A power transmission detection system based on AI video analysis

Abstract

The present invention provides a power transmission detection system based on AI video analysis, including: a front-end acquisition module, a cloud processing module and a background management module; the front-end acquisition module is used to fly along the power transmission line according to the set inspection route by a drone, and collect the power transmission line image data in real time, and transmit the collected power transmission line image data to the cloud processing module; the cloud processing module is used to integrate the power transmission line image data into the corresponding power transmission line position of the power transmission line map model for visual display; and further perform abnormal detection according to the power transmission line image data to obtain the power transmission line detection result; the background management module is used to access the cloud processing module to obtain real-time power transmission line map model data; and when the power transmission line detection result is abnormal, the corresponding on-site maintenance scheduling information is issued according to the abnormal detection result. The present invention helps to reduce the labor cost of power transmission line detection and improve the reliability level of the power transmission line of the power supply and distribution system.

Description

Technical Field

[0001] The present invention relates to the technical field of power transmission detection, and in particular to a power transmission detection system based on AI video analysis. Background Art

[0002] In the power supply and distribution system, transmission lines are an important component to ensure the reliable operation of large-scale power grids.

[0003] Since most transmission lines are set up outdoors, the outdoor environment is complex, which can easily lead to damage and safety problems. At present, power companies need to regularly send special maintenance personnel to the transmission line site to inspect and repair the lines. However, due to the wide range of transmission lines, maintenance personnel often need to climb mountains and wade through water to reach the transmission line site for maintenance. Not only is the working environment harsh, but there are also certain risks; at the same time, the maintenance efficiency is also low, which cannot meet the needs of large-scale power line inspections in cities. Summary of the invention

[0004] In view of the above problems, the present invention aims to provide a power transmission detection system based on AI video analysis.

[0005] The purpose of the present invention is achieved by the following technical solutions:

[0006] The present invention shows a power transmission detection system based on AI video analysis, including: a front-end acquisition module, a cloud processing module and a background management module;

[0007] The front-end acquisition module is used to fly the UAV along the transmission line according to the set inspection route, and collect the transmission line image data in real time during the flight, and transmit the collected transmission line image data to the cloud processing module, wherein the collected transmission line image data carries the positioning information at the time of data collection;

[0008] The cloud processing module is used to integrate the transmission line image data into the corresponding transmission line position of the transmission line map model for visual display according to the positioning information in the received transmission line image data; and further perform anomaly detection according to the transmission line image data to obtain the transmission line detection result, and integrate the transmission line detection result into the transmission line map model;

[0009] The background management module is used to access the cloud processing module to obtain real-time transmission line map model data; and when the transmission line detection results are abnormal, the corresponding on-site maintenance scheduling information is issued according to the abnormal detection results.

[0010] In one embodiment, the front-end acquisition module includes a drone unit, a positioning unit, a camera unit, and a communication unit;

[0011] The drone unit is used to obtain the preset inspection route and complete the flight inspection according to the training route;

[0012] The positioning unit is arranged in the drone unit, and is used to obtain the real-time positioning information of the drone, and send the obtained real-time positioning information to the camera unit;

[0013] The camera unit is used to collect the transmission line image data in real time while the UAV is flying along the inspection route, and integrate the acquired real-time positioning information into the transmission line image data;

[0014] The communication unit is used to transmit the power transmission line image data integrated with real-time positioning information to the cloud processing module through a wireless network.

[0015] In one embodiment, the cloud processing module includes a receiving unit, a preprocessing unit, an intelligent analysis unit, and a visualization unit;

[0016] The receiving unit is used to obtain the transmission line image data transmitted in real time by the front-end acquisition module;

[0017] The preprocessing unit is used to first preprocess the acquired power transmission line image data, and transmit the preprocessed power transmission line image data to the intelligent analysis unit and the visualization unit respectively;

[0018] The intelligent analysis unit is used to perform abnormality detection and analysis on the processed transmission line image data based on the intelligent image analysis technology to obtain the detection result of the transmission line;

[0019] The visualization unit is used to integrate the processed transmission line image data and the corresponding transmission line detection results into the corresponding positions in the transmission line map model based on the positioning information carried in the transmission line image data, and to visualize the transmission line map model.

[0020] In one embodiment, the cloud processing module further includes a data management unit;

[0021] The data management unit is used to store and manage the received transmission line image data and the corresponding detection and analysis results, and to build a historical detection database.

[0022] In one embodiment, the visualization unit further includes a model setting unit;

[0023] The model setting unit is used to construct a transmission line map model based on the actual transmission line layout information and the map model; wherein the transmission line map model includes the topological connection relationship of the transmission line, and the corresponding positions of the transmission line nodes and lines in the map.

[0024] In one embodiment, the intelligent analysis unit includes an input normalization unit and a model analysis unit;

[0025] The input normalization unit is used to construct an input set according to the preprocessed transmission line image data;

[0026] The model analysis unit is used to input the input set into the trained intelligent detection and analysis model, and the intelligent detection and analysis model performs abnormality detection on the transmission line according to the input set and outputs the detection result of the transmission line.

[0027] In one implementation, the intelligent analysis unit further includes a model training unit;

[0028] The model training unit is used to train the intelligent detection and analysis model according to a pre-prepared training set, and to test the trained intelligent detection and analysis model. When the model test result meets the preset standard, the model training is stopped and the trained intelligent detection and analysis model is output.

[0029] In one embodiment, the intelligent detection and analysis model is built based on a CNN convolutional neural network, wherein the model includes an input layer, a first convolutional layer, a second convolutional layer, a pooling layer, a first fully connected layer, a second fully connected layer, an activation layer, and an output layer;

[0030] The input layer is used to input the constructed input set into the model; the first convolution layer is connected to the input layer, wherein the first convolution layer contains 32 convolution kernels, and the size of each convolution kernel is 3×3; the second convolution layer is connected to the first convolution layer, wherein the second convolution layer contains 64 convolution kernels, and the size of each convolution kernel is 5×5; the pooling layer is connected to the second convolution layer, and the average pooling method is adopted, wherein the size of the pooling kernel is 3×3; the first fully connected layer is connected to the pooling layer, wherein the first fully connected layer contains 8 neurons; the second fully connected layer contains 16 neurons; that is, the first fully connected layer and the second fully connected layer contain a total of 512 fully connected layers; the activation layer is connected to the second fully connected layer, wherein the activation layer adopts the softmax function, and the activation layer calculates the probability of each transmission line detection classification according to the feature vector output by the fully connected layer; the output layer transforms according to the probability output by the activation layer, obtains the transmission line detection result and outputs it.

[0031] In one implementation, the background management module includes an access unit, a prompt unit, and a scheduling unit;

[0032] The access unit is used to access the cloud processing module to obtain the real-time transmission line model and display it;

[0033] The prompt unit is used to send a corresponding prompt message when the transmission line detection result in the transmission line model is abnormal;

[0034] The dispatching unit is used to generate on-site maintenance dispatching information according to the abnormal maintenance results when the transmission line detection results are abnormal, and send the on-site maintenance dispatching information to the corresponding inspection terminal.

[0035] In one implementation, the background management module also includes a calling unit;

[0036] The retrieval unit is used to obtain and play historical transmission line image data from the database of the cloud processing module, so that experts can perform manual inspection based on the played transmission line image data to obtain corresponding transmission line manual inspection results.

[0037] The beneficial effects of the present invention are as follows: the front-end acquisition module cruises along the route arranged by the unmanned aerial vehicle according to the transmission line, and collects the transmission line image data in real time during the flight, and transmits the collected data to the cloud processing module for intelligent analysis and processing; wherein the detection of the transmission line based on the unmanned aerial vehicle can improve the detection efficiency and detection safety of the transmission line, especially the transmission line in remote or secluded areas, and effectively reduce the labor cost of transmission line detection; the cloud processing module performs intelligent analysis on the collected transmission line image data, and with the help of the powerful computing and data storage capabilities of the cloud processing module, it can simultaneously perform real-time analysis and detection on a large amount of transmission line image data to obtain the transmission line detection results, which helps to improve the objectivity and reliability of the transmission line detection through intelligent detection, and avoids the occurrence of inaccurate subjective judgments in the manual detection process; at the same time, based on the transmission line map, the transmission line maintenance conditions in different areas are visualized, which helps to clearly display the detection conditions of complex transmission lines in the area and improve the management level of the transmission line detection results. Through the management module, managers can remotely view and manage the detection data of the transmission lines, and make corresponding operation and maintenance dispatches according to the abnormal detection results, so as to deal with the abnormal conditions of the transmission lines in time, thereby improving the reliability level of the transmission lines of the power supply and distribution system. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The present invention is further described using the accompanying drawings, but the embodiments in the accompanying drawings do not constitute any limitation to the present invention. A person skilled in the art can obtain other drawings based on the following drawings without creative work.

[0039] Figure 1 It is a framework structure diagram of the present invention;

[0040] Figure 2 This is a framework diagram of the front-end acquisition module of the present invention;

[0041] Figure 3 This is a framework diagram of the cloud processing module of the present invention;

[0042] Figure 4 It is a framework diagram of the background management module of the present invention. DETAILED DESCRIPTION

[0043] The present invention is further described in conjunction with the following application scenarios.

[0044] See also Figure 1 The embodiment shows a power transmission detection system based on AI video analysis, including: a front-end acquisition module, a cloud processing module and a background management module;

[0045] The front-end acquisition module is used to fly the UAV along the transmission line according to the set inspection route, and collect the transmission line image data in real time during the flight, and transmit the collected transmission line image data to the cloud processing module, wherein the collected transmission line image data carries the positioning information at the time of data collection;

[0046] The cloud processing module is used to integrate the transmission line image data into the corresponding transmission line position of the transmission line map model for visual display according to the positioning information in the received transmission line image data; and further perform anomaly detection according to the transmission line image data to obtain the transmission line detection result, and integrate the transmission line detection result into the transmission line map model;

[0047] The background management module is used to access the cloud processing module to obtain real-time transmission line map model data; and when the transmission line detection results are abnormal, the corresponding on-site maintenance scheduling information is issued according to the abnormal detection results.

[0048] In the above-mentioned embodiment of the present invention, the power transmission detection system proposed, the front-end acquisition module is based on the UAV cruising flight along the route arranged by the power transmission line, and the power transmission line image data is collected in real time during the flight, and the collected data is transmitted to the cloud processing module for intelligent analysis and processing; wherein the power transmission line detection based on the UAV can improve the detection efficiency and detection safety of the power transmission line, especially the power transmission line in remote or secluded areas, and effectively reduce the labor cost of power transmission line detection; based on the cloud processing module, the collected power transmission line image data is intelligently analyzed, and with the help of the powerful computing and data storage capabilities of the cloud processing module, a large amount of power transmission line image data can be analyzed and detected in real time at the same time to obtain the power transmission line detection results, which helps to improve the objectivity and reliability of the power transmission line detection through intelligent detection, and avoids the occurrence of inaccurate subjective judgments in the manual detection process; at the same time, based on the transmission line map, the maintenance conditions of the transmission lines in different areas are visualized, which helps to clearly display the detection conditions of complex transmission lines in the area and improve the management level of the transmission line detection results. Through the management module, managers can remotely view and manage the detection data of the transmission lines, and make corresponding operation and maintenance dispatches according to the abnormal detection results, so as to deal with the abnormal conditions of the transmission lines in time, thereby improving the reliability level of the transmission lines of the power supply and distribution system.

[0049] In one embodiment, see Figure 2 ,The front-end acquisition module includes a drone unit, a positioning unit, a camera unit and a communication unit;

[0050] The drone unit is used to obtain the preset inspection route and complete the flight inspection according to the training route;

[0051] The positioning unit is arranged in the drone unit, and is used to obtain the real-time positioning information of the drone, and send the obtained real-time positioning information to the camera unit;

[0052] The camera unit is used to collect the transmission line image data in real time while the UAV is flying along the inspection route, and integrate the acquired real-time positioning information into the transmission line image data;

[0053] The communication unit is used to transmit the power transmission line image data integrated with real-time positioning information to the cloud processing module through a wireless network.

[0054] In one scenario, the inspection flight route of the UAV is set according to the layout of the transmission line, wherein the inspection route is set according to the distance between 1 meter and 2 meters above the transmission line. The UAV adopts a four-wing UAV, and the flight speed of the UAV is limited to 10-30km / h; a camera unit is set under the UAV, and the camera unit adopts a CCD high-definition camera. The camera unit is aimed at the transmission line below the UAV during the flight of the UAV to collect image data of the transmission line, and the current real-time positioning information of the UAV is integrated into the image data to calibrate the transmission line data in the image data; finally, the collected transmission line image data is transmitted to the cloud processing module through wireless communication for further storage, display and analysis.

[0055] The use of drones to inspect power transmission lines can help improve the efficiency and flexibility of inspections and meet the needs of power transmission line inspections in different scenarios.

[0056] In one embodiment, see Figure 3 ,The cloud processing module includes a receiving unit, a preprocessing unit, an intelligent analysis unit and a visualization unit;

[0057] The receiving unit is used to obtain the transmission line image data transmitted in real time by the front-end acquisition module;

[0058] The preprocessing unit is used to first preprocess the acquired power transmission line image data, and transmit the preprocessed power transmission line image data to the intelligent analysis unit and the visualization unit respectively;

[0059] The intelligent analysis unit is used to perform abnormality detection and analysis on the processed transmission line image data based on the intelligent image analysis technology to obtain the detection result of the transmission line;

[0060] The visualization unit is used to integrate the processed transmission line image data and the corresponding transmission line detection results into the corresponding positions in the transmission line map model based on the positioning information carried in the transmission line image data, and to visualize the transmission line map model.

[0061] In one embodiment, the cloud processing module further includes a data management unit;

[0062] The data management unit is used to store and manage the received transmission line image data and the corresponding detection and analysis results, and to build a historical detection database.

[0063] In one scenario, a cloud processing module is built based on a cloud platform or a private cloud server, wherein the cloud processing module is wirelessly connected to the drone of the front-end acquisition module, and receives the transmission line image data uploaded by the drone in real time; based on the obtained transmission line image data, the image data is first preprocessed to perform data cleaning, data standardization, data analysis, enhancement processing, target extraction and other preprocessing on the received data, so that the acquired transmission line image data meets the needs of subsequent intelligent analysis and visual display.

[0064] In one implementation, the preprocessing unit first preprocesses the acquired power transmission line image data, including performing enhancement processing on the acquired power transmission line image data, specifically including:

[0065] The received power transmission line image data is framed to obtain each frame image, and each frame image is enhanced in turn, including:

[0066] Perform edge detection based on the obtained frame image, obtain edge pixel information in the frame image, and count the edge pixel set U;

[0067] Perform a noise point detection on the pixels in the set U according to the following primary detection function, and mark the pixels that meet the following primary detection function as suspected pixels:

[0068]

[0069] Among them, hgp(x,y) represents the gray level of the pixel (x,y), hgp min Indicates the minimum gray level of each pixel in the current video frame, hgp max Indicates the maximum gray level of each pixel in the current video frame;

[0070] The suspected pixels are further tested for secondary noise points according to the following primary detection function, and the pixels that meet the following secondary detection function are marked as noise pixels:

[0071]

[0072] Among them, hgp(x,y) represents the gray level of the pixel (x,y), Indicates the minimum gray level of other pixels except (x, y) in the neighborhood centered on the pixel (x, y). Represents the average gray level of other pixels except (x, y) in the neighborhood centered on the pixel (x, y); It represents the maximum gray level of other pixels except (x, y) in the neighborhood centered on the pixel (x, y); hgpTe represents the set gray level threshold;

[0073] Remove the noise pixels in the edge pixel set U;

[0074] The frame image is further converted from the RGB color space to the Lab color space to obtain the brightness component L, color component a and color component b of the image respectively;

[0075] For the pixel points in the edge pixel point set, the following first brightness enhancement function is used to perform brightness enhancement processing:

[0076] lumi u (x,y)=0.5×lumiSt+0.5×lumi(x,y)

[0077] Among them, lumi u (x, y) represents the brightness component level of the pixel point (x, y) after the first enhancement process, where (x, y)∈[0, 100]; lumiSt represents the set standard brightness component level, and lumi(x, y) represents the brightness component level of the pixel point (x, y) in the brightness component L;

[0078] For the pixels in the non-edge pixel set, the following second brightness enhancement function is used to perform brightness enhancement processing:

[0079]

[0080] lumi u (x, y) represents the brightness component level of the pixel point (x, y) after the first enhancement process, lumiSt represents the set standard brightness component level, the pixel point (a, b) represents the edge pixel point closest to the pixel point (x, y) in the pixel point set U, lumi argmin(||(x,y)-(a,b)||),(a,b)∈U (a,b) represents the brightness component level of pixel (a,b); Represents the average brightness component level of each pixel in the neighborhood centered on the pixel point (x, y);

[0081] After completing the brightness enhancement for all pixels, the brightness component L after brightness enhancement is obtained. u , and reconstruct according to the corresponding color component a and color component b to obtain an enhanced frame image;

[0082] The pre-processed power transmission line image data is reassembled according to the obtained enhanced frame images of each frame.

[0083] For the transmission line images taken from an upward or downward angle, the transmission line is affected by sunlight or ground lights, which causes backlighting of the transmission line, affecting the clarity of the transmission line area in the transmission line image data. The above-mentioned implementation method particularly proposes a technical solution for preprocessing the transmission line image. First, edge detection is performed on the transmission line image to obtain edge information in the image. Considering that the traditional edge detection algorithm through grayscale gradient or LBP is easy to misjudge noise points as edge pixels, therefore, after completing the edge pixel detection, the edge pixel is further subjected to primary and secondary noise point detection, which can accurately eliminate the noise points misjudged as edge pixels, and provide a guarantee for the subsequent overall brightness enhancement of the image based on the edge brightness information.

[0084] After acquiring the edge information of the image, the Lab brightness component of the image is obtained to perform brightness enhancement processing on the edge pixels and non-edge pixels respectively. When enhancing the non-edge pixels, the brightness information of the surrounding edge pixels is specially referred to as a basis for brightness compensation adjustment, which helps to complete adaptive brightness adjustment for the backlight area. At the same time, adaptive equalization adjustment is performed for the exposure area, which helps to improve the overall clarity and viewing level of the image, and improve the accuracy and reliability of subsequent intelligent analysis based on the transmission line image.

[0085] In one embodiment, the visualization unit further includes a model setting unit;

[0086] The model setting unit is used to construct a transmission line map model based on the actual transmission line layout information and the map model; wherein the transmission line map model includes the topological connection relationship of the transmission line, and the corresponding positions of the transmission line nodes and lines in the map.

[0087] By combining the regional transmission line layout with CIS map data, a transmission line map model corresponding to the actual situation is constructed, and the detection status of the transmission line is visualized based on the constructed transmission line map model, which helps to improve the intuitiveness of data display and indirectly improves the management level of transmission line detection.

[0088] In one embodiment, the intelligent analysis unit includes an input normalization unit and a model analysis unit;

[0089] The input normalization unit is used to construct an input set according to the preprocessed transmission line image data;

[0090] The model analysis unit is used to input the input set into the trained intelligent detection and analysis model, and the intelligent detection and analysis model performs abnormality detection on the transmission line according to the input set and outputs the detection result of the transmission line.

[0091] Among them, the intelligent detection and analysis of transmission line image data is realized by building an intelligent detection and analysis model based on an artificial intelligence network. The detection of transmission lines is completed through the intelligent analysis model, which helps to improve the intelligence level and objectivity of transmission line detection. At the same time, the intelligent detection of transmission lines based on AI analysis also helps to improve the efficiency of data processing.

[0092] In one embodiment, the intelligent detection and analysis model is built based on a CNN convolutional neural network, wherein the model includes an input layer, a first convolutional layer, a second convolutional layer, a pooling layer, a first fully connected layer, a second fully connected layer, an activation layer, and an output layer;

[0093] The input layer is used to input the constructed input set into the model; the first convolution layer is connected to the input layer, wherein the first convolution layer contains 32 convolution kernels, and the size of each convolution kernel is 3×3; the second convolution layer is connected to the first convolution layer, wherein the second convolution layer contains 64 convolution kernels, and the size of each convolution kernel is 5×5; the pooling layer is connected to the second convolution layer, and the average pooling method is adopted, wherein the size of the pooling kernel is 3×3; the first fully connected layer is connected to the pooling layer, wherein the first fully connected layer contains 8 neurons; the second fully connected layer contains 16 neurons; that is, the first fully connected layer and the second fully connected layer contain a total of 512 fully connected layers; the activation layer is connected to the second fully connected layer, wherein the activation layer adopts the softmax function, and the activation layer calculates the probability of each transmission line detection classification according to the feature vector output by the fully connected layer; the output layer transforms according to the probability output by the activation layer, obtains the transmission line detection result and outputs it.

[0094] In one scenario, the model input is a picture frame image obtained based on the transmission line image data and a transmission line area image obtained through target extraction. The picture frame image is extracted by performing frame processing based on the transmission line image, and the transmission line target is extracted based on the picture frame image to obtain the area where the transmission line is located in the picture frame image, and the extracted area image is input as an input image into the intelligent detection and analysis model. The intelligent detection and analysis model detects the external condition of the transmission line based on the input image to detect whether the transmission line has abnormal conditions such as surface rupture, cracks, breakage, and detachment, and obtains the detection and analysis results.

[0095] In one implementation, the intelligent analysis unit further includes a model training unit;

[0096] The model training unit is used to train the intelligent detection and analysis model according to a pre-prepared training set, and to test the trained intelligent detection and analysis model. When the model test result meets the preset standard, the model training is stopped and the trained intelligent detection and analysis model is output.

[0097] In one scenario, images of transmission line areas that have been preprocessed based on transmission line image data collected by the same front-end acquisition module are used as training set images, and experts identify abnormal situations based on the training set images to complete the construction of the training set; 80% of the training set is input into the intelligent detection and analysis model as training data to train the model, and 20% of the training set is used as test data to test the accuracy of the model. When the test result of the model meets the preset standard, for example, the accuracy rate is higher than 98%, the training of the model is stopped and the trained intelligent detection and analysis model is output. Otherwise, the training set is continued to be constructed to train and test the model until the model meets the preset standard.

[0098] In one embodiment, see Figure 4 ,The background management module includes an access unit, a prompt unit and a scheduling unit;

[0099] The access unit is used to access the cloud processing module to obtain the real-time transmission line model and display it;

[0100] The prompt unit is used to send a corresponding prompt message when the transmission line detection result in the transmission line model is abnormal;

[0101] The dispatching unit is used to generate on-site maintenance dispatching information according to the abnormal maintenance results when the transmission line detection results are abnormal, and send the on-site maintenance dispatching information to the corresponding inspection terminal.

[0102] The background management module can be built based on the intelligent terminal of the power smart center. The management personnel access the cloud processing platform through the background management module to obtain the transmission line map model for visual display, which can comprehensively understand the detection status of the transmission lines in the control area, which is helpful for the managers to complete the overall management of the regional transmission lines; at the same time, based on the lines with abnormal detection results, abnormal prompt messages will appear to remind the managers to deal with the abnormal detection results in time, generate on-site maintenance scheduling information through the scheduling unit, and send the on-site maintenance scheduling information to the responsible maintenance personnel. The maintenance personnel will arrive at the site of the abnormal transmission line to carry out on-site maintenance to deal with the abnormal situation, which will help improve the efficiency and management level of abnormal handling of transmission lines.

[0103] In one implementation, the background management module also includes a calling unit;

[0104] The retrieval unit is used to obtain and play historical transmission line image data from the database of the cloud processing module, so that experts can perform manual inspection based on the played transmission line image data to obtain corresponding transmission line manual inspection results.

[0105] Through the retrieval unit, the collected historical transmission line image data can be played, which helps experts or operation and maintenance personnel to conduct secondary manual inspection (remote inspection) based on the played image data to meet the needs of transmission line inspection in different scenarios.

[0106] In one implementation, the backend management module further includes a frontend management unit;

[0107] The front-end management unit is used to manage the drone units in the front-end management module, including setting inspection flight routes and assigning inspection tasks.

[0108] It should be noted that each functional unit / module in each embodiment of the present invention may be integrated into one processing unit / module, or each unit / module may exist physically separately, or two or more units / modules may be integrated into one unit / module. The above-mentioned integrated unit / module may be implemented in the form of hardware or in the form of software functional unit / module.

[0109] Through the description of the above implementation mode, it can be clearly understood by those skilled in the art that the embodiments described herein can be implemented in hardware, software, firmware, middleware, code or any appropriate combination thereof. For hardware implementation, the processor can be implemented in one or more of the following units: application specific integrated circuit (ASIC), digital signal processor (DSP), digital signal processing device (DSPD), programmable logic device (PLD), field programmable gate array (FPGA), processor, controller, microcontroller, microprocessor, other electronic units designed to implement the functions described herein or their combination. For software implementation, part or all of the process of the embodiment can be completed by instructing the relevant hardware through a computer program. When implemented, the above program can be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein the communication medium includes any medium that facilitates the transmission of a computer program from one place to another. The storage medium can be any available medium that can be accessed by a computer. Computer-readable media can include, but are not limited to, RAM, ROM, EEPROM, CD-ROM or other optical disk storage, disk storage media or other magnetic storage devices, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer.

[0110] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention, rather than to limit the scope of protection of the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should analyze that the technical solution of the present invention can be modified or replaced by equivalents without departing from the essence and scope of the technical solution of the present invention.

Claims

1. A power transmission detection system based on AI video analysis, characterized in that: include: Front-end acquisition module, cloud processing module and back-end management module; The front-end acquisition module is used to fly the UAV along the transmission line according to the set inspection route, and collect the transmission line image data in real time during the flight, and transmit the collected transmission line image data to the cloud processing module, wherein the collected transmission line image data carries the positioning information at the time of data collection; The cloud processing module is used to integrate the transmission line image data into the corresponding transmission line position of the transmission line map model for visual display according to the positioning information in the received transmission line image data; Furthermore, anomaly detection is performed based on the transmission line image data to obtain a transmission line detection result, and the transmission line detection result is integrated into the transmission line map model; The backend management module is used to access the cloud processing module to obtain real-time transmission line map model data; When the transmission line detection results are abnormal, the corresponding on-site maintenance dispatch information is issued according to the abnormal detection results; The cloud processing module pre-processes the received power transmission line image data, that is, performs frame processing on the received power transmission line image data to obtain each frame image, and sequentially performs enhancement processing on each frame image, including: Perform edge detection based on the obtained frame image, obtain edge pixel information in the frame image, and count the edge pixel set U; Perform a noise point detection on the pixels in the set U according to the following primary detection function, and mark the pixels that meet the following primary detection function as suspected pixels: Among them, hgp(x,y) represents the gray level of the pixel (x,y), hgp min Indicates the minimum gray level of each pixel in the current video frame, hgp max Indicates the maximum gray level of each pixel in the current video frame; Further secondary noise point detection is performed on the suspected pixels according to the following secondary detection function, and the pixels that meet the following secondary detection function are marked as noise pixels: Among them, hgp(x,y) represents the gray level of the pixel (x,y), They represent the minimum gray level, average gray level, and maximum gray level of other pixels except (x, y) in the neighborhood centered on the pixel (x, y); hgpTe represents the set gray level threshold; Remove the noise pixels in the edge pixel set U; Further converting the picture frame image from RGB color space to Lab color space; Brightness enhancement processing is performed on edge pixels and non-edge pixels respectively, wherein when the non-edge pixels are enhanced, brightness information of the surrounding edge pixels is referred to as a basis to perform brightness compensation adjustment.

2. According to claim 1, a power transmission detection system based on AI video analysis is characterized in that: The front-end acquisition module includes a drone unit, a positioning unit, a camera unit, and a communication unit; The drone unit is used to obtain the preset inspection route and complete the flight inspection according to the inspection route; The positioning unit is arranged in the drone unit, and is used to obtain the real-time positioning information of the drone, and send the obtained real-time positioning information to the camera unit; The camera unit is used to collect the transmission line image data in real time while the UAV is flying along the inspection route, and integrate the acquired real-time positioning information into the transmission line image data; The communication unit is used to transmit the power transmission line image data integrated with real-time positioning information to the cloud processing module through a wireless network.

3. The power transmission detection system based on AI video analysis according to claim 2 is characterized in that: The cloud processing module includes a receiving unit, a preprocessing unit, an intelligent analysis unit, and a visualization unit; The receiving unit is used to obtain the transmission line image data transmitted in real time by the front-end acquisition module; The preprocessing unit is used to first preprocess the acquired power transmission line image data, and transmit the preprocessed power transmission line image data to the intelligent analysis unit and the visualization unit respectively; The intelligent analysis unit is used to perform abnormality detection and analysis on the processed transmission line image data based on the intelligent image analysis technology to obtain the detection result of the transmission line; The visualization unit is used to integrate the processed transmission line image data and the corresponding transmission line detection results into the corresponding positions in the transmission line map model based on the positioning information carried in the transmission line image data, and to visualize the transmission line map model.

4. The power transmission detection system based on AI video analysis according to claim 3 is characterized in that: The cloud processing module also includes a data management unit; The data management unit is used to store and manage the received transmission line image data and the corresponding detection and analysis results, and to build a historical detection database.

5. The power transmission detection system based on AI video analysis according to claim 3 is characterized in that: The visualization unit also includes a model setting unit; The model setting unit is used to construct a transmission line map model based on the actual transmission line layout information and the map model; wherein the transmission line map model includes the topological connection relationship of the transmission line, and the corresponding positions of the transmission line nodes and lines in the map.

6. The power transmission detection system based on AI video analysis according to claim 3 is characterized in that: The intelligent analysis unit includes an input standardization unit and a model analysis unit; The input normalization unit is used to construct an input set according to the preprocessed transmission line image data; The model analysis unit is used to input the input set into the trained intelligent detection and analysis model, and the intelligent detection and analysis model performs abnormality detection on the transmission line according to the input set and outputs the detection result of the transmission line.

7. The power transmission detection system based on AI video analysis according to claim 6, characterized in that: The intelligent analysis unit also includes a model training unit; The model training unit is used to train the intelligent detection and analysis model according to a pre-prepared training set, and to test the trained intelligent detection and analysis model. When the model test result meets the preset standard, the model training is stopped and the trained intelligent detection and analysis model is output.

8. The power transmission detection system based on AI video analysis according to claim 6, characterized in that: The intelligent detection and analysis model is built based on the CNN convolutional neural network, where the model includes an input layer, a first convolutional layer, a second convolutional layer, a pooling layer, a first fully connected layer, a second fully connected layer, an activation layer, and an output layer; The input layer is used to input the constructed input set into the model; the first convolution layer is connected to the input layer, wherein the first convolution layer contains 32 convolution kernels, and the size of each convolution kernel is 3×3; the second convolution layer is connected to the first convolution layer, wherein the second convolution layer contains 64 convolution kernels, and the size of each convolution kernel is 5×5; the pooling layer is connected to the second convolution layer, using average pooling, wherein the size of the pooling kernel is 3×3; the first fully connected layer is connected to the pooling layer, wherein the first fully connected layer contains 8 neurons; the second fully connected layer contains 16 neurons; that is, the first fully connected layer and the second fully connected layer contain a total of 512 fully connected layers; the activation layer is connected to the second fully connected layer, wherein the activation layer uses the softmax function, and the activation layer calculates the probability of each transmission line detection classification according to the feature vector output by the fully connected layer; The output layer transforms the probability output by the activation layer to obtain and output the transmission line detection results.

9. The power transmission detection system based on AI video analysis according to claim 3 is characterized in that: The background management module includes an access unit, a prompt unit and a scheduling unit; The access unit is used to access the cloud processing module to obtain the real-time transmission line model and display it; The prompt unit is used to send a corresponding prompt message when the transmission line detection result in the transmission line model is abnormal; The dispatching unit is used to generate on-site maintenance dispatching information according to the abnormal maintenance results when the transmission line detection results are abnormal, and send the on-site maintenance dispatching information to the corresponding inspection terminal.

10. The power transmission detection system based on AI video analysis according to claim 9, characterized in that: The background management module also includes a calling unit; The retrieval unit is used to obtain and play historical transmission line image data from the database of the cloud processing module, so that experts can perform manual inspection based on the played transmission line image data to obtain corresponding transmission line manual inspection results.

Images