A power distribution network panoramic intelligent sensing method

Abstract

The application discloses a power distribution network panoramic intelligent sensing method, and belongs to the technical field of power grid detection. The method comprises the following steps: dividing a power distribution network into multiple sections, collecting voltage data, current data and image data of each section; determining voltage mutation variables and current mutation variables of each section based on the voltage data and the current data of each section, and generating a mutation variable correlation one-dimensional vector corresponding to each section; constructing and training a fault positioning model based on a neural network; inputting the mutation variable correlation one-dimensional vector corresponding to each section and image data of the power distribution network into the trained fault positioning model to obtain a fault position of the power distribution network. The method can quickly and accurately locate a fault point based on panoramic sensing technology, thereby shortening a fault repair time and reducing a power outage range and time.

Description

Technical Field

[0001] This application relates to the field of power grid detection technology, and in particular to a panoramic intelligent sensing method for distribution networks. Background Technology

[0002] Distribution network panoramic perception refers to the comprehensive, multi-level, real-time, and accurate monitoring and perception of the distribution network's operating status, equipment status, and environmental information through the integrated use of various sensors, intelligent devices, and communication technologies. This allows for the acquisition of comprehensive, detailed, and accurate distribution network information, providing data support for reliable operation, optimized scheduling, and rapid fault handling of the distribution network.

[0003] Currently, the application of panoramic sensing technology for power distribution networks is mostly limited to the data monitoring stage, that is, real-time acquisition of electrical data of each node in the power distribution network; however, when a fault occurs in the power distribution network, it is impossible to quickly and accurately locate the fault point based on panoramic sensing technology. Summary of the Invention

[0004] To address the aforementioned shortcomings in existing technologies, this application provides a panoramic intelligent sensing method for power distribution networks, which solves the problem that current methods cannot quickly and accurately locate fault points based on panoramic sensing technology when a fault occurs in the power distribution network.

[0005] To achieve the aforementioned objectives, the technical solution adopted in this application is as follows: This application provides a panoramic intelligent sensing method for power distribution networks, including: S1: Divide the distribution network into multiple sections and collect voltage data, current data, and image data for each section; S2: Based on the voltage and current data of each segment, determine the voltage and current mutation amounts of each segment, and generate a one-dimensional vector of mutation amount correlation for each segment. S3: Construct and train a fault location model based on neural networks; S4: Input the one-dimensional vector of the correlation between the mutation amount of each section and the image data of the distribution network into the trained fault location model to obtain the fault location of the distribution network.

[0006] Further, S2 includes: S201: Calculate the voltage fluctuation in the voltage data of each section; S202: Calculate the current abrupt change in the current data of each section; S203: The correlation coefficient between voltage and current surges in each segment within one cycle after a fault occurs is calculated using a sliding window algorithm. S204: Based on the correlation coefficients of voltage and current mutations in each segment, form a one-dimensional vector of mutation correlation for each segment.

[0007] Furthermore, the formula for calculating the voltage surge is as follows:

[0008] In the formula, For the first Voltage fluctuations in voltage data for each segment For the first Voltage data for each section At the time of failure, The power frequency period is m, where m is an integer.

[0009] Furthermore, the formula for calculating the sudden change in current is:

[0010] In the formula, For the first The amount of sudden changes in current in the current data of each segment. For the first Current data for each section.

[0011] Furthermore, the formula for calculating the correlation coefficient between the voltage surge and the current surge is as follows:

[0012] In the formula, This is the correlation coefficient between voltage and current mutations. For each section within one power frequency cycle after a distribution network fault occurs, in the 1st... Voltage fluctuation at each sampling point For each section within one power frequency cycle after a distribution network fault occurs, in the 1st... Current change at each sampling point For the first time in one power frequency cycle One sampling point, This represents the number of sampling points within one power frequency cycle.

[0013] Furthermore, S2 also includes: The segment where the correlation coefficient between voltage and current mutations is not less than zero is marked as a healthy segment. The segments where the correlation coefficient between voltage and current mutations is less than zero are marked as unhealthy segments. Acquire image data of the non-healthy section, perform image recognition on the image data of the non-healthy section, and determine whether there are abnormally bright pixels in the electrical equipment in the image data of the non-healthy section; If so, mark the number of abnormally bright pixels in the image data of the non-healthy segment as the abnormal number.

[0014] Furthermore, the training process of the neural network-based fault location model includes: Obtain historical operation data for each segment within a preset time period. The historical operation data includes a one-dimensional vector of the correlation between historical mutations for each segment, the number of anomalies for each segment, and whether a fault occurred in each segment. The historical mutation correlation vectors of each segment within a preset time period and the number of anomalies in each segment are used as input parameters for a neural network-based fault location model. The model is trained by using whether a fault occurs in each segment as the output parameter.

[0015] Further, S4 includes: S401: Input the one-dimensional vector of mutation correlation corresponding to the non-healthy segment and the number of anomalies corresponding to the non-healthy segment into the trained neural network-based fault location model to obtain the fault prediction result of the non-healthy segment, wherein the fault prediction result includes whether the fault has occurred or not. S402: Mark the non-healthy section where the fault prediction result indicates a fault as the predicted fault section; S403: The location of the predicted fault section is taken as the fault location of the distribution network.

[0016] The beneficial effects of this application are: The present application provides a panoramic intelligent sensing method for power distribution networks that can quickly and accurately locate fault points based on panoramic sensing technology, thereby shortening fault repair time and reducing the scope and duration of power outages. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other embodiments can be obtained based on these drawings.

[0018] Figure 1 This is a flowchart illustrating a panoramic intelligent sensing method for a power distribution network provided in an embodiment of this application. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art based on this application are within the scope of protection of this application.

[0020] Example 1: This application provides a panoramic intelligent sensing method for power distribution networks. This method is applied to a panoramic intelligent sensing system for power distribution networks. The system includes a data acquisition module, a server, a drone, and a management terminal. The data acquisition module, the drone, and the management terminal are all communicatively connected to the server. The server controls the flight of the drone, and the drone is equipped with a camera module. This method can be found in [reference needed]. Figure 1 ,include: S1: Divide the power distribution network into multiple sections and collect voltage data, current data, and image data for each section.

[0021] In one embodiment of this application, the server divides the power distribution network into multiple segments, each segment being equipped with a corresponding acquisition module. The acquisition module includes a voltage sensor for acquiring voltage data of the corresponding segment and a current sensor for acquiring current data of the corresponding segment. The server acquires the voltage and current data corresponding to each segment of the power distribution network collected by the acquisition module. The server controls a drone to cruise the power distribution network and acquires image data of each segment captured by the drone's camera module during the cruise. This image data includes overhead photographs of the electrical equipment in each segment.

[0022] S2: Based on the voltage and current data of each segment, the server determines the voltage and current mutation rates of each segment and generates a one-dimensional vector of mutation rate correlation for each segment.

[0023] Specifically, it includes: S201: The server calculates the voltage fluctuation in the voltage data of each segment using the following formula:

[0024] In the formula, For the first Voltage fluctuations in voltage data for each segment For the first Voltage data for each section At the time of failure, The power frequency period is m, where m is an integer.

[0025] S202: The server calculates the current surge in the current data of each section using the following formula:

[0026] In the formula, For the first The amount of sudden changes in current in the current data of each segment. For the first Current data for each section.

[0027] S203: The server forms a one-dimensional vector of correlation between the voltage and current mutations in each segment, based on the correlation coefficient between the voltage and current mutations in each segment. The calculation formula is as follows:

[0028] In the formula, The correlation coefficient between voltage and current mutations, with subscripts. , These represent the voltage and current abrupt change sequences involved in the correlation calculation, respectively. For each section within one power frequency cycle after a distribution network fault occurs, in the 1st... The voltage fluctuation at each sampling point, the voltage fluctuation sequence is composed of voltage fluctuations Based on sampling points, For each section within one power frequency cycle after a distribution network fault occurs, in the 1st... The current abrupt change at each sampling point, the current abrupt change sequence is composed of current abrupt changes. Based on sampling points, For the first time in one power frequency cycle One sampling point, This represents the number of sampling points within one power frequency cycle.

[0029] In one embodiment of this application, the server marks segments where the correlation coefficient between voltage and current fluctuations is not less than 0 as healthy segments; and marks other segments as unhealthy segments. The server acquires image data of the unhealthy segments and performs image recognition on the image data of the unhealthy segments to determine whether there are abnormally bright pixels in the electrical equipment in the image data of the unhealthy segments. If so, the server marks the number of abnormally bright pixels in the image data of the unhealthy segments as the abnormal number. Here, abnormally bright pixels indicate that the electrical equipment has malfunctioned. The more abnormally bright pixels there are, the higher the degree of malfunction of the electrical equipment.

[0030] S3: Build and train a neural network-based fault location model.

[0031] In one embodiment of this application, the server obtains historical operational data for each segment over a preset time period (e.g., the past month). This historical operational data includes a one-dimensional vector of historical mutation correlation for each segment, the number of anomalies for each segment, and whether a fault occurred in each segment. The server uses the one-dimensional vector of historical mutation correlation for each segment over the preset time period and the number of anomalies for each segment as input parameters to a neural network-based fault location model, and uses whether a fault occurred in each segment as the output parameter to train the neural network-based fault location model.

[0032] S4: Input the one-dimensional vector of the correlation between the mutation amount of each section and the image data of the distribution network into the trained fault location model to obtain the fault location of the distribution network.

[0033] In one embodiment of this application, the server inputs the one-dimensional vector of mutation correlation corresponding to the non-healthy section and the number of anomalies corresponding to the non-healthy section into a fault location model based on a neural network to obtain the fault prediction result of the non-healthy section, wherein the fault prediction result includes whether a fault has occurred or not; the server marks the non-healthy section with the fault prediction result indicating a fault as the estimated fault section; the location of the estimated fault section is used as the fault location of the distribution network and sent to the management terminal, which includes a display module to display the location of the estimated fault section.

[0034] This application first divides the power distribution network into multiple sections. Then, it collects voltage and current data corresponding to each section of the power distribution network through a data acquisition module. Simultaneously, it controls a drone to patrol the power distribution network and acquires image data of each section captured by a camera module during the patrol. Based on the voltage and current data of each section, it determines the voltage and current fluctuations of each section and generates a one-dimensional vector of the fluctuation correlation for each section. Finally, it inputs the one-dimensional vector of the fluctuation correlation for each section and the image data of the power distribution network into a fault location model based on a neural network to obtain the fault location of the power distribution network. Therefore, this application can quickly and accurately locate fault points based on panoramic perception technology, thereby shortening fault repair time and reducing the scope and duration of power outages.

[0035] It should be noted that those skilled in the art will recognize that the embodiments described herein are for the purpose of helping readers understand the principles of this application, and should be understood as not limiting the scope of protection of this application to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this application without departing from the essence of this application, and these modifications and combinations are still within the scope of protection of this application.

Claims

1. A panoramic intelligent sensing method for power distribution networks, characterized in that, include: S1: Divide the distribution network into multiple sections and collect voltage data, current data, and image data for each section; S2: Based on the voltage and current data of each segment, determine the voltage and current mutation amounts of each segment, and generate a one-dimensional vector of mutation amount correlation for each segment. S3: Construct and train a fault location model based on neural networks; S4: Input the one-dimensional vector of the correlation between the mutation amount of each section and the image data of the distribution network into the trained fault location model to obtain the fault location of the distribution network.

2. The distribution network panoramic intelligent sensing method according to claim 1, characterized in that, S2 includes: S201: Calculate the voltage fluctuation in the voltage data of each section; S202: Calculate the current abrupt change in the current data of each section; S203: The correlation coefficient between voltage and current surges in each segment within one cycle after a fault occurs is calculated using a sliding window algorithm. S204: Based on the correlation coefficients of voltage and current mutations in each segment, form a one-dimensional vector of mutation correlation for each segment.

3. The distribution network panoramic intelligent sensing method according to claim 2, characterized in that, The formula for calculating the voltage surge is: In the formula, For the first Voltage fluctuations in voltage data for each segment For the first Voltage data for each section At the time of failure, The power frequency period is m, where m is an integer.

4. The distribution network panoramic intelligent sensing method according to claim 3, characterized in that, The formula for calculating the sudden change in current is: In the formula, For the first The amount of sudden changes in current in the current data of each segment. For the first Current data for each section.

5. The distribution network panoramic intelligent sensing method according to claim 4, characterized in that, The formula for calculating the correlation coefficient between the voltage surge and the current surge is: In the formula, This is the correlation coefficient between voltage and current mutations. For each section within one power frequency cycle after a distribution network fault occurs, in the 1st... Voltage fluctuation at each sampling point For each section within one power frequency cycle after a distribution network fault occurs, in the 1st... Current change at each sampling point For the first time in one power frequency cycle One sampling point, This represents the number of sampling points within one power frequency cycle.

6. The distribution network panoramic intelligent sensing method according to claim 5, characterized in that, S2 further includes: The segment where the correlation coefficient between voltage and current mutations is not less than zero is marked as a healthy segment. The segments where the correlation coefficient between voltage and current mutations is less than zero are marked as unhealthy segments. Acquire image data of the non-healthy section, perform image recognition on the image data of the non-healthy section, and determine whether there are abnormally bright pixels in the electrical equipment in the image data of the non-healthy section; If so, mark the number of abnormally bright pixels in the image data of the non-healthy segment as the abnormal number.

7. The distribution network panoramic intelligent sensing method according to claim 6, characterized in that, The training process of the neural network-based fault location model includes: Obtain historical operation data for each segment within a preset time period. The historical operation data includes a one-dimensional vector of the correlation between historical mutations for each segment, the number of anomalies for each segment, and whether a fault occurred in each segment. The historical mutation correlation vectors of each segment within a preset time period and the number of anomalies in each segment are used as input parameters for a neural network-based fault location model. The model is trained by using whether a fault occurs in each segment as the output parameter.

8. The distribution network panoramic intelligent sensing method according to claim 6, characterized in that, The S4 includes: S401: Input the one-dimensional vector of mutation correlation corresponding to the non-healthy segment and the number of anomalies corresponding to the non-healthy segment into the trained neural network-based fault location model to obtain the fault prediction result of the non-healthy segment, wherein the fault prediction result includes whether the fault has occurred or not. S402: Mark the non-healthy section where the fault prediction result indicates a fault as the predicted fault section; S403: The location of the predicted fault section is taken as the fault location of the distribution network.

Images