CN112070715A - Transmission line small-size hardware defect detection method based on improved SSD model - Google Patents
Transmission line small-size hardware defect detection method based on improved SSD model Download PDFInfo
- Publication number
- CN112070715A CN112070715A CN202010750096.9A CN202010750096A CN112070715A CN 112070715 A CN112070715 A CN 112070715A CN 202010750096 A CN202010750096 A CN 202010750096A CN 112070715 A CN112070715 A CN 112070715A
- Authority
- CN
- China
- Prior art keywords
- small
- detection
- network model
- size
- size hardware
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
- G06T7/0008—Industrial image inspection checking presence/absence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30136—Metal
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
- G06T2207/30164—Workpiece; Machine component
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
The invention relates to a method for detecting the defect of a small-size hardware fitting of a power transmission line based on an improved SSD model, which detects a network model structure by improving an SSD target; collecting samples for model training; and after the training parameters are adjusted; and carrying out corrosion state detection and/or pin missing detection on the small-size hardware fittings of the class a and the class b detected by the network model. By the detection method, the defects of small-size hardware fittings in the inspection image are intelligently identified, and the inspection efficiency of the power transmission line is improved.
Description
Technical Field
The invention relates to the field of power transmission line detection, in particular to a method for detecting defects of small-size hardware fittings of a power transmission line based on an improved SSD model.
Background
The power transmission line is an artery of a power system, the power transmission line inspection is safe and stable in operation, the 2/3 terrain in China is a mountain land, a large number of power transmission lines pass through places which are difficult to reach manually, such as mountains, desert and the like, and the manual inspection efficiency is low; the length of the 110kV and above transmission line is nearly 100 kilometers, compared with the fixed member of the power transmission operation and maintenance personnel, the length of the transmission line is nearly 50 percent, the height of the ultra-high voltage iron tower exceeds 60m, and the defect of the part above the bottle mouth is difficult to find by manual inspection; therefore, the operation and maintenance tasks of the line can be completed only by means of intelligent routing inspection technology. Along with the development of the technology, unmanned aerial vehicle inspection and channel visual inspection have gradually replaced manual inspection and become the main mode of power transmission line inspection at present.
The small-size hardware fitting in the power transmission line mainly comprises a bolt, a nut and a pin. The bolt, the nut and the pin play a role in connecting key components in the power transmission line, and are exposed in an external environment for a long time, so that the defects of corrosion, loss and the like can be caused under the influence of factors such as human factors, severe weather, electrical flashover, mechanical tension, material aging and the like, and the defects directly cause component deformation, instability and even electric power failure accidents.
Unmanned aerial vehicle and the visual inspection of passageway produce a large amount of images, and the research about electric power inspection image processing technique is still in the starting stage at home and abroad at present. Because the small-size hardware in the image occupies fewer pixel points and belongs to a small target object, and a large number of small-size hardware exist on the power transmission line and the tower, how to accurately position the small-size hardware in a complex background by using an image processing technology is a difficult point. The small-size hardware defects comprise: the method is characterized by comprising the following steps of rusting, loosening, pin falling, improper pin installation and the like, and the method is mainly used for detecting the defects of rusting and falling of part types of pins.
Disclosure of Invention
Based on the above situation in the prior art, the invention aims to provide a method for detecting defects of small-size hardware fittings of a power transmission line based on an improved SSD model, which is used for intelligently identifying the defects of the small-size hardware fittings in inspection images and improving the inspection efficiency of the power transmission line.
In order to achieve the purpose, the invention provides a method for detecting the defect of the small-size hardware fitting of the power transmission line based on an improved SSD model, which comprises the following steps:
improving the structure of the SSD target detection network model;
collecting an image sample containing small-size hardware and carrying out model training;
adjusting training parameters to ensure model accuracy and generalization capability;
carrying out corrosion state detection on the detected small-size hardware fitting class a by adopting an improved network model;
and (3) carrying out corrosion state detection and pin missing detection on the detected small-size hardware fitting class b by adopting the improved network model.
Further, the improved SSD object detection network model structure includes: VGG _16 used for extracting image shallow features in the original SSD network structure is changed into ResNet101, and more shallow features are reserved while the receptive field is expanded by introducing a residual error network.
Further, the improved SSD object detection network model structure includes: and changing the input size of the model to reserve more features of the small-size hardware in each feature layer.
Further, the improved SSD object detection network model structure includes: adding a shallow feature layer on the basis of the original six feature layers, fusing the specific feature layers to form new four feature layers, and performing convolution twice on the four formed feature layers to obtain the positions and the types of the prediction frames.
Further, the improved SSD object detection network model structure includes: the anchor generation mechanism is modified, and the ratio of the anchor area to the corresponding feature layer area is adjusted.
Furthermore, the image samples containing the small-size fittings are collected, and the unmanned aerial vehicle inspection images are used as the image samples.
Further, the training parameter is adjusted, specifically, the batch size is 16, when the number of first 10000 iterations is larger, the learning rate is set to 0.001, and then the learning rate is adjusted to 0.0001, and the superparameters _ optimizer and the decapay _ factor are respectively set to 0.9 and 0.95.
Further, the small-size hardware a detected by the network model is subjected to corrosion state detection, specifically, an image is converted from an RGB color space to an HIS color space, and a corrosion area is identified based on hue H and saturation S.
Further, the small-size hardware fitting b detected by the network model is subjected to corrosion state detection, specifically, an image is converted from an RGB color space to an HIS color space, and a corrosion area is identified based on hue H and saturation S.
Further, the pin missing detection is performed on the small-size hardware fitting class b detected by the network model, and specifically includes:
carrying out graying processing on the image;
fitting the outlines of the nut, the bolt and the pin through edge detection;
judging whether a fitted line at the other end of the opposite nut is intersected with the axis or not according to the central axis of the bolt, wherein the distance between the vertex and the axis is longer than the distance between the edge of the nut and the axis;
if yes, the pin is present, otherwise, the pin is detached.
In summary, the invention provides a method for detecting the defect of the small-size hardware fitting of the power transmission line based on an improved SSD model, which detects a network model structure by improving an SSD target; collecting samples for model training; and after the training parameters are adjusted; and carrying out corrosion state detection and/or pin missing detection on the small-size hardware fittings of the class a and the class b detected by the network model. The technical scheme of the invention has the following beneficial technical effects: the small-size hardware defect in the inspection image is intelligently identified, and the inspection efficiency of the power transmission line is improved.
Drawings
FIG. 1 is a schematic diagram of an improved SSD network architecture;
FIG. 2 is a schematic diagram of an annotation image and a corresponding label, and FIG. 2-1 shows an annotation with a category of 1: boltnuts, fig. 2-2, label category 2: dowel;
FIG. 3 is a class a hardware fitting detection result and a rust identification result, wherein FIG. 3-a is a network-detected small-size hardware fitting, and FIG. 3-b is a rust identification result;
FIG. 4 is a schematic diagram of a pin detection process, wherein FIG. 4-a is a class b small-size fitting detected by a model, FIG. 4-b is a diagram after graying, and FIG. 4-c is a diagram of an edge detection result;
fig. 5 is a schematic diagram of a conventional SSD network structure.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the accompanying drawings in conjunction with the following detailed description. It should be understood that the description is intended to be exemplary only, and is not intended to limit the scope of the present invention. Moreover, in the following description, descriptions of well-known structures and techniques are omitted so as to not unnecessarily obscure the concepts of the present invention.
The invention provides a method for detecting the defects of small-size hardware fittings of a power transmission line based on an improved SSD model, which detects a network model structure by improving an SSD target; collecting samples for model training; and after the training parameters are adjusted; and carrying out corrosion state detection and/or pin missing detection on the small-size hardware fittings of the class a and the class b detected by the network model.
First, a brief description of the SSD primitive model is given. The SSD algorithm integrates the explicit-free candidate box extraction of YOLO and the Anchor mechanism in Faster R-CNN, and integrates the characteristics of different convolutional layers in a characteristic space for prediction. The model structure mainly comprises three parts, wherein VGG _16 is used for extracting multilayer image features, a group of cascade CNNs further extract feature information under different size conditions, a bounding box with a fixed size is generated on each input image, the multilevel features are simultaneously input into a detector, and the simple description of the regression calculation and maximum suppression structure is shown in FIG. 5.
The terms used in the present invention are explained as follows:
SSD: single Shot MultiBox Detector, an open source target detection model.
an anchor: in the SSD network structure, a plurality of candidate frames are generated in advance on each feature layer, then the candidate frames are screened through a series of operations, the coordinate positions are adjusted, and finally the frame for target detection is obtained, wherein anchors approximately represent the candidate frames.
momentum _ optimizer, decapy _ factor: and in the training process, parameters for accelerating convergence speed and preventing gradient explosion are used.
The technical solution of the present invention will be described in detail below with reference to the accompanying drawings. The invention provides a method for detecting defects of small-size hardware fittings of a power transmission line based on an improved SSD model, which comprises the following steps:
the first step is as follows: improving the structure of the SSD target detection network model;
the second step is that: collecting an image sample containing small-size hardware and carrying out model training;
the third step: adjusting training parameters to ensure model accuracy and generalization capability;
the fourth step: carrying out corrosion state detection on the detected small-size hardware fitting class a by using the improved network model;
the fifth step: and (3) carrying out corrosion state detection and pin missing detection on the detected small-size hardware fitting class b by adopting the improved network model.
The small-size hardware a and b involved in the method are further explained: two labels exist in the small-size hardware after model identification, one label is a type, and the other label is b type; class a indicates that the bolt and nut with the pin are not needed to be subjected to pin detection, and specifically, reference is made to the labeled class 1 of fig. 2-1: boltnuts; the b type represents that the bolt and the nut with the pin need to be subjected to pin detection, and particularly, the reference types shown in the figures 2-2 are 1: dowel (r).
According to some embodiments, the first step of improving the SSD object detection network structure and improving the small object detection performance includes four aspects. On the first hand, VGG _16 used for extracting image shallow features in the original SSD network structure is changed into ResNet101, and more shallow features are reserved while the receptive field is expanded by introducing a residual error network, so that the detection of small-size hardware of the power transmission line is facilitated. In the second aspect, the input size of the model is changed from 300 × 300 to 800 × 800, and more features of small-size hardware in each feature layer are reserved. In certain other embodiments, the improved SSD network model input image may not necessarily be 800 × 800, as well as 10% fine-tuning. In the third aspect, a shallow feature layer conv3_4 is added on the basis of six original feature layers conv4_3, fc7, conv6_2, conv7_2, conv8_2 and conv9_2, conv9_2 is up-sampled and then fused with conv6_2 to form a new feature layer conv6_9, conv8_2 is up-sampled and then fused with fc7 to form a feature layer conv7_8, conv7_2 is up-sampled and then fused with conv4_3 to form a feature layer conv4_7, and finally conv3_4, conv4_7, conv7_8 and conv6_9 are used as final feature layers to be convolved twice respectively to obtain the predicted frame position and the type. In a fourth aspect, correspondingly, the anchor generation mechanism is modified, the anchor area to corresponding feature area ratio Sk is adjusted to 0.1, 0.2, 0.3, 0.4, the aspect ratio of each pixel is {1,2,3,1/2,1/3}, and when the aspect ratio is 1, a scale 1.414 x Sk is added. In certain other embodiments, the improved SSD network model modification anchor size Sk may not necessarily be 0.1, 0.2, 0.3, 0.4, but may be an inner size of ± 0.1. A modified SSD network structure is shown in fig. 1.
According to some embodiments, in the second step, image samples containing small-size hardware are collected and model training is performed. The image acquisition mode mainly patrols and examines the image for unmanned aerial vehicle, guarantees that the image has higher definition, and the pixel specification is close. And the image annotation adopts an open source image annotation tool LabelImg, the annotation generates an xml file, and the annotation category is 1: boltnuts, 2: a schematic diagram of the labeled image and the corresponding label is shown in fig. 2.
According to some embodiments, the training parameters are adjusted in the third step, so that model accuracy and generalization ability are guaranteed. During training, the batch size is 16, the learning rate is set to 0.001 for the first 10000 iterations, then the learning rate is adjusted to 0.0001, and the superparameters _ optimizer and decapay _ factor are set to 0.9 and 0.95 respectively. In some other embodiments, the relevant parameters may not be as described above, but may be fine-tuned by 10%.
According to some embodiments, in the fourth step, the small-size hardware detected by the network model is subjected to corrosion state detection, the small-size hardware detected by the network is shown in fig. 3-a below, then the image is converted from the RGB color space to the HSI color space based on the formulas (1) - (4), the corrosion area is identified based on the hue H and the saturation S, and after repeated adjustment, when the H value is 0-30 and 330-360, and the S is 0.1, the corrosion identification effect is better, and the identification result is shown in fig. 3-b.
I＝(R+G+B)/3 (4)
According to some embodiments, in the fifth step, the small-size hardware detected by the network model is subjected to corrosion state detection and pin missing detection, where the corrosion state detection process is as described in the specific embodiment of the fourth step, and the pin missing detection step is as follows: firstly, carrying out gray processing on an image, then fitting the outlines of the nut, the bolt and the pin through edge detection, judging whether a fitted line is intersected with an axis at the other end of the nut or not according to the central axis of the bolt, wherein the distance between a vertex and the axis is longer than the distance between the edge of the nut and the axis, if so, indicating that the pin exists, and otherwise, indicating that the pin falls off. The detection flow is shown in fig. 4, where fig. 4-a is a b-class small-size hardware detected by the model, fig. 4-b is a graph after graying, and fig. 4-c is an edge detection result graph.
In summary, the invention relates to a method for detecting the defect of the small-size hardware fitting of the power transmission line based on an improved SSD model, which detects a network model structure by improving an SSD target; collecting samples for model training; and after the training parameters are adjusted; and carrying out corrosion state detection and/or pin missing detection on the small-size hardware fittings of the class a and the class b detected by the network model. The small-size hardware defect in the inspection image is intelligently identified, and the inspection efficiency of the power transmission line is improved.
It is to be understood that the above-described embodiments of the present invention are merely illustrative of or explaining the principles of the invention and are not to be construed as limiting the invention. Therefore, any modification, equivalent replacement, improvement and the like made without departing from the spirit and scope of the present invention should be included in the protection scope of the present invention. Further, it is intended that the appended claims cover all such variations and modifications as fall within the scope and boundaries of the appended claims or the equivalents of such scope and boundaries.
Claims (10)
1. A method for detecting defects of small-size hardware fittings of a power transmission line based on an improved SSD model is characterized by comprising the following steps:
improving the structure of the SSD target detection network model;
collecting an image sample containing small-size hardware and carrying out model training;
adjusting training parameters to ensure model accuracy and generalization capability;
carrying out corrosion state detection on the detected small-size hardware fitting class a by adopting an improved network model;
and (3) carrying out corrosion state detection and pin missing detection on the detected small-size hardware fitting class b by adopting the improved network model.
2. The detection method according to claim 1, wherein the improving the SSD destination detection network model structure comprises: VGG _16 used for extracting image shallow features in the original SSD network structure is changed into ResNet101, and more shallow features are reserved while the receptive field is expanded by introducing a residual error network.
3. The detection method according to claim 1, wherein the improving the SSD destination detection network model structure comprises: and changing the input size of the model to reserve more features of the small-size hardware in each feature layer.
4. The detection method according to claim 1, wherein the improving the SSD destination detection network model structure comprises: adding a shallow feature layer on the basis of the original six feature layers, fusing the specific feature layers to form new four feature layers, and performing convolution twice on the four formed feature layers to obtain the positions and the types of the prediction frames.
5. The detection method according to claim 1, wherein the improving the SSD destination detection network model structure comprises: the anchor generation mechanism is modified, and the ratio of the anchor area to the corresponding feature layer area is adjusted.
6. The detection method according to claim 1, wherein the image samples containing the small-size hardware are collected, and unmanned aerial vehicle inspection images are adopted as the image samples.
7. The detection method according to claim 1, wherein the learning rate is set to 0.001 when the training parameters, specifically the batch size is 16 and the first 10000 iterations are performed, and then the learning rate is adjusted to 0.0001, and the momentum _ optimizer and the decapay _ factor are respectively set to 0.9 and 0.95.
8. The detection method according to claim 1, wherein the rust state detection is carried out on the small-size hardware a detected by the network model, specifically, an image is converted from an RGB color space to an HIS color space, and a rust area is identified based on hue H and saturation S.
9. The detection method according to claim 1, wherein the rust state detection is carried out on the small-size hardware detected by the network model, specifically, an image is converted from an RGB color space to an HIS color space, and a rust area is identified based on hue H and saturation S.
10. The detection method according to claim 1, wherein the pin missing detection is performed on the class b small-size hardware detected by the network model, specifically:
carrying out graying processing on the image;
fitting the outlines of the nut, the bolt and the pin through edge detection;
judging whether a fitted line at the other end of the opposite nut is intersected with the axis or not according to the central axis of the bolt, wherein the distance between the vertex and the axis is longer than the distance between the edge of the nut and the axis;
if yes, the pin is present, otherwise, the pin is detached.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010750096.9A CN112070715A (en) | 2020-07-30 | 2020-07-30 | Transmission line small-size hardware defect detection method based on improved SSD model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010750096.9A CN112070715A (en) | 2020-07-30 | 2020-07-30 | Transmission line small-size hardware defect detection method based on improved SSD model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112070715A true CN112070715A (en) | 2020-12-11 |
Family
ID=73656896
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010750096.9A Pending CN112070715A (en) | 2020-07-30 | 2020-07-30 | Transmission line small-size hardware defect detection method based on improved SSD model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112070715A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112966788A (en) * | 2021-04-19 | 2021-06-15 | 扬州大学 | Power transmission line spacer fault detection method based on deep learning |
| CN113129300A (en) * | 2021-05-10 | 2021-07-16 | 深圳市水务工程检测有限公司 | Drainage pipeline defect detection method, device, equipment and medium for reducing false detection rate |
| CN113361473A (en) * | 2021-06-30 | 2021-09-07 | 北京百度网讯科技有限公司 | Image processing method, model training method, device, apparatus, storage medium, and program |
| CN113515829A (en) * | 2021-05-21 | 2021-10-19 | 华北电力大学（保定） | Situation sensing method for transmission line hardware defects under extremely cold disasters |
| CN114049354A (en) * | 2022-01-12 | 2022-02-15 | 山东仲良格环保技术有限公司 | Rust remover optimized proportioning method and system based on metal corrosion degree |
| CN114782679A (en) * | 2022-05-05 | 2022-07-22 | 国家电网有限公司 | Hardware defect detection method and device in power transmission line based on cascade network |
| CN115841460A (en) * | 2022-11-21 | 2023-03-24 | 国网湖北省电力有限公司超高压公司 | High-precision hardware crack image detection and feature extraction method under complex background |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107741427A (en) * | 2017-11-30 | 2018-02-27 | 南京航空航天大学 | The machine vision detection device and method of automobile engine turbine case surface defect |
| CN108389197A (en) * | 2018-02-26 | 2018-08-10 | 上海赛特斯信息科技股份有限公司 | Transmission line of electricity defect inspection method based on deep learning |
| CN108921844A (en) * | 2018-07-06 | 2018-11-30 | 许继集团有限公司 | A kind of defects of insulator detection method and device |
| CN110569887A (en) * | 2019-08-20 | 2019-12-13 | 高新兴科技集团股份有限公司 | Target detection method based on prediction layer feature enhancement, storage medium and device |
| CN110827251A (en) * | 2019-10-30 | 2020-02-21 | 江苏方天电力技术有限公司 | Power transmission line locking pin defect detection method based on aerial image |
| CN111080598A (en) * | 2019-12-12 | 2020-04-28 | 哈尔滨市科佳通用机电股份有限公司 | Bolt and nut missing detection method for coupler yoke key safety crane |
| CN111121709A (en) * | 2019-12-26 | 2020-05-08 | 湖北航天技术研究院总体设计所 | System and method for measuring thrust line of small-size large-expansion-ratio nozzle engine |
-
2020
- 2020-07-30 CN CN202010750096.9A patent/CN112070715A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107741427A (en) * | 2017-11-30 | 2018-02-27 | 南京航空航天大学 | The machine vision detection device and method of automobile engine turbine case surface defect |
| CN108389197A (en) * | 2018-02-26 | 2018-08-10 | 上海赛特斯信息科技股份有限公司 | Transmission line of electricity defect inspection method based on deep learning |
| CN108921844A (en) * | 2018-07-06 | 2018-11-30 | 许继集团有限公司 | A kind of defects of insulator detection method and device |
| CN110569887A (en) * | 2019-08-20 | 2019-12-13 | 高新兴科技集团股份有限公司 | Target detection method based on prediction layer feature enhancement, storage medium and device |
| CN110827251A (en) * | 2019-10-30 | 2020-02-21 | 江苏方天电力技术有限公司 | Power transmission line locking pin defect detection method based on aerial image |
| CN111080598A (en) * | 2019-12-12 | 2020-04-28 | 哈尔滨市科佳通用机电股份有限公司 | Bolt and nut missing detection method for coupler yoke key safety crane |
| CN111121709A (en) * | 2019-12-26 | 2020-05-08 | 湖北航天技术研究院总体设计所 | System and method for measuring thrust line of small-size large-expansion-ratio nozzle engine |
Non-Patent Citations (2)
| Title |
|---|
| 戚银城 等: "基于改进SSD模型的输电线路巡检图像金具检测方法", 《电测与仪表》, 25 November 2019 (2019-11-25), pages 7 - 12 * |
| 祁婕 等: "改进SSD的输电铁塔鸟窝检测", 《计算机系统应用》, 7 May 2020 (2020-05-07), pages 202 - 208 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112966788A (en) * | 2021-04-19 | 2021-06-15 | 扬州大学 | Power transmission line spacer fault detection method based on deep learning |
| CN113129300A (en) * | 2021-05-10 | 2021-07-16 | 深圳市水务工程检测有限公司 | Drainage pipeline defect detection method, device, equipment and medium for reducing false detection rate |
| CN113515829A (en) * | 2021-05-21 | 2021-10-19 | 华北电力大学（保定） | Situation sensing method for transmission line hardware defects under extremely cold disasters |
| CN113361473A (en) * | 2021-06-30 | 2021-09-07 | 北京百度网讯科技有限公司 | Image processing method, model training method, device, apparatus, storage medium, and program |
| CN113361473B (en) * | 2021-06-30 | 2023-12-08 | 北京百度网讯科技有限公司 | Image processing method, model training method, image processing device, model training apparatus, storage medium, and program |
| CN114049354A (en) * | 2022-01-12 | 2022-02-15 | 山东仲良格环保技术有限公司 | Rust remover optimized proportioning method and system based on metal corrosion degree |
| CN114782679A (en) * | 2022-05-05 | 2022-07-22 | 国家电网有限公司 | Hardware defect detection method and device in power transmission line based on cascade network |
| CN115841460A (en) * | 2022-11-21 | 2023-03-24 | 国网湖北省电力有限公司超高压公司 | High-precision hardware crack image detection and feature extraction method under complex background |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112070715A (en) | Transmission line small-size hardware defect detection method based on improved SSD model | |
| CN112200178B (en) | Transformer substation insulator infrared image detection method based on artificial intelligence | |
| CN105160669B (en) | The detection of defects of insulator and localization method in a kind of unmanned plane inspection transmission line of electricity image | |
| CN109712127B (en) | Power transmission line fault detection method for machine inspection video stream | |
| CN107679495B (en) | Detection method for movable engineering vehicles around power transmission line | |
| CN115240075B (en) | Construction and training method of electric power vision multi-granularity pre-training large model | |
| CN106600580B (en) | Hough transform-based power line anomaly identification method and system | |
| Liu et al. | Summary of insulator defect detection based on deep learning | |
| CN114862768A (en) | Improved YOLOv5-LITE lightweight-based power distribution assembly defect identification method | |
| Yuan et al. | Identification method of typical defects in transmission lines based on YOLOv5 object detection algorithm | |
| CN110705397A (en) | YOLOv3 pruning identification insulator defect method suitable for small field sample amount | |
| CN115984672B (en) | Detection method and device for small target in high-definition image based on deep learning | |
| CN110059076A (en) | A kind of Mishap Database semi-automation method for building up of power transmission and transformation line equipment | |
| CN115147383A (en) | Insulator state rapid detection method based on lightweight YOLOv5 model | |
| CN112837281B (en) | Pin defect identification method, device and equipment based on cascade convolution neural network | |
| CN113450318A (en) | Primary power distribution system porcelain insulator detection method based on unmanned aerial vehicle autonomous vision | |
| Hong-Bin et al. | Target tracking method of transmission line insulator based on multi feature fusion and adaptive scale filter | |
| CN116452848A (en) | Hardware classification detection method based on improved attention mechanism | |
| CN113989209B (en) | Power line foreign matter detection method based on Faster R-CNN | |
| Qi et al. | Optimized YOLOX based transmission line bolt cascade detection | |
| CN111930982A (en) | Intelligent labeling method for power grid images | |
| Sheng et al. | A YOLOX-Based Detection Method of Triple-Cascade Feature Level Fusion for Power System External Defects | |
| Zhang et al. | Infrared Image Target Detection Algorithm for Hardware Fittings at Tension Clamp of Qishao UHV Transmission Line | |
| Li et al. | Improved YOLOV7 algorithm for transmission line metal fittings equipment detection: algorithm improvement for integrating attention mechanism | |
| Liu et al. | Defect insulator detection method based on deep learning |
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 |
Application publication date: 20201211 |
|
| RJ01 | Rejection of invention patent application after publication |