CN116363818A - Power transmission line external damage prevention monitoring system and method based on deep learning - Google Patents
Power transmission line external damage prevention monitoring system and method based on deep learning Download PDFInfo
- Publication number
- CN116363818A CN116363818A CN202310246590.5A CN202310246590A CN116363818A CN 116363818 A CN116363818 A CN 116363818A CN 202310246590 A CN202310246590 A CN 202310246590A CN 116363818 A CN116363818 A CN 116363818A
- Authority
- CN
- China
- Prior art keywords
- transmission line
- image
- alarm
- power transmission
- external damage
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B13/00—Burglar, theft or intruder alarms
- G08B13/18—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength
- G08B13/189—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems
- G08B13/194—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems using image scanning and comparing systems
- G08B13/196—Actuation by interference with heat, light, or radiation of shorter wavelength; Actuation by intruding sources of heat, light, or radiation of shorter wavelength using passive radiation detection systems using image scanning and comparing systems using television cameras
- G08B13/19602—Image analysis to detect motion of the intruder, e.g. by frame subtraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/766—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using regression, e.g. by projecting features on hyperplanes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B21/00—Alarms responsive to a single specified undesired or abnormal condition and not otherwise provided for
- G08B21/18—Status alarms
- G08B21/182—Level alarms, e.g. alarms responsive to variables exceeding a threshold
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B7/00—Signalling systems according to more than one of groups G08B3/00 - G08B6/00; Personal calling systems according to more than one of groups G08B3/00 - G08B6/00
- G08B7/06—Signalling systems according to more than one of groups G08B3/00 - G08B6/00; Personal calling systems according to more than one of groups G08B3/00 - G08B6/00 using electric transmission, e.g. involving audible and visible signalling through the use of sound and light sources
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Emergency Management (AREA)
- Business, Economics & Management (AREA)
- Alarm Systems (AREA)
Abstract
A deep learning-based transmission line external damage prevention monitoring system and method relate to the technical field of transmission line monitoring, and the technical scheme includes: mounting power transmission line monitoring equipment, collecting image information in the range of the power transmission line, and capturing image content features; determining an application scene, analyzing the application scene type through the characteristics of the image content; misjudgment analysis, namely comparing and analyzing with historical information of a database based on image features and application scenes; determining an alarm level, and determining the alarm level based on a false judgment analysis result and an image re-acquisition process; uploading the alarm signal to a client and on-site staff, and judging an uploading path of the alarm signal based on the alarm grade. The invention saves the labor cost of the power transmission line external damage inspection, improves the safety degree of the power transmission line for preventing the external damage, and the training model established by deep learning can improve the accuracy of the risk judgment of the external damage, reduce the misjudgment and the recognition error and improve the applicability of the whole system.
Description
Technical Field
The invention relates to the technical field of power transmission line monitoring, in particular to a power transmission line external damage prevention monitoring system and method based on deep learning.
Background
The safe and stable operation of the power transmission line has great significance for daily production and life power supply of people, and the large and small power transmission lines of the whole country span the big river, the south and the north, the mountain river and the urban road, often suffer from external damage and have serious potential safety hazard. At present, most circuits are inspected by adopting traditional manual inspection, so that hidden danger cannot be monitored and removed in time, and the cost is high and the efficiency is low. For example, the construction machine spans across the urban road construction site and has different damage dangerous objects at any time, such as a kite, a dust-proof film and other conducting wire foreign matters, a beam crane, a steam crane and other construction machines. External hidden troubles such as common mountain fire, bird nest and the like in the mountain and wild zone are difficult to be found by inspection staff in the first time. Once the interference damage is caused, the power is cut off, and the economic loss is immeasurable.
With the development of technology, some monitoring devices are deployed on the power transmission line, and an attempt is made to replace manual inspection. The common practice is to use the camera to monitor on line, and the staff is required to monitor the video recording screen for a long time to check whether the dangerous object approaches. However, the efforts of the personnel are limited after all, and the personnel cannot concentrate on the monitoring picture all the time. If the hidden trouble picture can not be found in time, the lead is destroyed by interference, and the alarm timeliness is not high. Most of the hardware devices of the monitoring systems rely on wired power supply, and the hardware devices are generally placed on the ground or a tower, so that the view of a camera and the monitoring range are limited. Or powered by solar energy, however, the cruising may be affected by weather.
The invention patent CN108460939A of China proposes an intelligent warning ball for preventing an electric transmission line from being broken, which comprises a ball body, a mounting device, an electromagnetic induction electricity taking device, an interface, a voice alarm, an LED lamp and a double camera, wherein the mounting device is arranged in the middle of the top end of the ball body, the electromagnetic induction electricity taking device is arranged in the mounting device, the voice alarm is arranged in the middle of the bottom end of the ball body, and the LED lamp is arranged on the outer diameter of the bottom end of the ball body. Although the invention can be installed on a power transmission line to remove broken objects for shooting monitoring, based on an installation mode and a self structure, the monitoring range of the shot image is smaller than actual requirements, and the invention can only react to conventional images and cannot adapt to complex broken object environments.
The Chinese patent No. CN111881760A discloses an identification method for preventing external damage of a power transmission line and a terminal method thereof, comprising the following steps: establishing an external broken sample library of the power transmission line, and formulating an evaluation standard; establishing an external damage detector; training the outer break detector by using a coco data set until convergence to obtain a pre-training model; migrating the pre-training model to the external broken sample library for training to obtain a training model; evaluating the training model according to the evaluation standard to obtain an optimal model; and identifying the image to be identified by using the optimal model. However, the training model method and the image characteristic acquisition equipment adopted by the method are inaccurate in measurement, the established Gaussian model and the measurement offset method do not exclude interference factors in the high-altitude environment, and the measurement effect is poor.
Disclosure of Invention
The invention aims to provide a deep learning-based power transmission line external damage prevention monitoring system and a deep learning-based power transmission line external damage prevention monitoring method, which are used for monitoring external damage of a power transmission line in real time, and establishing a system for identifying, monitoring and alarming conventional and non-conventional external damage objects through deep learning and model training, so that the identification monitoring capability and accuracy of the system are improved, the labor consumption is reduced, and the problems in the background technology are solved.
In order to achieve the above purpose, the invention provides a deep learning-based transmission line external damage prevention monitoring method, which comprises the following steps:
mounting power transmission line monitoring equipment, collecting image information in the range of the power transmission line, and capturing image content features;
determining an application scene, analyzing the application scene type through the characteristics of the image content;
misjudgment analysis, namely comparing and analyzing with historical information of a database based on image features and application scenes;
determining an alarm level, and determining the alarm level based on a false judgment analysis result and an image re-acquisition process;
uploading the alarm signal to a client and on-site staff, and judging an uploading path of the alarm signal based on the alarm grade.
Further, the method for collecting the image information comprises the following steps:
shooting images and videos of the external environment of the power transmission line through a binocular camera;
the brightness, resolution, angle, color difference and other factors of the image are regulated to make the characteristic contrast in the image obvious;
obtaining feature recognition based on image segmentation calculation of a convolution network;
and carrying out segmentation frame selection on different image features, and grabbing distinguishing features.
Further, the image segmentation step based on the convolution network comprises the following steps:
compressing the processed image, dividing the processed image into equally divided grids, and distributing each grid to image sample characteristics to be predicted according to a threshold value determined by the IOU of the group Truth;
each grid predicts a conditional probability value for each feature class, and generates B boxes on the basis of the grid, each box predicts five regression values, four characterization positions, and the fifth characterization of the probability that the box contains an object and the accuracy of the position are represented by the IOU; the score was calculated as follows at the time of testing:
the first term on the left side of the equation is predicted by a grid, the second two terms are predicted by each box, and the score of each box containing objects of different categories is obtained in a conditional probability mode; thus, the number of predicted values commonly output by the convolutional network is sxs× (b× 5+C), where S is the number of meshes, B is the number of boxes generated for each mesh, and C is the number of categories;
finally, NMS filtering is used to obtain the final prediction frame.
Further, the determining step of the application scene type includes:
determining image features, wherein the image features comprise the shape, color, distinguishing features, moving speed and contrast height of the features to judge the specific types of the features;
judging the current environment of the external object of the power transmission line and the risk of the matching environment on the monitoring equipment based on the relative position of the position distribution of the feature object in the image and the power transmission line;
and transmitting the type conclusion of the analysis application scene to a database, comparing the type conclusion with historical data of the database, and establishing an external damage environment model of the power transmission line after multiple times of comparison and elimination of misjudgment analysis by combining different images.
Further, the process of determining the alarm level includes the following steps:
the images acquired at the same position have the same characteristics, the image acquisition is triggered for a plurality of times, the risk of external damage is determined according to the occurrence frequency of the characteristics, and an alarm mechanism is triggered;
for the image characteristic analysis result, defining a non-monitoring area in the image, and triggering an alarm mechanism only for broken objects inside and outside the monitoring area in the image;
the IOU overlap of the same physical position is more than 80%, only 3 alarms are allowed within 1 day, the time interval between each alarm can be set, and more than 3 alarms are ignored;
establishing an alarm blacklist mechanism, wherein targets in the blacklist do not trigger an alarm, and abnormal characteristic targets which occur at the same position for a period of time are added into the blacklist; the similarity measurement standard IOU is excessively large in overlapping degree, is regarded as the same target, and is added into a blacklist; the blacklist is provided with an elimination mechanism, abnormal target identification is not generated for a period of time, and the abnormal target identification is eliminated from the blacklist.
Further, the step of determining the uploading path of the alarm signal based on the alarm level includes:
screening and marking the alarm information, and directly uploading a signal conforming to the alarm level to a field staff client for receiving; and manually checking and recording the signals which do not accord with the alarm level to a database as the optimization information of the monitoring model.
The technical scheme of the invention also provides a transmission line external damage prevention monitoring system based on deep learning, which comprises the following steps:
the power transmission line monitoring module is used for acquiring images and videos of the power transmission line external damage environment, performing feature frame selection on the acquired images and analyzing the application scene type;
the model building module is used for building a power transmission line external damage environment model based on the collected image characteristics and the application scene types, and carrying out characteristic model training upgrading on the collected images;
the database module is used for recording the comparison between the historical acquired image data and the image model characteristics and sending out the comparison result in an alarm mode;
the server module is used for receiving the image to perform outward breaking risk analysis, forming alarm grade and issuing instruction regulation and control, and is in communication connection with the power transmission line monitoring module, electrically connected with the model building module and electrically connected with the database module;
and the alarm module is used for displaying the monitoring scene picture, viewing the report and the history record, sending alarm information to the user terminal based on the instruction of the server module, and communicating with the server module.
Further, the power transmission line monitoring module comprises a monitor mounted on the power transmission line, the monitor comprises a left box assembly and a right box assembly which are clamped on the power transmission line in a symmetrical hinged mode, a GPRS communication module and a three-gesture coordinate sensor are mounted in the left box assembly, the power transmission line monitoring module further comprises a processor for analyzing image features, gaps are formed in the middle of the left box assembly and the middle of the right box assembly, the power transmission line penetrates through the gaps, an induction power taking sensor and a current measuring sensor are arranged on the gaps, and high-definition binocular cameras are arranged on the sides of the left box assembly and the right box assembly.
Further, the monitor is discoid, installs microphone, loudspeaker and LED lamp on it for supplementary binocular camera carries out image acquisition, and sends out the alarm.
The beneficial effects of the invention are as follows: the monitoring instrument mounted on the power transmission line is used for collecting the external damage image of the power transmission line, identifying the image characteristics and establishing an external damage object training model to judge the external damage risk level of the power transmission line, sending alarm information to staff, saving the labor cost of external damage inspection of the power transmission line, improving the safety degree of external damage prevention of the power transmission line, improving the accuracy of external damage object risk judgment by the training model established by deep learning, reducing misjudgment and identification errors and improving the applicability of the whole system.
Drawings
Fig. 1 is a flowchart of a method for monitoring an electric transmission line for preventing an external damage.
Fig. 2 is a composition diagram of the transmission line anti-external damage monitoring system of the present invention.
Fig. 3 is a schematic structural diagram of a monitor according to an embodiment of the present invention.
The three-dimensional coordinate sensor comprises a 1-left box component, a 2-right box component, a 3-binocular camera, a 4-three-dimensional coordinate sensor, a 5-notch, a 6-camera, a 7-induction pickup sensor, an 8-loudspeaker, a 9-microphone, a 10-LED lamp and an 11-current measurement sensor.
Detailed Description
In order to make the technical problems, technical schemes and beneficial effects to be solved more clear, the invention is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
Examples
Referring to a flow chart of a deep learning-based transmission line external damage prevention monitoring method in fig. 1, the invention monitors external damage of a transmission line in real time, establishes a system for identifying, monitoring and alarming conventional and non-conventional external damage objects through deep learning and model training, improves the identification monitoring capability and accuracy of the system, and reduces the labor consumption, and the invention provides the deep learning-based transmission line external damage prevention monitoring method, which comprises the following technical scheme:
s1, mounting power transmission line monitoring equipment, collecting image information in a power transmission line range, and capturing image content features;
s2, determining an application scene, analyzing the type of the application scene through the characteristics of the content of the image;
s3, misjudgment analysis is carried out, and comparison analysis is carried out on the image features and the application scene and the historical information of the database;
s4, determining an alarm level, and determining the alarm level based on a misjudgment analysis result and an image re-acquisition process;
and S5, uploading the alarm signal to the client and the field staff, and judging an uploading path of the alarm signal based on the alarm grade.
Referring to the structural schematic diagram of the monitor in fig. 3, the device for acquiring image information acquires images and videos of an external damage environment of a power transmission line for the monitor mounted on the power transmission line, and the monitor includes: the monitor comprises a left box assembly 1 and a right box assembly 2 which are clamped on a power transmission line in a symmetrical hinged mode, wherein a GPRS communication module and a three-gesture coordinate sensor 4 are installed in the left box assembly 1, the monitor further comprises a processor for analyzing image features, a notch 5 is formed in the middle of the left box assembly 1 and the middle of the right box assembly 2, the power transmission line penetrates through the notch 5, an induction pickup sensor 7 and a current measurement sensor 11 are arranged on the notch 5, a high-definition binocular camera 3 is arranged on the side faces of the left box assembly 1 and the right box assembly 2, the monitor is disc-shaped, and a microphone 9, a loudspeaker 8 and an LED lamp 10 are installed on the monitor and used for assisting the binocular camera in image acquisition and giving out an alarm.
The monitoring instrument is mounted on the power transmission line, the state of the power transmission line is obtained in real time, and the binocular camera can shoot on-site photos and videos and communicate with background service through a mobile network. The device can deploy a lightweight operating system, a target detection service and a model, perform preliminary outward-breaking risk analysis on a shot picture, and also can send a photo to a background, and perform outward-breaking risk analysis by a server. The hardware configuration and the calculation power of the server are stronger, and a target detection model with higher precision can be carried. The service can be carried out by determining that the target detection service is placed in the equipment or the service end according to the actual situation, or can be carried out on both sides.
The method for acquiring the image comprises the following steps:
shooting images and videos of the external environment of the power transmission line through a binocular camera;
the brightness, resolution, angle, color difference and other factors of the image are regulated to make the characteristic contrast in the image obvious;
obtaining feature recognition based on image segmentation calculation of a convolution network;
and carrying out segmentation frame selection on different image features, and grabbing distinguishing features.
Further, the image segmentation step based on the convolution network comprises the following steps:
compressing the processed image, dividing the processed image into equally divided grids, and distributing each grid to image sample characteristics to be predicted according to a threshold value determined by the IOU of the group Truth;
each grid predicts a conditional probability value for each feature class, and generates B boxes on the basis of the grid, each box predicts five regression values, four characterization positions, and the fifth characterization of the probability that the box contains an object and the accuracy of the position are represented by the IOU; the score was calculated as follows at the time of testing:
the first term on the left side of the equation is predicted by a grid, the second two terms are predicted by each box, and the score of each box containing objects of different categories is obtained in a conditional probability mode; thus, the number of predicted values commonly output by the convolutional network is sxs× (b× 5+C), where S is the number of meshes, B is the number of boxes generated for each mesh, and C is the number of categories;
finally, NMS filtering is used to obtain the final prediction frame.
Further, the determining step of the application scene type includes:
determining image features, wherein the image features comprise the shape, color, distinguishing features, moving speed and contrast height of the features to judge the specific types of the features;
judging the current environment of the external object of the power transmission line and the risk of the matching environment on the monitoring equipment based on the relative position of the position distribution of the feature object in the image and the power transmission line;
and transmitting the type conclusion of the analysis application scene to a database, comparing the type conclusion with historical data of the database, and establishing an external damage environment model of the power transmission line after multiple times of comparison and elimination of misjudgment analysis by combining different images.
Further, the process of determining the alarm level includes the following judgment principles and steps:
the recognition result of the single abnormal image or the single out-of-limit of the sensor data and the like are not directly pushed to the user as an alarm. In order to improve the accuracy of the alarm, false alarm is avoided to the greatest extent, insignificant alarm (such as construction machinery far away from a line) is avoided, and the identification result of a single abnormal image or single out-of-limit of sensor data and the like are not directly pushed to a user as alarm as much as possible; and the recognition result of the single image or the single out-of-limit of the sensor data is used as the trigger of the suspected alarm event, and when the recognition of the single abnormal image or the single out-of-limit trigger of the sensor data is carried out, the equipment end enters an alarm event processing flow and starts an image acquisition recognition or sensor data acquisition mode with higher frequency.
After the identification result of the single abnormal image or the single out-of-limit of the sensor data is triggered, the single abnormal image or the single out-of-limit of the sensor data is confirmed to be a harmless event through an alarm event processing flow, and the product can upload and identify the identification result of the single abnormal image or the single out-of-limit of the sensor data and can be identified as classification such as suspected, caution and the like;
the same physical location (IOU overlap is more than 80%), only 3 alarms are allowed within 1 day, and the time interval between each alarm can be set, such as 15/30/60 minutes; more than 3 alarms are ignored.
And establishing an alarm blacklist mechanism, wherein targets in the blacklist do not trigger an alarm. The abnormal targets occurring at the same position for a continuous period of time (which can be set by user definition according to specific scenes, and the following is the same) are false alarms and added into a blacklist. The similarity measurement standard IOU is excessively overlapped, is regarded as the same target, and is added into a blacklist. The blacklist is provided with a elimination mechanism, and abnormal target identification is eliminated from the blacklist after a period of time.
Further, the step of determining the uploading path of the alarm signal based on the alarm level includes:
screening and marking the alarm information, and directly uploading a signal conforming to the alarm level to a field staff client for receiving; and manually checking and recording the signals which do not accord with the alarm level to a database as the optimization information of the monitoring model.
Referring to fig. 2, the invention further provides a transmission line external damage prevention monitoring system based on deep learning, which comprises:
the power transmission line monitoring module is used for acquiring images and videos of the power transmission line external damage environment, performing feature frame selection on the acquired images and analyzing the application scene type;
the model building module is used for building a power transmission line external damage environment model based on the collected image characteristics and the application scene types, and carrying out characteristic model training upgrading on the collected images;
the database module is used for recording the comparison between the historical acquired image data and the image model characteristics and sending out the comparison result in an alarm mode;
the server module is used for receiving the image to perform outward breaking risk analysis, forming alarm grade and issuing instruction regulation and control, and is in communication connection with the power transmission line monitoring module, electrically connected with the model building module and electrically connected with the database module;
and the alarm module is used for displaying the monitoring scene picture, viewing the report and the history record, sending alarm information to the user terminal based on the instruction of the server module, and communicating with the server module.
The workflow of the system is as follows:
when the equipment camera shoots a photo, the self-integrated target detection service is called to infer the photo, the category and the position of the target are judged, and the result is stored in the database. In order to prevent misjudgment, the historical results are queried in the database, and when targets with the same category exist in the same position, the photographing frequency is accelerated. When the same target is found to exist in the same position three times continuously, the system considers that the target has an external damage risk to the power transmission line, and the alarm level analysis is carried out on the target.
After the system sends out the alarm information, the user can carry out further manual checking on the information at the client to decide to push the alarm to the field operation personnel, and the system can also be set to directly push the alarm information to the field operation personnel. And alarming for individual misjudgment, and making a corresponding mark when manually checking, and not alarming to operators. The system will record such scenarios for later use in upgrading the target detection model.
The server side is provided with a communication interface, a message queue, a stream processing component, an artificial intelligent service, a database and the like, receives and processes instructions and information pushed by the equipment, performs external damage risk analysis on photos shot by the equipment in real time, pushes alarm information as appropriate, and stores historical information such as photos, alarms and the like. The server adopts containerized deployment, can be rapidly deployed on a cloud server or a private server, and performs cluster scaling, rolling updating and the like according to the traffic.
The client side mainly comprises a WEB large screen of a computer desktop and a mobile phone APP, and can adapt to different scene requirements. The WEB large screen is suitable for complex requirements of users, such as comparing monitoring pictures of multiple time periods, checking reports, historical records and the like. The mobile phone APP is more suitable for on-site attended and operators, is convenient for receiving alarm information at the first time and takes corresponding measures.
It will be appreciated by those skilled in the art that the foregoing description of the service device is merely an example and is not meant to be limiting, and may include more or fewer components than the foregoing description, or may combine certain components, or different components, such as may include input-output devices, network access devices, buses, etc.
The processor may be a central processing unit (Central Processing Unit, CPU), other general purpose processors, digital signal processors (Digital Signal Processor, DSP), application specific integrated circuits (Application Specific Integrated Circuit, ASIC), off-the-shelf programmable gate arrays (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or the like. The general purpose processor may be a microprocessor or the processor may be any conventional processor or the like, which is the control center of the terminal device described above, and which connects the various parts of the entire user terminal using various interfaces and lines.
The foregoing description is only of the preferred embodiments of the present invention, and is not intended to limit the scope of the invention, but rather is intended to cover any equivalents of the structures or equivalent processes disclosed herein or in the alternative, which may be employed directly or indirectly in other related arts.
Claims (9)
1. The power transmission line external damage prevention monitoring method based on deep learning is characterized by comprising the following steps of:
mounting power transmission line monitoring equipment, collecting image information in the range of the power transmission line, and capturing image content features;
determining an application scene, analyzing the application scene type through the characteristics of the image content;
misjudgment analysis, namely comparing and analyzing with historical information of a database based on image features and application scenes;
determining an alarm level, and determining the alarm level based on a false judgment analysis result and an image re-acquisition process;
uploading the alarm signal to a client and on-site staff, and judging an uploading path of the alarm signal based on the alarm grade.
2. The deep learning-based transmission line damage-prevention monitoring method as claimed in claim 1, wherein the image information acquisition method comprises the following steps:
shooting images and videos of the external environment of the power transmission line through a binocular camera;
the brightness, resolution, angle, color difference and other factors of the image are regulated to make the characteristic contrast in the image obvious;
obtaining feature recognition based on image segmentation calculation of a convolution network;
and carrying out segmentation frame selection on different image features, and grabbing distinguishing features.
3. The deep learning-based transmission line damage-prevention monitoring method as claimed in claim 2, wherein the image segmentation step based on the convolutional network comprises:
compressing the processed image, dividing the processed image into equally divided grids, and distributing each grid to image sample characteristics to be predicted according to a threshold value determined by the IOU of the group Truth;
each grid predicts a conditional probability value for each feature class, and generates B boxes on the basis of the grid, each box predicts five regression values, four characterization positions, and the fifth characterization of the probability that the box contains an object and the accuracy of the position are represented by the IOU; the score was calculated as follows at the time of testing:
the first term on the left side of the equation is predicted by a grid, the second two terms are predicted by each box, and the score of each box containing objects of different categories is obtained in a conditional probability mode; thus, the number of predicted values commonly output by the convolutional network is sxs× (b× 5+C), where S is the number of meshes, B is the number of boxes generated for each mesh, and C is the number of categories;
finally, NMS filtering is used to obtain the final prediction frame.
4. The deep learning-based transmission line damage-prevention monitoring method as claimed in claim 3, wherein the determining of the application scene type includes:
determining image features, wherein the image features comprise the shape, color, distinguishing features, moving speed and contrast height of the features to judge the specific types of the features;
judging the current environment of the external object of the power transmission line and the risk of the matching environment on the monitoring equipment based on the relative position of the position distribution of the feature object in the image and the power transmission line;
and transmitting the type conclusion of the analysis application scene to a database, comparing the type conclusion with historical data of the database, and establishing an external damage environment model of the power transmission line after multiple times of comparison and elimination of misjudgment analysis by combining different images.
5. The deep learning-based transmission line damage prevention monitoring method as claimed in claim 4, wherein the process of determining the alarm level comprises the following steps:
the images acquired at the same position have the same characteristics, the image acquisition is triggered for a plurality of times, the risk of external damage is determined according to the occurrence frequency of the characteristics, and an alarm mechanism is triggered;
for the image characteristic analysis result, defining a non-monitoring area in the image, and triggering an alarm mechanism only for broken objects inside and outside the monitoring area in the image;
the IOU overlap of the same physical position is more than 80%, only 3 alarms are allowed within 1 day, the time interval between each alarm can be set, and more than 3 alarms are ignored;
establishing an alarm blacklist mechanism, wherein targets in the blacklist do not trigger an alarm, and abnormal characteristic targets which occur at the same position for a period of time are added into the blacklist; the similarity measurement standard IOU is excessively large in overlapping degree, is regarded as the same target, and is added into a blacklist; the blacklist is provided with an elimination mechanism, abnormal target identification is not generated for a period of time, and the abnormal target identification is eliminated from the blacklist.
6. The deep learning-based transmission line damage prevention monitoring method according to claim 1, wherein the step of determining an uploading path of an alarm signal based on an alarm level comprises:
screening and marking the alarm information, and directly uploading a signal conforming to the alarm level to a field staff client for receiving; and manually checking and recording the signals which do not accord with the alarm level to a database as the optimization information of the monitoring model.
7. Power transmission line external damage prevention monitoring system based on deep learning, which is characterized by comprising:
the power transmission line monitoring module is used for acquiring images and videos of the power transmission line external damage environment, performing feature frame selection on the acquired images and analyzing the application scene type;
the model building module is used for building a power transmission line external damage environment model based on the collected image characteristics and the application scene types, and carrying out characteristic model training upgrading on the collected images;
the database module is used for recording the comparison between the historical acquired image data and the image model characteristics and sending out the comparison result in an alarm mode;
the server module is used for receiving the image to perform outward breaking risk analysis, forming alarm grade and issuing instruction regulation and control, and is in communication connection with the power transmission line monitoring module, electrically connected with the model building module and electrically connected with the database module;
and the alarm module is used for displaying the monitoring scene picture, viewing the report and the history record, sending alarm information to the user terminal based on the instruction of the server module, and communicating with the server module.
8. The deep learning-based transmission line external damage prevention monitoring system according to claim 7, wherein the transmission line monitoring module comprises a monitor mounted on the transmission line, the monitor comprises a left box assembly (1) and a right box assembly (2) which are clamped on the transmission line in a symmetrical hinged manner, a GPRS communication module and a three-gesture coordinate sensor (4) are mounted in the left box assembly (1), the deep learning-based transmission line external damage prevention monitoring system further comprises a processor for analyzing image features, a notch (5) is formed in the middle of the left box assembly (1) and the right box assembly (2) so that the transmission line passes through, an induction pickup sensor (7) and a current measurement sensor (11) are arranged on the notch (5), and high-definition binocular cameras (3) are arranged on the sides of the left box assembly (1) and the right box assembly (2).
9. The deep learning-based transmission line external damage prevention monitoring system according to claim 8, wherein the monitoring instrument is disc-shaped, and is provided with a microphone (9), a loudspeaker (8) and an LED lamp (10) for assisting the binocular camera in image acquisition and giving an alarm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310246590.5A CN116363818A (en) | 2023-03-15 | 2023-03-15 | Power transmission line external damage prevention monitoring system and method based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310246590.5A CN116363818A (en) | 2023-03-15 | 2023-03-15 | Power transmission line external damage prevention monitoring system and method based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116363818A true CN116363818A (en) | 2023-06-30 |
Family
ID=86927567
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310246590.5A Pending CN116363818A (en) | 2023-03-15 | 2023-03-15 | Power transmission line external damage prevention monitoring system and method based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116363818A (en) |
-
2023
- 2023-03-15 CN CN202310246590.5A patent/CN116363818A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20200072814A1 (en) | Support Structure Inspection Devices, Systems and Methods | |
| CN102842211B (en) | Monitoring and early warning system and monitoring and early warning method for prevention of external force of transmission line based on image recognition | |
| CN106326808B (en) | Method for detecting bird nest on power transmission line tower based on unmanned aerial vehicle image | |
| CN107331097A (en) | The periphery intrusion preventing apparatus and method merged based on target position information | |
| US20230005176A1 (en) | Throwing position acquisition method and apparatus, computer device and storage medium | |
| CN104394361A (en) | Pedestrian crossing intelligent monitoring device and detection method | |
| CN111523397B (en) | Intelligent lamp post visual identification device, method and system and electronic equipment thereof | |
| CN111507235B (en) | Railway perimeter foreign matter intrusion detection method based on video | |
| CN112776856A (en) | Track foreign matter intrusion monitoring method, device and system and monitoring host equipment | |
| CN111476964A (en) | Remote forest fire prevention monitoring system and method | |
| CN110868568A (en) | Video monitoring method and device for station outside air pipe | |
| CN116416579A (en) | Intelligent analysis management system for park security monitoring storage information | |
| CN110691224A (en) | Transformer substation perimeter video intelligent detection system | |
| CN114627388B (en) | Foreign matter detection equipment and foreign matter detection method for power transmission line | |
| CN209433517U (en) | It is a kind of based on more flame images and the fire identification warning device for combining criterion | |
| CN116958764A (en) | Method for detecting data of mounting equipment based on deep learning | |
| CN116363818A (en) | Power transmission line external damage prevention monitoring system and method based on deep learning | |
| CN115393340A (en) | AI vision product quality detection system based on 5G algorithm | |
| CN113569956A (en) | Mountain fire disaster investigation and identification method based on AI algorithm | |
| CN113222548A (en) | Community property eagle eye monitoring system and use method | |
| CN111382697A (en) | Image data processing method and first electronic equipment | |
| CN113920767B (en) | Operation and maintenance alarming method, system, device and computer readable storage medium | |
| CN220543137U (en) | Radar, vibration and machine vision fusion intrusion detection host and intrusion detection system | |
| CN117978969A (en) | AI video management platform applied to aquaculture | |
| CN116434114A (en) | Forest smoke intelligent detection method and system based on deep neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |