 1), the voltage waveform diagram has time as the horizontal axis and voltage value as the vertical axis, and records the voltage on the substation side and the voltage on the user side within the time period.

[0010] Preferably, in step 2), a box plot method is used to eliminate abnormal points;

[0011] 2.1) Calculate the quartiles of the voltage data (take the first quartile Q1 and the third quartile Q3), as well as the interquartile range IQR (IQR = Q3-Q1);

[0012] 2.2) Define the upper and lower limits of the voltage data, the upper limit is Q3+1.5*IQR, and the lower limit is Q1-1.5*IQR;

[0013] 2.3) Data points outside the upper and lower limits are considered as outliers and removed.

[0014] Preferably, in step 3), the waveform curve with abnormal values ​​is compared with the historical waveform curve to perform image recognition, and the most similar historical waveform curve is identified, and the voltage value at the same time in the identified waveform curve is the correct value of the abnormal point, which is added to the curve that eliminates the abnormal point.

[0015] Preferably, several similar waveform curves are selected from the historical waveform curves and entered into a test library, and the most similar historical waveform curve is identified from the test library during image recognition.

[0016] Preferably, the specific method for selecting historical waveform curves into the standby inspection library is: the voltage waveform data is sorted and stored in a manner corresponding to time and voltage, a number of time moments and their corresponding voltage values ​​are randomly selected as input on the waveform curve where abnormal points exist, and the waveforms of the same voltage at the same time are retrieved from the voltage waveform database and sent to the standby inspection library.

[0017] Preferably, in step 4), the minimum distance is set for the Hausdorff distance corresponding to the waveform image segment, and after the Hausdorff distance corresponding to the waveform image segment exceeds the minimum distance, it is determined that the household change relationship recognition is completed.

[0018] Preferably, a low voltage distribution network topology identification device based on voltage similarity evaluation comprises:

[0019] Waveform acquisition module: used to obtain waveforms of the voltage on the transformer side and the voltage on the user side;

[0020] Outlier removal module: removes outliers on the waveform;

[0021] Difference module: used to interpolate abnormal points to supplement the complete waveform diagram;

[0022] Segment comparison module: used to segment the completed waveform and compare the segmented waveforms one by one through Hausdorff distance calculation. When a change relationship is identified during the comparison of several waveforms, subsequent identification is stopped.

[0023] The technical effects and advantages of the present invention are as follows: 1. Improve recognition accuracy: Through the comprehensive application of segmented comparison, abnormal data point elimination and intelligent optimization algorithm, the recognition accuracy of the phase relationship between the station area and the user side under complex working conditions is significantly improved.

[0024] 2. Improve recognition efficiency: The application of successive comparison strategy, adaptive segmentation method and deep learning algorithm has greatly reduced unnecessary calculations, improved overall recognition efficiency and met real-time requirements.

[0025] 3. Enhanced adaptability: This method is not only applicable to standard low-voltage distribution network environments, but also shows good adaptability and robustness for areas with special load conditions, uneven data quality or complex grid structures.

[0026] 4. Promote intelligent operation and maintenance: The implementation of the present invention provides strong support for the intelligent operation and maintenance of low-voltage distribution networks, and helps to achieve rapid fault location, optimal resource allocation and continuous optimization of the grid structure. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 is a flow chart of the present invention;

[0028] Figure 2 Schematic diagram of the module of the device of the present invention. DETAILED DESCRIPTION

[0029] The specific embodiments of the present invention are further described below in conjunction with the accompanying drawings. It should be noted that the description of these embodiments is used to help understand the present invention, but does not constitute a limitation of the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0030] Embodiment 1

[0031] The present invention provides Figure 1 A low voltage distribution network topology identification method based on voltage similarity evaluation is shown, comprising the following steps:

[0032] Step 1: Obtain waveform diagrams of different substation-side voltages and user-side voltages; the voltage waveform diagram uses time as the horizontal axis and voltage value as the vertical axis, and records the substation-side voltage and user-side voltage within the time period.

[0033] Step 2: To solve the problem of abnormal points in the waveform data, the box plot method is used to eliminate the abnormal points. The specific steps of the box plot method are as follows:

[0034] Calculate the quartiles of the data (take the first quartile Q1 and the third quartile Q3), as well as the interquartile range IQR (IQR = Q3-Q1);

[0035] Define the upper and lower limits, the upper limit is Q3+1.5*IQR, and the lower limit is Q1-1.5*IQR;

[0036] Data points outside the upper and lower limits are considered outliers and removed.

[0037] Step 3: Establish a voltage waveform database, which contains historical voltage waveforms. By comparing the voltage waveform with abnormal points with the voltage waveform database, identify the most similar waveform curve in the voltage waveform database, and according to the time corresponding to the abnormal point, use the voltage value at the same time in the identified waveform curve as the correct value of the abnormal point, and add it into the curve excluding the abnormal point, so that the waveform voltage values ​​are accurate, avoiding the error in identifying the household change relationship due to abnormal values;

[0038] Since the voltage waveform is too large to be conducive to image recognition, the voltage waveform in the voltage waveform database is now sorted and stored in the form of (time, voltage), and several moments and their corresponding voltage values ​​are randomly selected on the waveform curve where abnormal points exist as input. The waveforms of the same voltage at the same time are retrieved from the voltage waveform database and sent to the inspection library. The most similar waveforms are then identified from the inspection library through a convolutional neural network based on deep learning. Through the setting of the inspection library, the number of recognitions is reduced, and the recognition efficiency of the convolutional neural network is greatly improved.

[0039] Step 4: Divide the obtained voltage waveforms on the substation side and the user side into several small segments according to time and voltage, each segment containing a certain amount of time and its corresponding voltage data; this can more accurately capture the local features of the voltage waveform and improve the accuracy of similarity comparison.

[0040] In order to avoid excessive computational burden caused by too many segments, the present invention adopts a successive comparison strategy. That is, when calculating the Hausdorff distance for each small segment of the waveform, once a clear user-to-user relationship (i.e., the corresponding relationship between the user and the corresponding area and phase) is identified, the comparison of subsequent segments is immediately stopped, thereby significantly reducing the amount of calculation while ensuring the recognition accuracy and improving the recognition efficiency;

[0041] The specific method of calculating Hausdorff distance is as follows:

[0042] First, set the voltage curve point set V of each terminal node to be identified bo (i) and the target transformer side voltage curve point set V t (j), i and j are node voltage numbers.

[0043] V bo (i) = {u 1 (i),u2 (i),...,u n (i)}

[0044] V t (j) = {u 1 (j),u 2 (j),...,u n (j)}

[0045] Thus, the hausdroff distance between each end node voltage curve point set to be identified and the known target substation transformer side voltage curve point set can be obtained:

[0046] H(V bo (i),V t (j)) = max[h(V bo (i),V t (j)),h(V t (j),V bo (i))]

[0047] in:

[0048]

[0049] a is the data point in the voltage curve point set of the terminal node, and b is the data point in the voltage curve point set of the transformer side of the target substation.

[0050] By calculating the Hausdorff distance between the voltage curve on the user side and the voltage curves on the sides of different substations as described above, the substation corresponding to the smallest Hausdorff distance is selected as the substation to which the correct user belongs.

[0051] It should be noted that the Hausdorff distances corresponding to several waveform segments should be set with a minimum distance, that is, the household transformer relationship identification is judged to be completed after exceeding the minimum distance. If the Hausdorff distances between the user-side voltage curve of the current waveform segment and the voltage curves on the different substation sides are all within the minimum distance, the Hausdorff distances between the user-side voltage curve of the next waveform segment and the voltage curves on the different substation sides are calculated.

[0052] The present invention proposes a method for accurately identifying the topology of a low-voltage distribution network based on voltage similarity evaluation and multi-strategy optimization. Through the comprehensive application of segmented Hausdorff comparison, abnormal data point elimination mechanism and intelligent optimization algorithm, the present invention effectively solves the problems of low recognition accuracy, large calculation amount and susceptibility to abnormal data interference of traditional recognition technology in complex power grid environments, thus providing a strong technical guarantee for the intelligent and efficient management of low-voltage distribution networks.

[0053] Embodiment 2

[0054] like Figure 2As shown, the present invention also provides a low-voltage distribution network topology identification device based on voltage similarity evaluation. This embodiment can be applied to identifying the household-to-substation relationship, that is, by comparing the waveform curves of the voltage on the substation side and the voltage on the user side, the user-side voltage curve and the substation side voltage curve with the smallest Hausdorff distance are taken to obtain the most accurate household-to-substation relationship; the low-voltage distribution network topology identification device based on voltage similarity evaluation can be implemented in the form of hardware and / or software, and the low-voltage distribution network topology identification device based on voltage similarity evaluation can be configured in an electronic device, such as a computer. Figure 2 As shown, the device comprises:

[0055] Waveform acquisition module: used to obtain waveforms of the voltage on the transformer side and the voltage on the user side;

[0056] Outlier removal module: removes outliers on the waveform;

[0057] Difference module: used to interpolate abnormal points to supplement the complete waveform diagram;

[0058] Segment comparison module: used to segment the completed waveform and compare the segmented waveforms one by one through Hausdorff distance calculation. When a change relationship is identified during the comparison of several waveforms, subsequent identification is stopped.

[0059] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

1. A low voltage distribution network topology identification method based on voltage similarity evaluation, characterized in that: The steps include: 1) Obtain the waveform diagram of the voltage on the station side and the voltage on the user side; 2) Remove abnormal points on the waveform; 3) Perform interpolation processing on abnormal points to supplement the complete waveform diagram; 4) The completed waveform is segmented, and the segmented waveforms are compared one by one by calculating the Hausdorff distance. When a household change relationship is identified in the process of comparing several segments of the waveforms, the subsequent identification is stopped.

2. A low voltage distribution network topology identification method based on voltage similarity evaluation according to claim 1, characterized in that: In the step 1), the voltage waveform diagram takes time as the horizontal axis and voltage value as the vertical axis, and records the voltage on the substation side and the voltage on the user side within the time period.

3. The method for identifying low-voltage distribution network topology based on voltage similarity evaluation according to claim 1, characterized in that: In the step 2), the box plot method is used to eliminate abnormal points; 2.1) Calculate the quartiles of the voltage data (take the first quartile Q1 and the third quartile Q3), as well as the interquartile range IQR (IQR = Q3-Q1); 2.2) Define the upper and lower limits of the voltage data, the upper limit is Q3+1.5*IQR, and the lower limit is Q1-1.5*IQR; 2.3) Data points outside the upper and lower limits are considered as outliers and removed.

4. The method for identifying low-voltage distribution network topology based on voltage similarity evaluation according to claim 1, characterized in that: In the step 3), the waveform curve with abnormal values ​​is compared with the historical waveform curve to perform image recognition, and the most similar historical waveform curve is identified. The voltage value at the same time in the identified waveform curve is the correct value of the abnormal point, which is added into the curve that eliminates the abnormal point.

5. A low voltage distribution network topology identification method based on voltage similarity evaluation according to claim 4, characterized in that: A number of similar waveform curves are selected from the historical waveform curves and entered into the inspection library. During image recognition, the most similar historical waveform curve is identified from the inspection library.

6. A low voltage distribution network topology identification method based on voltage similarity evaluation according to claim 5, characterized in that: The specific method for selecting historical waveform curves into the standby inspection library is as follows: the voltage waveform data is sorted and stored in the manner of corresponding time and voltage, a number of time points and their corresponding voltage values ​​are randomly selected as input on the waveform curve where abnormal points exist, and the waveforms of the same voltage at the same time are retrieved from the voltage waveform database and sent to the standby inspection library.

7. The method for identifying low-voltage distribution network topology based on voltage similarity evaluation according to claim 1, characterized in that: In the step 4), the minimum distance is set for the Hausdorff distance corresponding to the waveform image segment, and when the Hausdorff distance corresponding to the waveform image segment exceeds the minimum distance, it is determined that the household change relationship recognition is completed.

8. A low voltage distribution network topology identification device based on voltage similarity evaluation, characterized in that: include: Waveform acquisition module: used to obtain waveforms of the voltage on the transformer side and the voltage on the user side; Outlier removal module: removes outliers on the waveform; Difference module: used to interpolate abnormal points to supplement the complete waveform diagram; Segment comparison module: used to segment the completed waveform and compare the segmented waveforms one by one through Hausdorff distance calculation. When a change relationship is identified during the comparison of several waveforms, subsequent identification is stopped.

Images