CN114612855A - Power line hazard source detection system and method fusing residual error and multi-scale network - Google Patents
Power line hazard source detection system and method fusing residual error and multi-scale network Download PDFInfo
- Publication number
- CN114612855A CN114612855A CN202210128670.6A CN202210128670A CN114612855A CN 114612855 A CN114612855 A CN 114612855A CN 202210128670 A CN202210128670 A CN 202210128670A CN 114612855 A CN114612855 A CN 114612855A
- Authority
- CN
- China
- Prior art keywords
- module
- tower crane
- network
- identification model
- rpn
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 23
- 238000000034 method Methods 0.000 title abstract description 9
- 238000012549 training Methods 0.000 claims abstract description 22
- 230000005540 biological transmission Effects 0.000 claims abstract description 18
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 11
- 238000000605 extraction Methods 0.000 claims description 7
- 230000004927 fusion Effects 0.000 claims description 3
- 238000003062 neural network model Methods 0.000 claims description 3
- 238000004519 manufacturing process Methods 0.000 abstract description 3
- 238000005516 engineering process Methods 0.000 description 8
- 238000007689 inspection Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 238000012544 monitoring process Methods 0.000 description 4
- 230000007547 defect Effects 0.000 description 3
- 238000012423 maintenance Methods 0.000 description 3
- 208000037170 Delayed Emergence from Anesthesia Diseases 0.000 description 1
- 208000025274 Lightning injury Diseases 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 230000008014 freezing Effects 0.000 description 1
- 238000007710 freezing Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 239000002689 soil Substances 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computational Linguistics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a power line hazard source detection system and method integrating residual errors and a multi-scale network, and the system comprises an image acquisition module and a tower crane identification model, wherein the image acquisition module is connected with a control module and is used for acquiring video images in a power transmission line hazard source range and outputting the video images to the control module; the tower crane identification model is used for identifying whether a tower crane enters the power transmission line danger source range or not; and training a convolutional neural network by using a plurality of pre-collected pictures containing the tower crane in the range of the dangerous source to obtain the tower crane identification model. The user can quickly identify the tower crane in the range of the hazard source by using the convolutional neural network, thereby avoiding the generation of safety accidents and preventing the influence of unexpected power failure on the normal production and life order of enterprises and common people.
Description
Technical Field
The invention relates to the technical field of computer vision, in particular to a power line hazard source detection system and method fusing residual error and a multi-scale network.
Background
The transmission line is exposed in an outdoor environment for a long time, not only bears normal mechanical load and current impact, but also is inevitably subjected to various external damages, such as strong wind, freezing, lightning stroke, sandy soil, flood, insolation, birds and beasts and the like in a natural environment if the environment is severe and the places where people are rare; in the population residence, the living space is vulnerable to both natural and artificial damages.
The safe operation of the power transmission channel is the basis for ensuring the safe and stable operation of the power transmission line, and in recent years, the trip-out rate of the power transmission line caused by external factors such as mechanical construction and the like in the power transmission channel accounts for the first of various trips. In addition, with economic development, high-speed railways, expressways and high-voltage-level lines in line channels are more and more, and because the operation and maintenance of the lines are not in place, the accidents of major social influences are more and more, and the traditional operation and maintenance mode is difficult to effectively control.
At present, the technical means of power transmission channel inspection mainly comprise a helicopter inspection technology, an unmanned aerial vehicle inspection technology, a laser scanning technology and an online monitoring technology, and the method is applied to the operation and maintenance of a power transmission line to a certain extent, but the single technical means is difficult to realize timely identification and continuous tracking of a dangerous source, and the following problems exist.
The helicopter inspection technology, the unmanned aerial vehicle inspection technology and the laser scanning technology can effectively discover line body defects and channel defects, are limited by inspection frequency, cannot guarantee the timeliness of discovering the defects, and have long monitoring period and high investment cost.
Secondly, the online monitoring technology carries out dynamic monitoring and diagnosis on the line through a sensor, has certain capability of predicting equipment faults, cannot acquire measurement of distances of static and dynamic targets in a channel, and cannot accurately identify and dynamically track a hazard source.
Disclosure of Invention
In order to solve the problems, the invention provides a power line hazard source detection system and method fusing residual errors and a multi-scale network, a user can quickly identify a tower crane in a hazard source range by using a convolutional neural network, so that safety accidents are avoided, and the influence of unexpected power failure on normal production and living order of enterprises and common people is prevented.
In order to achieve the above purpose, the invention adopts a technical scheme that:
the power line dangerous source detection system fusing the residual error and the multi-scale network comprises an image acquisition module and a tower crane identification model, wherein the image acquisition module is connected with a control module and is used for acquiring video images within the dangerous source range of the power transmission line and outputting the video images to the control module; the tower crane identification model is used for identifying whether a tower crane enters the power transmission line danger source range or not; and training a convolutional neural network by using a plurality of pre-collected pictures containing the tower crane in the range of the dangerous source to obtain the tower crane identification model.
Furthermore, the image acquisition module is a monocular camera, and the monocular camera is arranged in the dangerous source range area.
Further, the convolutional neural network comprises a backhaul Module, an RPN Module, a Module1 Module and a Module2 Module, a picture containing a tower crane is output to the backhaul Module and the RPN Module, the backhaul Module is a shared feature extraction convolutional network, the RPN Module is an area generation network, the RPN Module is used for distinguishing a foreground target and a background target, the backhaul Module is connected with the RPN Module, the backhaul Module is output to the Module1 Module, the RPN Module is output to the Module2 Module, the Module1 Module is a loss function for target positioning and classification of a prediction frame generated based on the same RPN, and the Module2 Module generates different positioning and classification loss functions for prediction frames based on the RPN.
Further, the loss function in the Module1 is defined as follows:
wherein H1(·)＝{f(·)，C(·)}，H2(. f), R (f) and f (f) are common features extracted by the Backbone module, C (g) and R (g) are functions for classifying and positioning the features respectively, P represents a prediction frame output from the Backbone module and the RPN module together, and y and P are values
Respectively representing the true category and location coordinates of the object.
Further, the loss function in the Module2 is defined as follows:
wherein P iscAnd PrThe prediction frames obtained by the Module2 Module through different 5-layer full connections are used for classification and positioning respectively.
The invention also provides a detection method of the power line hazard source detection system based on any one of the fusion residual errors and the multi-scale network, which comprises the following steps: s10, acquiring pictures, and acquiring real-time video data in the range of the power transmission line hazard source through a monocular camera; s20, training a tower crane identification model, manually collecting pictures containing tower cranes in a plurality of danger source ranges as training pictures, and training a neural network model by adopting the training pictures to obtain the tower crane identification model; s30, the real-time video data are output to the tower crane identification model after being processed by the control module, the tower crane identification model is used for identifying whether the real-time video data contain the dangerous source tower crane or not, and identification results are output to the control module.
Further, the step of S20 includes: s21, manually collecting pictures containing tower cranes in a plurality of danger source ranges as training pictures; s22, constructing a convolution application network; s23 constructing a WBlock module; s24, constructing a residual error network module Rblock, and when the input of the residual error network module Rblock is x, the output is F (x)) + x, wherein weight layer is the WBlock module; s25, constructing a Backbone feature extraction model; s26, model training is carried out by adopting a gradient descent algorithm to obtain a tower crane identification model, the tower crane identification model is deployed to a rear-end server, and a module1 is taken out of a detection network during deployment.
Compared with the prior art, the technical scheme of the invention has the following advantages:
according to the power line hazard source detection system and method integrating the residual error and the multi-scale network, a user can rapidly identify the tower crane in the hazard source range by using the convolutional neural network, so that safety accidents are avoided, and the influence of unexpected power failure on the normal production and life order of enterprises and common people is prevented.
Drawings
The technical solution and the advantages of the present invention will be apparent from the following detailed description of the embodiments of the present invention with reference to the accompanying drawings.
FIG. 1 is a system block diagram of an embodiment of the present invention;
FIG. 2 is a block diagram of a convolutional neural network in accordance with an embodiment of the present invention;
FIG. 3 is a block diagram of a WBlock module according to an embodiment of the invention;
fig. 4 is a structural diagram of a residual block Rblock according to an embodiment of the present invention;
fig. 5 is a structural diagram of a backhaul feature extraction model according to an embodiment of the present invention;
FIG. 6 is a flow chart of a method according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The embodiment provides a power line hazard source detection system integrating residual errors and a multi-scale network, as shown in fig. 1, the system comprises an image acquisition module connected with a control module and a tower crane identification model.
The image acquisition module is used for acquiring video images within the range of the power transmission line dangerous source and outputting the video images to the control module. The tower crane identification model is used for identifying whether a tower crane enters the power transmission line dangerous source range or not. And training a convolutional neural network by using a plurality of pre-collected pictures containing the tower crane in the range of the dangerous source to obtain the tower crane identification model. The image acquisition module is a monocular camera, and the monocular camera is arranged in the dangerous source range area.
As shown in fig. 2, the convolutional neural network includes a backhaul Module, an RPN Module, a Module1 Module, and a Module2 Module, a picture including a tower crane is output to the backhaul Module and the RPN Module, the backhaul Module is a shared feature extraction convolutional network, the RPN Module is an area generation network, the RPN Module is used for distinguishing a foreground target and a background target, the backhaul Module is connected to the RPN Module, the backhaul Module is output to the Module1 Module, the RPN Module is output to the Module2 Module, the Module1 Module is a loss function for target positioning and classification of a prediction frame generated based on the same RPN, and the Module2 Module generates different positioning and classification loss functions for prediction frames based on the RPN.
The loss function in the Module1 Module is defined as follows:
wherein H1(·)＝{f(·)，C(·)}，H2(. f), R (f) and f (f) are common features extracted by the Backbone module, C (g) and R (g) are functions for classifying and positioning the features respectively, P represents a prediction frame output from the Backbone module and the RPN module together, and y and P are values
Respectively representing the true category and location coordinates of the object.
The loss function in the Module2 Module is defined as follows:
wherein P iscAnd PrThe prediction frames obtained by the Module2 Module P via different 5-layer full connections are used for classification and positioning respectively.
As shown in fig. 3, the invention further provides a detection method of the power line hazard source detection system based on the above fusion residual and multi-scale network, which includes the following steps: and S10, acquiring pictures, and acquiring real-time video data in the range of the power transmission line hazard source through a monocular camera. S20 training the tower crane identification model, manually collecting pictures containing the tower crane in a plurality of danger source ranges as training pictures, and acquiring the tower crane identification model by training the neural network model through the training pictures. S30, the real-time video data are output to the tower crane identification model after being processed by the control module, the tower crane identification model is used for identifying whether the real-time video data contain the dangerous source tower crane or not, and identification results are output to the control module.
As shown in fig. 4 to 6, the step S20 includes: s21, manually collecting pictures containing tower cranes in a plurality of danger source ranges as training pictures; s22, constructing a convolution application network; s23 constructing a WBlock module; s24, constructing a residual error network module Rblock, and when the input of the residual error network module Rblock is x, the output is F (x)) + x, wherein weight layer is the WBlock module; s25, constructing a Backbone feature extraction model; s26, model training is carried out by adopting a gradient descent algorithm to obtain a tower crane identification model, the tower crane identification model is deployed to a rear-end server, and a module1 is taken out of a detection network during deployment.
The above description is only an exemplary embodiment of the present invention, and not intended to limit the scope of the present invention, and all equivalent structures or equivalent processes that are transformed by the content of the present specification and the attached drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.
Claims (7)
1. The power line hazard source detection system integrating residual errors and a multi-scale network is characterized by comprising an image acquisition module and a tower crane identification model which are connected with a control module,
the image acquisition module is used for acquiring video images within the dangerous source range of the power transmission line and outputting the video images to the control module;
the tower crane identification model is used for identifying whether a tower crane enters the power transmission line danger source range or not;
and training a convolutional neural network by using a plurality of pre-collected pictures containing the tower crane in the range of the dangerous source to obtain the tower crane identification model.
2. The residual error and multi-scale network fused power line hazard detection system of claim 1, wherein the image acquisition module is a monocular camera, and the monocular camera is disposed in the hazard range area.
3. The residual error and multi-scale network fused power line hazard detection system of claim 1, wherein the convolutional neural network comprises a backhaul Module, an RPN Module, a Module1 Module and a Module2 Module, a picture containing a tower crane is output to the backhaul Module and the RPN Module, the backhaul Module is a shared feature extraction convolutional network, the RPN Module is an area generation network, the RPN Module is used for distinguishing a foreground target from a background target, the backhaul Module is connected with the RPN Module, the backhaul Module is output to the Module1 Module, the RPN Module is output to the Module2 Module, the Module1 Module is a loss function for target positioning and classification based on a prediction frame generated by the same RPN, and the Module2 Module generates different positioning and classification loss functions for a prediction frame based on RPN.
4. The residual error and multi-scale network fused power line hazard detection system of claim 3, wherein the loss function in the Module1 Module is defined as follows:
wherein H1(·)＝{f(·),C(·)},Hz(f), R (f) is a common feature extracted by the Backbone module, C (g) and R (g) are functions for classifying and positioning features respectively, P represents a prediction frame output from the Backbone module and the RPN module together, and y and R are values
Respectively representing the true category and location coordinates of the object.
5. The residual error and multi-scale network fused power line hazard detection system of claim 3, wherein the loss function in the Module2 Module is defined as follows:
wherein P iscAnd PrThe prediction frames obtained by the Module2 Module P via different 5-layer full connections are used for classification and positioning respectively.
6. The detection method of the power line hazard source detection system based on the fusion residual error and multi-scale network of any one of claims 1 to 5 is characterized by comprising the following steps:
s10, acquiring pictures, and acquiring real-time video data in the range of the power transmission line hazard source through a monocular camera;
s20, training a tower crane identification model, manually collecting pictures containing tower cranes in a plurality of danger source ranges as training pictures, and training a neural network model by adopting the training pictures to obtain the tower crane identification model;
s30, the real-time video data are output to the tower crane identification model after being processed by the control module, the tower crane identification model is used for identifying whether the real-time video data contain the dangerous source tower crane or not, and identification results are output to the control module.
7. The residual error and multi-scale network fused power line hazard detection method according to claim 6, wherein the step of S20 comprises:
s21, manually collecting pictures containing tower cranes in a plurality of danger source ranges as training pictures;
s22, constructing a convolution application network;
s23 constructing a WBlock module;
s24, constructing a residual error network module Rblock, and when the input of the residual error network module Rblock is x, the output is F (x)) + x, wherein weight layer is the WBlock module;
s25, constructing a Backbone feature extraction model;
s26, model training is carried out by adopting a gradient descent algorithm, a tower crane identification model is obtained and is deployed to a back-end server, and when the model is deployed, a detection network takes away a module 1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210128670.6A CN114612855A (en) | 2022-02-11 | 2022-02-11 | Power line hazard source detection system and method fusing residual error and multi-scale network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210128670.6A CN114612855A (en) | 2022-02-11 | 2022-02-11 | Power line hazard source detection system and method fusing residual error and multi-scale network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114612855A true CN114612855A (en) | 2022-06-10 |
Family
ID=81859074
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210128670.6A Pending CN114612855A (en) | 2022-02-11 | 2022-02-11 | Power line hazard source detection system and method fusing residual error and multi-scale network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114612855A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117268344A (en) * | 2023-11-17 | 2023-12-22 | 航天宏图信息技术股份有限公司 | Method, device, equipment and medium for predicting high-risk source of electric tower line tree |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112232133A (en) * | 2020-09-18 | 2021-01-15 | 许继集团有限公司 | Power transmission line image identification method and device based on deep convolutional neural network |
-
2022
- 2022-02-11 CN CN202210128670.6A patent/CN114612855A/en active Pending
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112232133A (en) * | 2020-09-18 | 2021-01-15 | 许继集团有限公司 | Power transmission line image identification method and device based on deep convolutional neural network |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117268344A (en) * | 2023-11-17 | 2023-12-22 | 航天宏图信息技术股份有限公司 | Method, device, equipment and medium for predicting high-risk source of electric tower line tree |
| CN117268344B (en) * | 2023-11-17 | 2024-02-13 | 航天宏图信息技术股份有限公司 | Method, device, equipment and medium for predicting high-risk source of electric tower line tree |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108109385B (en) | System and method for identifying and judging dangerous behaviors of power transmission line anti-external damage vehicle | |
| CN111537515A (en) | Iron tower bolt defect display method and system based on three-dimensional live-action model | |
| CN108957240A (en) | Electric network fault is remotely located method and system | |
| CN108549862A (en) | Abnormal scene detection method and device | |
| CN113947555A (en) | Infrared and visible light fused visual system and method based on deep neural network | |
| CN111768417B (en) | Railway wagon overrun detection method based on monocular vision 3D reconstruction technology | |
| CN115620239B (en) | Point cloud and video combined power transmission line online monitoring method and system | |
| CN113589837A (en) | Electric power real-time inspection method based on edge cloud | |
| CN113252701A (en) | Cloud edge cooperation-based power transmission line insulator self-explosion defect detection system and method | |
| CN115909092A (en) | Light-weight power transmission channel hidden danger distance measuring method and hidden danger early warning device | |
| CN114612855A (en) | Power line hazard source detection system and method fusing residual error and multi-scale network | |
| CN115082813A (en) | Detection method, unmanned aerial vehicle, detection system and medium | |
| CN116740833A (en) | Line inspection and card punching method based on unmanned aerial vehicle | |
| CN115690730A (en) | High-speed rail contact net foreign matter detection method and system based on single classification and abnormal generation | |
| CN115546664A (en) | Cascaded network-based insulator self-explosion detection method and system | |
| CN113095160B (en) | Power system personnel safety behavior identification method and system based on artificial intelligence and 5G | |
| CN106686354A (en) | Drone patrol measuring system and measuring method thereof | |
| Zheng et al. | Forest farm fire drone monitoring system based on deep learning and unmanned aerial vehicle imagery | |
| CN113723701A (en) | Forest fire monitoring and predicting method and system, electronic equipment and storage medium | |
| CN118038153A (en) | Method, device, equipment and medium for identifying external damage prevention of distribution overhead line | |
| CN116866520B (en) | AI-based monorail crane safe operation real-time monitoring management system | |
| CN116297472A (en) | Unmanned aerial vehicle bridge crack detection method and system based on deep learning | |
| CN105551017A (en) | Transmission line forest fire target extraction method on the basis of spatio-temporal union | |
| Rong et al. | A joint faster RCNN and stereovision algorithm for vegetation encroachment detection in power line corridors | |
| CN212084165U (en) | Equipment for detecting cracks of asphalt road pavement based on 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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20220610 |