CN116229278B - Method and system for detecting rust defect of vibration damper of power transmission line - Google Patents
Method and system for detecting rust defect of vibration damper of power transmission line Download PDFInfo
- Publication number
- CN116229278B CN116229278B CN202310518795.4A CN202310518795A CN116229278B CN 116229278 B CN116229278 B CN 116229278B CN 202310518795 A CN202310518795 A CN 202310518795A CN 116229278 B CN116229278 B CN 116229278B
- Authority
- CN
- China
- Prior art keywords
- transmission line
- damper
- data set
- image
- detection model
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
-
- 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
-
- 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
-
- 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
- G06V10/765—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects using rules for classification or partitioning the feature space
-
- 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
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06V20/70—Labelling scene content, e.g. deriving syntactic or semantic representations
-
- 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)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Medical Informatics (AREA)
- Databases & Information Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention discloses a method and a system for detecting rust defects of a transmission line damper, which relate to the technical field of defect detection of transmission line equipment, wherein an EESP module in an ESPNetv2 model is used for replacing a CBS module of a C2f module in a YOLOv8s detection model, a SiLU activation function in a convolution module which plays a role in downsampling in a main network in the YOLOv8s detection model is replaced by a PReLU activation function, a downsampling operation and 1X 1 standard convolution are used for replacing the CBS module of a head network in a YOLOv8s detection model feature pyramid, an improved YOLOv8s detection model is obtained, the model is trained, the obtained target detection model is used for detecting rust defects of the transmission line damper, and the technical effects of accurately and rapidly detecting rust defect target positions of the transmission line damper are achieved, and technical references are provided for operation and maintenance staff to develop the detection work of the transmission line damper.
Description
Technical Field
The invention relates to the technical field of defect detection of transmission line equipment, in particular to a method and a system for detecting rust defects of a damper of a transmission line.
Background
The vibration damper for the power transmission line is one of important component parts of the overhead power transmission line, can reduce vibration amplitude of the overhead conductor, avoids frequent vibration of the conductor due to wind power, further generates fatigue damage, and is beneficial to prolonging service life of the conductor. The damper is exposed outdoors for a long time, is often subjected to wind blowing and rain striking, is extremely easy to generate rust phenomenon, and is likely to generate conditions such as damage and falling off of the damper once rust occurs, so that the protection effect of the damper on a power transmission line is affected, and the reliability and the stability of the power transmission line are greatly affected. Therefore, the defect detection is carried out on the damper, the problem of rust corrosion of the damper is found in time to be one of important working contents of inspection of the transmission line, and the method is a primary task for guaranteeing safe and stable operation of the transmission line.
The method is characterized in that mass transmission line inspection images are acquired through machine vision, defect analysis on the transmission line vibration damper according to the inspection images is a main mode of rust detection on the transmission line vibration damper currently, and the mass transmission line inspection images can be analyzed by the target detection method based on the deep convolution neural network, so that defect automatic identification is realized. The current detection method for the rust defects of the anti-vibration hammers of the transmission line is more prone to improving the identification accuracy, and ignores the real-time requirement in the inspection process, so that operation and maintenance personnel cannot timely feed back and process after the trained detection network for the rust defects of the anti-vibration hammers of the transmission line detects the defects. Therefore, the novel method for detecting the rust defect of the transmission line damper is used for accurately and rapidly detecting the rust defect target position of the transmission line damper, provides technical reference for operation and maintenance personnel to develop the transmission line damper detection work, and is a technical problem to be solved urgently by the personnel in the field.
Disclosure of Invention
The invention provides a method and a system for detecting rust defects of a transmission line damper, which are used for accurately and rapidly detecting the target positions of the rust defects of the transmission line damper and provide technical references for operation and maintenance personnel to develop transmission line damper detection work.
In view of the foregoing, a first aspect of the present invention provides a method for detecting rust defects of a damper for a power transmission line, including:
constructing a transmission line anti-vibration hammer image sample data set, wherein the transmission line anti-vibration hammer image sample data set comprises a rust defect image data set and a normal image data set;
replacing a CBS module of a C2f module in the YOLOv8s detection model by using an EESP module in the ESPNetv2 model, replacing a SiLU activation function in a convolution module which plays a role in downsampling in a main network in the YOLOv8s detection model by using a PReLU activation function, and replacing a CBS module of a head network in a feature pyramid of the YOLOv8s detection model by using downsampling operation and 1X 1 standard convolution to obtain an improved YOLOv8s detection model;
training an improved YOLOv8s detection model according to the transmission line damper image sample data set to obtain a target detection model;
and inputting the transmission line image to be identified and containing the damper into a target detection model to obtain a transmission line damper rust defect detection result.
Optionally, constructing a transmission line anti-vibration hammer image sample data set includes:
acquiring an image containing a damper in a transmission line inspection image, preprocessing the image, and constructing a real transmission line damper image data set;
training a generated countermeasure network model based on a real transmission line anti-vibration hammer image data set, and generating a transmission line anti-vibration hammer image containing the anti-vibration hammer rust defect through a trained generation network in the generated countermeasure network based on an image containing the anti-vibration hammer rust defect in the real transmission line anti-vibration hammer image data set;
adding an image containing the damper rust defect in the real power transmission line damper image data set and a power transmission line damper image containing the damper rust defect generated by the generation network into the rust defect image data set, and adding an image which does not contain the damper rust defect in the real power transmission line damper image data set into the normal image data set.
Optionally, the number of parallel deep hole separable volume modules in EESP modules in the ESPNetv2 model is 4, and the hole rates are 3,5,7 and 3, respectively.
Optionally, training the improved YOLOv8s detection model according to the transmission line anti-vibration hammer image sample data set to obtain a target detection model, including:
pre-training the improved YOLOv8s detection model by using a Gogle Open Image public data set, and taking model parameters obtained by pre-training as initial parameters of the improved YOLOv8s detection model;
and training the pre-trained improved YOLOv8s detection model by using the transmission line damper image sample data set to obtain a target detection model.
Alternatively, the generation countermeasure network model is a SinGAN network model.
The second aspect of the invention provides a transmission line damper rust defect detection system, comprising:
the data set construction module is used for constructing a power transmission line anti-vibration hammer image sample data set, wherein the power transmission line anti-vibration hammer image sample data set comprises a rust defect image data set and a normal image data set;
the improved detection model construction module is used for replacing a CBS module of a C2f module in the YOLOv8s detection model by using an EESP module in the ESPNetv2 model, replacing a SiLU activation function in a convolution module which plays a role in downsampling in a main network in the YOLOv8s detection model, and replacing a CBS module of a head network in a feature pyramid of the YOLOv8s detection model by using downsampling operation and 1X 1 standard convolution to obtain an improved YOLOv8s detection model;
the detection model training module is used for training the improved YOLOv8s detection model according to the transmission line damper image sample data set to obtain a target detection model;
and the rust defect identification module is used for inputting the transmission line image to be identified and containing the damper into the target detection model to obtain a transmission line damper rust defect detection result.
Optionally, the data set construction module is specifically configured to:
acquiring an image containing a damper in a transmission line inspection image, preprocessing the image, and constructing a real transmission line damper image data set;
training a generated countermeasure network model based on a real transmission line anti-vibration hammer image data set, and generating a transmission line anti-vibration hammer image containing the anti-vibration hammer rust defect through a trained generation network in the generated countermeasure network based on an image containing the anti-vibration hammer rust defect in the real transmission line anti-vibration hammer image data set;
adding an image containing the damper rust defect in the real power transmission line damper image data set and a power transmission line damper image containing the damper rust defect generated by the generation network into the rust defect image data set, and adding an image which does not contain the damper rust defect in the real power transmission line damper image data set into the normal image data set.
Optionally, the number of parallel deep hole separable volume modules in EESP modules in the ESPNetv2 model is 4, and the hole rates are 3,5,7 and 3, respectively.
Optionally, the detection model training module is specifically configured to:
pre-training the improved YOLOv8s detection model by using a Gogle Open Image public data set, and taking model parameters obtained by pre-training as initial parameters of the improved YOLOv8s detection model;
and training the pre-trained improved YOLOv8s detection model by using the transmission line damper image sample data set to obtain a target detection model.
Alternatively, the generation countermeasure network model is a SinGAN network model.
According to the technical scheme, the method and the system for detecting the rust defect of the vibration damper of the power transmission line have the following advantages:
according to the method for detecting the rust defect of the transmission line vibration damper, EESP modules in ESPNetv2 models with smaller parameter are used for replacing CBS modules of C2f modules in YOLOv8s detection models, parameter of the models is reduced, PReLU activation functions with lower calculation complexity are used for replacing SiLU activation functions in convolution modules playing a role in downsampling in a main network in the YOLOv8s detection models, downsampling operation and 1X 1 standard convolution are used for replacing CBS modules of head networks in feature pyramids of the YOLOv8s detection models, an improved YOLOv8s detection model is obtained, an improved YOLOv8s detection model is trained by using a transmission line vibration damper image sample data set, a target detection model is obtained, and finally rust defects of the transmission line vibration damper are detected by using the target detection model, so that the technical reference for carrying out vibration damper detection of the transmission line vibration damper is achieved.
The power transmission line damper rust defect detection system provided by the invention is used for executing the power transmission line damper rust defect detection method provided by the invention, and the principle and the obtained technical effects are the same as those of the power transmission line damper rust defect detection method provided by the invention, and are not repeated herein.
Drawings
For a clearer description of embodiments of the invention or of solutions according to the prior art, the figures which are used in the description of the embodiments or of the prior art will be briefly described, it being obvious that the figures in the description below are only some embodiments of the invention, from which, without the aid of inventive efforts, other relevant figures can be obtained for a person skilled in the art.
FIG. 1 is a schematic flow chart of a method for detecting rust defects of a damper of a power transmission line;
FIG. 2 is a network architecture diagram of EESP modules in the ESPNetv2 model provided in the present invention;
FIG. 3 is a network architecture diagram of the improved YOLOv8s detection model provided in the present invention;
fig. 4 is a schematic structural diagram of the rust defect detection system for the damper of the transmission line provided by the invention.
Detailed Description
In order to make the present invention better understood by those skilled in the art, the following description will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings, and it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
For easy understanding, referring to fig. 1, an embodiment of a method for detecting rust defects of a damper of a power transmission line according to the present invention includes:
step 101, constructing a transmission line anti-vibration hammer image sample data set, wherein the transmission line anti-vibration hammer image sample data set comprises a rust defect image data set and a normal image data set.
In the embodiment of the invention, the rust defect image data set and the normal image data set are acquired first, and the vibration damper image sample data set of the power transmission line is constructed. The method comprises the steps of selecting an image containing the damper from an unmanned aerial vehicle inspection image, preprocessing the image, cutting the image into preset sizes such as 800 multiplied by 800, and classifying and labeling the rust defect image and the normal image, so that a real power transmission line damper image data set is constructed. In order to realize the comprehensiveness of the rust defect image sample, the rust defect image sample can be subjected to sample expansion, specifically, a generated countermeasure network model is trained based on a real power transmission line damper image data set, the generated countermeasure network model is a SinGAN network model, an image containing damper rust defects in the real power transmission line damper image data set is based on the trained generation network in the generated countermeasure network, and a power transmission line damper image containing damper rust defects is generated. Therefore, the image data set of the rust defect formed by the image containing the rust defect of the damper in the real power transmission line damper image data set and the power transmission line damper image containing the rust defect of the damper generated by the generation network can be obtained. And the image which does not contain the vibration damper rust defect in the real power transmission line vibration damper image data set is classified as a normal image data set.
Step 102, replacing a CBS module of a C2f module in the Yolov8s detection model by an EESP module in the ESPNetv2 model, replacing a SiLU activation function in a convolution module playing a role in downsampling in a backbone network in the Yolov8s detection model by a PReLU activation function, and replacing a CBS module of a head network in a feature pyramid of the Yolov8s detection model by downsampling operation and standard convolution of 1×1 to obtain an improved Yolov8s detection model.
It should be noted that, in the embodiment of the present invention, the conventional YOLOv8s detection model is improved, the EESP module in the ESPNetv2 model with a smaller parameter is used to replace the CBS module of the C2f module in the YOLOv8s detection model (the CBS module is composed of a 3x3 standard convolution+batch normalization+silu activation function), the prerlu activation function with a lower computational complexity is used to replace the SiLU activation function in the convolution module playing the role of downsampling in the backbone network in the YOLOv8s detection model, and the downsampling operation and a 1 x 1 standard convolution are used to replace the CBS module of the head network in the YOLOv8s detection model feature pyramid (a 3x3 standard convolution+batch normalization+silu activation function with a step length of 2), so as to obtain the improved YOLOv8s detection model.
Specifically, as shown in fig. 2, the network structure of the EESP unit module in the present invention has 4 parallel deep hole separable convolution modules, and the hole rates are 3,5,7 and 3 respectively. In fig. 2, H and W are the width and height of the input feature map, respectively, ci is the number of input channels of the input feature map, co is the number of output channels of the output feature map, s is the convolution kernel size, and r is the void fraction. The principle is as follows: firstly, processing input features by using a 1 multiplied by 1 point group convolution, adjusting the number of channels to Co/4, facilitating cross-channel feature information interaction, using a batch normalization layer to avoid network over fitting, and using a ReLU activation function to increase nonlinear expression capacity of the network. And then, extracting the features by using the depth hole separable convolutions of different receptive fields (namely four 3x3 depth hole separable convolutions with the hole rates r of 3,5,7 and 3 in fig. 2), so that the receptive fields of the network are effectively increased, and the network is facilitated to extract more effective feature information. And then combining and splicing the features extracted from different receptive fields to integrate effective feature information. And then, carrying out information fusion (the number of channels is unchanged) on the fused features by using a 1X 1 point group-by-point convolution, connecting the generated features with the input features through a shortcut, effectively utilizing the original effective feature information, and finally, increasing the nonlinear expression capacity of the module through a PReLU activation function to obtain an output feature map.
The network structure of the improved YOLOv8s detection model obtained in the invention is shown in fig. 3, a convolution module con_p_3 in the main network has the function of downsampling, the channel number is adjusted, and finally, a characteristic diagram is generated (the characteristic diagram sizes are respectively 100×100×256, 50×50×512 and 25×25×512). The c2f_1 and c2f_2 modules (in the backbone network) play a role in feature extraction, extracting valid features. The SPPF module is used for realizing fusion of local features and global features. The up-sampling module is used for expanding the input feature map by two times, and generating the feature map through two up-samplings (the feature map sizes are respectively 50×50×512 and 100×100×256). The feature map 100×100×256 generated by the upsampling operation and the feature map 100×100×256 generated by the con_p_3 module are subjected to feature fusion, and the fused feature map is input into the detection head (100×100×256). The two downsampling+1x1 standard convolution modules play a role in downsampling, the feature map 100×100×256 is adjusted to 50×50×512 after passing through one module, then the feature map 50×50×512 is adjusted to 25×25×512 after passing through the module again, and then the feature map 50×50×512 obtained after being respectively fused with the 50×50×512 feature map generated by the Con_P_3 module and the 50×50×512 feature map generated by the upsampling and the 25×25×512 feature map generated by the SPPF module are respectively input into the other two detection heads (50×50×512 and 25×25×512) through feature fusion. And finally, the detection head detects the rust defect of the vibration damper of the power transmission line according to the input characteristic diagram.
The total of 8C 2f modules in the traditional YOLOv8s detection model has 16 standard convolutions of 3 multiplied by 3, the corresponding 8C 2f modules in the improved YOLOv8s detection model has 64 depth holes with 3 multiplied by 3, the standard convolutions of 3 multiplied by 3 have the parameters of 9 multiplied by the number of output channels and 3 multiplied by 3, and the receptive field is 3 multiplied by 3; the parameter quantity of the separable convolution of each 3 multiplied by 3 depth hole is 9 multiplied by the number of input channels and the number of input channels multiplied by the number of input channels, and the receptive field is (2 multiplied by the void ratio+1) multiplied by (2 multiplied by the void ratio+1); the minimum input/output channel number of the C2f module in the Yolov8s model is 128, so that the improved model parameter is far lower than that of the traditional Yolov8s detection model; and the void ratio in the improved model is 3,5,7 and 3 respectively, so that the improved model receptive field is larger than that of the traditional YOLOv8s detection model, and the characteristics can be extracted more effectively.
And step 103, training the improved YOLOv8s detection model according to the transmission line damper image sample data set to obtain a target detection model.
The improved YOLOv8s detection model is trained by using the transmission line damper image sample data set. Specifically, the improved YOLOv8s detection model is pre-trained by using a Gogle Open Image public data set, model parameters obtained by pre-training are used as initial parameters of the improved YOLOv8s detection model, and then the pre-trained improved YOLOv8s detection model is trained by using a transmission line damper Image sample data set to obtain a target detection model. When the pre-trained improved YOLOv8s detection model is trained by using the transmission line anti-vibration hammer image sample data set, the transmission line anti-vibration hammer image sample data set can be divided into a training set and a testing set according to a preset proportion, the improved YOLOv8s detection model is trained by using the training set, and the improved YOLOv8s detection model is tested by using the testing set.
And 104, inputting the transmission line image to be identified and containing the damper into a target detection model to obtain a transmission line damper rust defect detection result.
It should be noted that, after the target detection model is obtained in step 103, the transmission line image to be identified containing the damper is input into the target detection model, so as to obtain the detection result of the rust defect of the transmission line damper, that is, whether the damper in the transmission line image to be identified containing the damper has the rust defect and the specific rust defect position are identified.
According to the method for detecting the rust defect of the transmission line vibration damper, EESP modules in ESPNetv2 models with smaller parameter are used for replacing CBS modules of C2f modules in YOLOv8s detection models, parameter of the models is reduced, PReLU activation functions with lower calculation complexity are used for replacing SiLU activation functions in convolution modules playing a role in downsampling in a main network in the YOLOv8s detection models, downsampling operation and 1X 1 standard convolution are used for replacing CBS modules of head networks in feature pyramids of the YOLOv8s detection models, an improved YOLOv8s detection model is obtained, an improved YOLOv8s detection model is trained by using a transmission line vibration damper image sample data set, a target detection model is obtained, and finally rust defects of the transmission line vibration damper are detected by using the target detection model, so that the technical reference for carrying out vibration damper detection of the transmission line vibration damper is achieved.
For ease of understanding, referring to fig. 4, an embodiment of a transmission line damper rust defect detection system is provided in the present invention, including:
the data set construction module is used for constructing a power transmission line anti-vibration hammer image sample data set, wherein the power transmission line anti-vibration hammer image sample data set comprises a rust defect image data set and a normal image data set;
the improved detection model construction module is used for replacing a CBS module of a C2f module in the YOLOv8s detection model by using an EESP module in the ESPNetv2 model, replacing a SiLU activation function in a convolution module which plays a role in downsampling in a main network in the YOLOv8s detection model, and replacing a CBS module of a head network in a feature pyramid of the YOLOv8s detection model by using downsampling operation and 1X 1 standard convolution to obtain an improved YOLOv8s detection model;
the detection model training module is used for training the improved YOLOv8s detection model according to the transmission line damper image sample data set to obtain a target detection model;
and the rust defect identification module is used for inputting the transmission line image to be identified and containing the damper into the target detection model to obtain a transmission line damper rust defect detection result.
The data set construction module is specifically used for:
acquiring an image containing a damper in a transmission line inspection image, preprocessing the image, and constructing a real transmission line damper image data set;
training a generated countermeasure network model based on a real transmission line anti-vibration hammer image data set, and generating a transmission line anti-vibration hammer image containing the anti-vibration hammer rust defect through a trained generation network in the generated countermeasure network based on an image containing the anti-vibration hammer rust defect in the real transmission line anti-vibration hammer image data set;
adding an image containing the damper rust defect in the real power transmission line damper image data set and a power transmission line damper image containing the damper rust defect generated by the generation network into the rust defect image data set, and adding an image which does not contain the damper rust defect in the real power transmission line damper image data set into the normal image data set.
The number of parallel deep hole separable volume modules in EESP modules in the ESPNetv2 model is 4, and the hole rates are 3,5,7 and 3 respectively.
The detection model training module is specifically used for:
pre-training the improved YOLOv8s detection model by using a Gogle Open Image public data set, and taking model parameters obtained by pre-training as initial parameters of the improved YOLOv8s detection model;
and training the pre-trained improved YOLOv8s detection model by using the transmission line damper image sample data set to obtain a target detection model.
The generated countermeasure network model is a SinGAN network model.
The power transmission line damper rust defect detection system provided by the invention is used for executing the power transmission line damper rust defect detection method provided by the invention, and the principle and the obtained technical effects are the same as those of the power transmission line damper rust defect detection method provided by the invention, and are not repeated herein.
The above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (8)
1. The method for detecting the rust defect of the vibration damper of the power transmission line is characterized by comprising the following steps of:
constructing a transmission line anti-vibration hammer image sample data set, wherein the transmission line anti-vibration hammer image sample data set comprises a rust defect image data set and a normal image data set;
the EESP module in the ESPNetv2 model is used for replacing the CBS module of the C2f module in the YOLOv8s detection model, the PReLU activation function is used for replacing the SiLU activation function in the convolution module which plays a role in downsampling in the main network in the YOLOv8s detection model, the downsampling operation and 1X 1 standard convolution are used for replacing the CBS module of the head network in the YOLOv8s detection model feature pyramid, an improved YOLOv8s detection model is obtained, the number of parallel depth void separable volume modules in the EESP module is 4, and the void ratio is 3,5,7 and 3 respectively;
training an improved YOLOv8s detection model according to the transmission line damper image sample data set to obtain a target detection model;
and inputting the transmission line image to be identified and containing the damper into a target detection model to obtain a transmission line damper rust defect detection result.
2. The transmission line anti-rattle rust defect detection method of claim 1, wherein constructing the transmission line anti-rattle image sample data set includes:
acquiring an image containing a damper in a transmission line inspection image, preprocessing the image, and constructing a real transmission line damper image data set;
training a generated countermeasure network model based on a real transmission line anti-vibration hammer image data set, and generating a transmission line anti-vibration hammer image containing the anti-vibration hammer rust defect through a trained generation network in the generated countermeasure network based on an image containing the anti-vibration hammer rust defect in the real transmission line anti-vibration hammer image data set;
adding an image containing the damper rust defect in the real power transmission line damper image data set and a power transmission line damper image containing the damper rust defect generated by the generation network into the rust defect image data set, and adding an image which does not contain the damper rust defect in the real power transmission line damper image data set into the normal image data set.
3. The method for detecting rust defects of a damper of a power transmission line according to claim 2, wherein training the improved YOLOv8s detection model according to the power transmission line damper image sample dataset to obtain a target detection model comprises:
pre-training the improved YOLOv8s detection model by using a Gogle Open Image public data set, and taking model parameters obtained by pre-training as initial parameters of the improved YOLOv8s detection model;
and training the pre-trained improved YOLOv8s detection model by using the transmission line damper image sample data set to obtain a target detection model.
4. The method for detecting rust defects of a damper of a transmission line according to claim 2, wherein the generated countermeasure network model is a SinGAN network model.
5. The utility model provides a transmission line damper corrosion defect detecting system which characterized in that includes:
the data set construction module is used for constructing a power transmission line anti-vibration hammer image sample data set, wherein the power transmission line anti-vibration hammer image sample data set comprises a rust defect image data set and a normal image data set;
the improved detection model construction module is used for replacing CBS modules of C2f modules in the YOLOv8s detection model by EESP modules in the ESPNetv2 model, replacing SiLU activation functions in convolution modules which play a role in downsampling in a main network in the YOLOv8s detection model by PReLU activation functions, replacing CBS modules of a head network in a feature pyramid of the YOLOv8s detection model by downsampling operation and standard convolution of 1X 1, and obtaining an improved YOLOv8s detection model, wherein the number of parallel depth void separable volume modules in the EESP modules in the ESPNetv2 model is 4, and the void ratio is 3,5,7 and 3 respectively;
the detection model training module is used for training the improved YOLOv8s detection model according to the transmission line damper image sample data set to obtain a target detection model;
and the rust defect identification module is used for inputting the transmission line image to be identified and containing the damper into the target detection model to obtain a transmission line damper rust defect detection result.
6. The transmission line anti-vibration hammer rust defect detection system of claim 5, wherein the data set construction module is specifically configured to:
acquiring an image containing a damper in a transmission line inspection image, preprocessing the image, and constructing a real transmission line damper image data set;
training a generated countermeasure network model based on a real transmission line anti-vibration hammer image data set, and generating a transmission line anti-vibration hammer image containing the anti-vibration hammer rust defect through a trained generation network in the generated countermeasure network based on an image containing the anti-vibration hammer rust defect in the real transmission line anti-vibration hammer image data set;
adding an image containing the damper rust defect in the real power transmission line damper image data set and a power transmission line damper image containing the damper rust defect generated by the generation network into the rust defect image data set, and adding an image which does not contain the damper rust defect in the real power transmission line damper image data set into the normal image data set.
7. The transmission line anti-vibration hammer rust defect detection system of claim 6, wherein the detection model training module is specifically configured to:
pre-training the improved YOLOv8s detection model by using a Gogle Open Image public data set, and taking model parameters obtained by pre-training as initial parameters of the improved YOLOv8s detection model;
and training the pre-trained improved YOLOv8s detection model by using the transmission line damper image sample data set to obtain a target detection model.
8. The transmission line anti-vibration hammer rust defect detection system of claim 6, wherein the generated countermeasure network model is a SinGAN network model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310518795.4A CN116229278B (en) | 2023-05-10 | 2023-05-10 | Method and system for detecting rust defect of vibration damper of power transmission line |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310518795.4A CN116229278B (en) | 2023-05-10 | 2023-05-10 | Method and system for detecting rust defect of vibration damper of power transmission line |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116229278A CN116229278A (en) | 2023-06-06 |
| CN116229278B true CN116229278B (en) | 2023-08-04 |
Family
ID=86570031
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310518795.4A Active CN116229278B (en) | 2023-05-10 | 2023-05-10 | Method and system for detecting rust defect of vibration damper of power transmission line |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116229278B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110634127A (en) * | 2019-07-24 | 2019-12-31 | 安徽南瑞继远电网技术有限公司 | Power transmission line vibration damper target detection and defect identification method and device |
| CN112906654A (en) * | 2021-03-29 | 2021-06-04 | 杭州电力设备制造有限公司 | Anti-vibration hammer detection method based on deep learning algorithm |
| CN112906769A (en) * | 2021-02-04 | 2021-06-04 | 国网河南省电力公司电力科学研究院 | Power transmission and transformation equipment image defect sample amplification method based on cycleGAN |
| CN114186234A (en) * | 2021-12-16 | 2022-03-15 | 西南民族大学 | Malicious code detection algorithm based on lightweight network ESPNet |
| CN114693614A (en) * | 2022-03-16 | 2022-07-01 | 武汉飞流智能技术有限公司 | Defect detection method, device and equipment for vibration damper and storage medium |
| CN115018818A (en) * | 2022-07-01 | 2022-09-06 | 南昌大学 | Power transmission line strain clamp defect detection method based on multi-network fusion model |
| CN115797357A (en) * | 2023-02-10 | 2023-03-14 | 智洋创新科技股份有限公司 | Transmission channel hidden danger detection method based on improved YOLOv7 |
| CN115995119A (en) * | 2023-03-23 | 2023-04-21 | 山东特联信息科技有限公司 | Gas cylinder filling link illegal behavior identification method and system based on Internet of things |
| CN116012601A (en) * | 2023-01-16 | 2023-04-25 | 苏州大学 | Yolo_sr system, target detection method and device for sweeping robot |
-
2023
- 2023-05-10 CN CN202310518795.4A patent/CN116229278B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110634127A (en) * | 2019-07-24 | 2019-12-31 | 安徽南瑞继远电网技术有限公司 | Power transmission line vibration damper target detection and defect identification method and device |
| CN112906769A (en) * | 2021-02-04 | 2021-06-04 | 国网河南省电力公司电力科学研究院 | Power transmission and transformation equipment image defect sample amplification method based on cycleGAN |
| CN112906654A (en) * | 2021-03-29 | 2021-06-04 | 杭州电力设备制造有限公司 | Anti-vibration hammer detection method based on deep learning algorithm |
| CN114186234A (en) * | 2021-12-16 | 2022-03-15 | 西南民族大学 | Malicious code detection algorithm based on lightweight network ESPNet |
| CN114693614A (en) * | 2022-03-16 | 2022-07-01 | 武汉飞流智能技术有限公司 | Defect detection method, device and equipment for vibration damper and storage medium |
| CN115018818A (en) * | 2022-07-01 | 2022-09-06 | 南昌大学 | Power transmission line strain clamp defect detection method based on multi-network fusion model |
| CN116012601A (en) * | 2023-01-16 | 2023-04-25 | 苏州大学 | Yolo_sr system, target detection method and device for sweeping robot |
| CN115797357A (en) * | 2023-02-10 | 2023-03-14 | 智洋创新科技股份有限公司 | Transmission channel hidden danger detection method based on improved YOLOv7 |
| CN115995119A (en) * | 2023-03-23 | 2023-04-21 | 山东特联信息科技有限公司 | Gas cylinder filling link illegal behavior identification method and system based on Internet of things |
Non-Patent Citations (1)
| Title |
|---|
| ESPNetv2: A Light-weight, Power Efficient, and General Purpose Convolutional Neural Network;Mehta Sachin 等;《2019 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2019)》;9182-9192 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116229278A (en) | 2023-06-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108154134B (en) | Pornographic image detection method is broadcast live in internet based on depth convolutional neural networks | |
| CN109344905A (en) | A kind of transmission facility automatic fault recognition methods based on integrated study | |
| CN111444939B (en) | Small-scale equipment component detection method based on weak supervision cooperative learning in open scene of power field | |
| CN111650204A (en) | Transmission line hardware defect detection method and system based on cascade target detection | |
| CN110084165B (en) | Intelligent identification and early warning method for abnormal events in open scene of power field based on edge calculation | |
| CN112102255B (en) | Intelligent defect grading method based on X-ray imaging image in industrial scene | |
| CN111401419A (en) | Improved RetinaNet-based employee dressing specification detection method | |
| CN111797890A (en) | Method and system for detecting defects of power transmission line equipment | |
| CN111339882A (en) | Power transmission line hidden danger detection method based on example segmentation | |
| CN114168906B (en) | Mapping geographic information data acquisition system based on cloud computing | |
| CN111222478A (en) | Construction site safety protection detection method and system | |
| CN111504675B (en) | On-line diagnosis method for mechanical fault of gas insulated switchgear | |
| CN111179262A (en) | Electric power inspection image hardware fitting detection method combined with shape attribute | |
| CN113642474A (en) | Hazardous area personnel monitoring method based on YOLOV5 | |
| CN115797357B (en) | Power transmission channel hidden danger detection method based on improved YOLOv7 | |
| CN110232334A (en) | A kind of steel construction corrosion recognition methods based on convolutional neural networks | |
| CN103886332A (en) | Method for detecting and identifying defects of metallic meshes | |
| CN111199213B (en) | Method and device for detecting defects of equipment for transformer substation | |
| CN115865483A (en) | Abnormal behavior analysis method and device based on machine learning | |
| CN111507398A (en) | Transformer substation metal instrument corrosion identification method based on target detection | |
| CN113962951A (en) | Training method and device for detecting segmentation model, and target detection method and device | |
| CN116229278B (en) | Method and system for detecting rust defect of vibration damper of power transmission line | |
| CN109829887B (en) | Image quality evaluation method based on deep neural network | |
| KR20210037199A (en) | Apparatus for dividing, tagging an image and for detecting defect of facilities using the same | |
| WO2023085992A1 (en) | Image analysis of cut flowers |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |