# Defect Detection Scheme for Key Equipment of Transmission Line for Complex Environment

^{1}

^{2}

^{*}

## Abstract

**:**

## 1. Introduction

- This paper uses a more lightweight network. The feature extraction network VGG-16 of Faster R-CNN is replaced by a lightweight MobileNet network, which greatly reduces redundant computation.
- In this paper, the soft-NMS algorithm is used to replace the original NMS algorithm to improve the detection accuracy, and CAROI is used to replace the original ROI pooling layer to maintain the original structure of small-sized components to improve the detection accuracy.

## 2. System Overview

## 3. Faster R-CNN Framework Improvement

#### 3.1. Basic Network

#### 3.2. RPN Network

#### 3.3. Context-Aware ROI Pooling

_{k}is the fixed size of the output feature map, F

_{k}is the input suggestion box, and h

_{k}is the kernel of the deconvolution. The size of the kernel is equal to the ratio of the output feature map to the input suggestion box. In addition, if one of the width or height of the suggested box is higher than the fixed value, the other is smaller than the fixed value. CAROI will use deconvolution to first expand the size of the proposal box, and then use max pooling to reduce the proposal box to a fixed size. Therefore, after using CAROI pooling, the proposals would have been resized to a fixed size while still extracting discriminative features from small proposals.

## 4. Kalman Filter Correction

_{k}is the Kalman gain, Z

_{k}is the system measurement value, and ${\widehat{X}}_{k-1}$ is the estimator at the previous moment. The predicted value is updated by the state measurement value at time k, and the update method is based on the minimum mean square error; finally, the best estimated value ${\widehat{X}}_{k}$ is obtained.

_{k}represents the control signal, its value is 0, w

_{k−1}is the noise function, x

_{k}is the signal, A and B are the matrix coefficients, and k represents the state index.

_{K}represents the measured value in state k; V

_{K}is a noise function, which usually obeys a Gaussian distribution; H is a coefficient matrix:

_{k}represents the Kalman gain at time k, and ${\widehat{x}}_{k}$ is the estimated value of the system at time k.

_{1}; update the error covariance matrix P

_{I}through Equation (9); recalculate the ${\widehat{X}}_{1}^{-}$ and P

_{I}obtained in the measurement update stage as the initial value of the time update, to realize the iteration.

_{k}in Kalman filtering. In the process of locating the defects of the key components of the actual transmission line, the initial prediction result is obtained through the improved Faster R-CNN and used as an observation value, which is equivalent to Z

_{K}in the Kalman filter, and then the five equations of the Kalman filter are used continuously. For iteration, each iteration process is the correction process of the Kalman filter algorithm, and finally, the optimal one is evaluated, and a set of optimal defect positions is output to further improve the defect detection performance.

## 5. Experimental Results and Analysis

#### 5.1. Experimental Environment

#### 5.2. Dataset Processing

#### 5.3. Model Training Parameter Settings

#### 5.4. Evaluation Indicators

#### 5.5. Display of Experimental Results

#### 5.6. Comparison and Analysis of Experimental Results of Different Algorithms

## 6. Conclusions

## Author Contributions

## Funding

## Conflicts of Interest

## References

- Liu, H. National Quality Infrastructure Supports Smart Grid Construction in China-Taking the State Grid as an Example. IOP Conf. Ser. Earth Environ. Sci.
**2020**, 531, 012011. [Google Scholar] [CrossRef] - Devices Create Smarter Grids with Accurate Line Fault Detection; Transmission & Distribution World: Overland Park, KS, USA, 2019.
- Wang, S.; Zhou, Z.; Zhao, W. Semantic Segmentation and Defect Detection of Aerial Insulators of Transmission Lines. J. Phys. Conf. Ser.
**2022**, 2185, 012086. [Google Scholar] [CrossRef] - Li, Y.; Song, Y.; Yang, Z.; Xie, X. Use of line laser scanning thermography for the defect detection and evaluation of composite material. Sci. Eng. Compos. Mater.
**2022**, 29, 74–83. [Google Scholar] [CrossRef] - Yücel, M.K.; Legg, M.; Kappatos, V.; Gan, T.-H. An ultrasonic guided wave approach for the inspection of overhead transmission line cables. Appl. Acoust.
**2017**, 122, 23–34. [Google Scholar] [CrossRef] - Zhang, Q.; Chang, X.; Meng, Z.; Li, Y. Equipment detection and recognition in electric power room based on faster R-CNN. Procedia Comput. Sci.
**2021**, 183, 324–330. [Google Scholar] [CrossRef] - Liu, X.; Lin, Y.; Jiang, H.; Miao, X.; Chen, J. Slippage fault diagnosis of dampers for transmission lines based on faster R-CNN and distance constraint. Electr. Power Syst. Res.
**2021**, 199, 107449. [Google Scholar] [CrossRef] - Rodriguez, F.M., Jr.; Bastos, G.B.; Seruffo, M.C.R.; Costa, F.A.R.; Figueiredo, K.; de Melo, H., Jr. Analysis of migration to the Brazilian free energy market based on statistical methods and artificial neural networks. SBIC
**2021**, 1–8. [Google Scholar] [CrossRef] - Zhou, J.H.; Liu, Z.Y.; Chen, G. Intelligent Inspection of the High-Speed Train Bogie Flaw Based on Eddy Current. Appl. Mech. Mater.
**2015**, 3785, 738–739. [Google Scholar] [CrossRef] - Huang, H.; Huang, Y.; Mu, X.; Wang, X. Research on Recognition and Location Method of Insulator in Infrared Image Based on Deep Learning. J. Phys. Conf. Ser.
**2021**, 2087, 012090. [Google Scholar] [CrossRef] - Wang, W.; Wei, J.; Zhu, Y.; Zhou, S. Power distribution equipment and defect identification technology based on deep learning. J. Phys. Conf. Ser.
**2021**, 2030, 012075. [Google Scholar] [CrossRef] - Tao, W.; Wang, W.; Yue, L.; Xie, B.; Yin, W.; Wang, H. Insulator Defect Detection Method for Lightweight YOLOV3. Comput. Eng.
**2019**, 45, 275–280. [Google Scholar] - Alahyari, A.; Hinneck, A.; Tariverdi, R.; Pozo, D. Segmentation and Defect Classification of the Power Line Insulators: A Deep Learning-based Approach. arXiv
**2020**, arXiv:2009.10163. [Google Scholar] - Lan, Y.; Xu, W. Insulator defect detection algorithm based on a lightweight network. J. Phys. Conf. Ser.
**2022**, 2181, 012007. [Google Scholar] [CrossRef] - Qi, Y.; Du, L.; Zhao, Z.; Cui, Y.; Si, W. Insulator Detection Based on SSD with the Default Box Adaptively Selection. Comput. Sci. Appl. Eng.
**2018**, 110, 1–4. [Google Scholar] [CrossRef] - Ni, H.; Wang, M.; Zhao, L. An improved Faster R-CNN for defect recognition of key components of transmission line. Math. Biosci. Eng.
**2021**, 18, 4679–4695. [Google Scholar] [CrossRef] - Zhao, Z.; Fan, X.; Xu, G.; Zhang, L.; Qi, Y.; Zhang, K. Aggregating Deep Convolutional Feature Maps for Insulator Detection in Infrared Images. IEEE Access
**2017**, 5, 21831–21839. [Google Scholar] [CrossRef] - Zhou, Y.; Wen, S.; Wang, D.; Mu, J.; Irampaye, R. Object Detection in Autonomous Driving Scenarios Based on an Improved Faster-RCNN. Appl. Sci.
**2021**, 11, 11630. [Google Scholar] [CrossRef] - Lv, L.; Tan, Y. Detection of cabinet in equipment floor based on AlexNet and SSD model. J. Eng.
**2019**, 2019, 605–608. [Google Scholar] [CrossRef] - Gao, S.; Kang, G.; Yu, L.; Zhang, D.; Wei, X.; Zhan, D. Adaptive Deep Learning for High-Speed Railway Catenary Swivel Clevis Defects Detection. IEEE Trans. Intell. Transp. Syst.
**2020**, 23, 1299–1310. [Google Scholar] [CrossRef] - Zhu, J.; Chang, X.; Zhang, X.; Su, Y.; Long, X. A Novel Method for the Reconstruction of Road Profiles from Measured Vehicle Responses Based on the Kalman Filter Method. Comput. Model. Eng. Sci.
**2022**, 130, 1719–1735. [Google Scholar] [CrossRef] - Ren, K.; Zhang, D.; Mohammed, S.; Calvi, A. A Kalman filtering fuzzy logic algorithm for recognition of lane departure. J. Intell. Fuzzy Syst.
**2021**, 41, 4855–4862. [Google Scholar] [CrossRef] - Liu, Z.; Liu, W.; Han, Z. A high-precision detection approach for catenary geometry parameters of electrical railway. IEEE Trans. Instrum. Meas.
**2017**, 66, 1798–1808. [Google Scholar] [CrossRef] - Li, Y.; Huang, H.; Xie, Q.; Yao, L.; Chen, Q. Research on a Surface Defect Detection Algorithm Based on MobileNet-SSD. Appl. Sci.
**2018**, 8, 1678. [Google Scholar] [CrossRef] [Green Version] - Bodla, N.; Singh, B.; Chellappa, R.; Davis, L.S. Soft-NMS—Improving Object Detection with One Line of Code. In Proceedings of the 2017 IEEE International Conference on Computer Vision (ICCV), Venice, Italy, 22–29 October 2017. [Google Scholar]
- Yao, L.; Sheng, Q.Z.; Qin, Y.; Wang, X.; Shemshadi, A.; He, Q. Context-aware Point-of-Interest Recommendation Using Tensor Factorization with Social Regularization. In Proceedings of the International ACM Sigir Conference, Santiago, Chile, 9–13 August 2015; ACM: New York, NY, USA, 2015. [Google Scholar]
- Xie, X.; Xiao, H.; Quan, L.; Shi, G. Visualization and Pruning of SSD with the base network VGG16. In Proceedings of the 2017 International Conference, Paris, France, 21–25 May 2017; ACM: New York, NY, USA, 2017. [Google Scholar]
- Shi, M.; Ouyang, P.; Yin, S.; Liu, L.; Wei, S. A Fast and Power-Efficient Hardware Architecture for Non-Maximum Suppression. IEEE Trans. Circuits Syst. II Express Briefs
**2019**, 66, 1870–1874. [Google Scholar] [CrossRef] - Wang, Y.; Liu, Z.; Deng, W. Anchor Generation Optimization and Region of Interest Assignment for Vehicle Detection. Sensors
**2019**, 19, 1089. [Google Scholar] [CrossRef] [PubMed] [Green Version] - Tun, N.L.; Gavrilov, A.; Tun, N.M.; Trieu, D.M.; Aung, H. Remote Sensing Data Classification Using a Hybrid Pre-Trained VGG16 CNN-SVM Classifier. In Proceedings of the 2021 IEEE Conference of Russian Young Researchers in Electrical and Electronic Engineering (ElConRus), Moscow, Russia, 26–29 January 2021. [Google Scholar]
- Zhao, W.; Yan, H.; Shao, X. Object detection based on improved non-maximum suppression algorithm. Appl. Soft Comput.
**2019**, 81, 105478. [Google Scholar] - Wang, K.; Liu, M.Z. Object Recognition at Night Scene Based on DCGAN and Faster R-CNN. IEEE Access
**2020**, 8, 193168–193182. [Google Scholar] [CrossRef] - Yan, C.; Chen, W.; Chen, P.C.Y.; Kendrick, A.S.; Wu, X. A new two-stage object detection network without RoI-Pooling. In Proceedings of the 2018 Chinese Control and Decision Conference (CCDC), Shenyang, China, 9–11 June 2018. [Google Scholar]
- Hu, X.; Xu, X.; Xiao, Y.; Chen, H. SINet: A Scale-insensitive Convolutional Neural Network for Fast Vehicle Detection. IEEE Trans. Intell. Transp. Syst.
**2018**, 3, 1–10. [Google Scholar] [CrossRef] [Green Version] - Luisier, F.; Blu, T.; Unser, M. Image Denoising in Mixed Poisson–Gaussian Noise. IEEE Trans. Image Process.
**2011**, 20, 696–708. [Google Scholar] [CrossRef] [Green Version] - Sun, K.; Simon, S. Bilateral Spectrum Weighted Total Variation for Real-World Super-Resolution and Image Denoising. arXiv
**2021**, arXiv:2106.00768. [Google Scholar] - Foi, A.; Trimeche, M.; Katkovnik, V.; Egiazarian, K. Practical Poissonian-Gaussian Noise Modeling and Fitting for Single-Image Raw-Data. IEEE Trans. Image Process.
**2008**, 17, 1737–1754. [Google Scholar] [CrossRef] [Green Version] - An, Y.; Wang, X.; Chu, R.; Yue, B.; Wu, L.; Cui, J.; Qu, Z. Event classification for natural gas pipeline safety monitoring based on long short-term memory network and Adam algorithm. Struct. Health Monit.
**2019**, 19, 1151–1159. [Google Scholar] [CrossRef] - Fraser, C.T.; Ulrich, S. Adaptive extended Kalman filtering strategies for spacecraft formation relative navigation. Acta Astronaut.
**2021**, 178, 700–721. [Google Scholar] [CrossRef] - Zhang, Y.; Wang, R.; Li, S.; Qi, S. Temperature Sensor Denoising Algorithm Based on Curve Fitting and Compound Kalman Filtering. Sensors
**2020**, 20, 1959. [Google Scholar] [CrossRef] [PubMed] [Green Version] - Kohli, H.; Agarwal, J.; Kumar, M. An Improved Method for Text Detection using Adam Optimization Algorithm. Glob. Transit. Proc.
**2022**, 3, 230–234. [Google Scholar] [CrossRef]

**Figure 7.**Defect detection results; (

**a**) insulator, (

**b**) wire clip and (

**c**) hardware. Among them, the left side is the preliminary detection result, and the right side is the corrected result.

Model | Precision (%) | Madd (Million) | Parameter (Million) |
---|---|---|---|

VGG-16 | 71.52 | 16,000 | 140 |

MobileNet | 70.83 | 570 | 4.3 |

Category | Version |
---|---|

operating system | Windows 10 |

CPU | Intel Core i9-10900 k |

GPU | NVIDIA GeForce GTX 3080 |

RAM | 32 GB |

Tensorflow-gpu | Tensorflow-gpu1.13.2 |

Keras | Keras2.1.5 |

Cuda | Cuda10.0 |

Cudnn | Cudnn7.4.1.5 |

Defet Detection Methods | P (%) | R (%) | mAP (%) | IOU (%) | Detection Time (s) |
---|---|---|---|---|---|

YOLO | 61.23 | 59.61 | 60.61 | 60.24 | 0.04 |

SSD | 77.96 | 73.18 | 77.42 | 76.55 | 0.06 |

MS-CNN | 85.83 | 83.06 | 86.53 | 83.36 | 0.40 |

Original Faster R-CNN | 80.36 | 78.62 | 80.05 | 80.16 | 2.00 |

Improve Faster R-CNN | 86.44 | 83.09 | 86.16 | 85.31 | 0.10 |

Improve Faster R-CNN + Kalman filter | 90.26 | 87.49 | 91.10 | 88.37 | 0.12 |

Number of Defective Samples | $\mathit{p}$/% | |||||
---|---|---|---|---|---|---|

1 | 5 | 10 | 20 | 30 | 40 | |

10 | 82.35 | 81.25 | 79.58 | 76.35 | 74.35 | 72.14 |

20 | 83.25 | 82.42 | 82.32 | 81.35 | 79.56 | 78.36 |

50 | 85.65 | 84.54 | 83.35 | 83.14 | 82.54 | 78.25 |

Denoising Methods | P (%) | R (%) | mAP (%) | IOU (%) | Detection Time (s) |
---|---|---|---|---|---|

mean filter | 86.25 | 87.26 | 84.58 | 88.65 | 0.14 |

median filter | 85.32 | 83.54 | 86.76 | 85.47 | 0.11 |

wavelet denoising | 84.35 | 85.25 | 87.54 | 84.35 | 0.13 |

Gaussian filter | 80.32 | 79.61 | 78.65 | 81.23 | 0.09 |

bilateral filter | 87.35 | 85.36 | 88.45 | 84.36 | 0.13 |

Kalman filter | 90.26 | 87.49 | 91.10 | 88.37 | 0.12 |

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |

© 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Wang, J.; Deng, F.; Wei, B.
Defect Detection Scheme for Key Equipment of Transmission Line for Complex Environment. *Electronics* **2022**, *11*, 2332.
https://doi.org/10.3390/electronics11152332

**AMA Style**

Wang J, Deng F, Wei B.
Defect Detection Scheme for Key Equipment of Transmission Line for Complex Environment. *Electronics*. 2022; 11(15):2332.
https://doi.org/10.3390/electronics11152332

**Chicago/Turabian Style**

Wang, Jian, Fangming Deng, and Baoquan Wei.
2022. "Defect Detection Scheme for Key Equipment of Transmission Line for Complex Environment" *Electronics* 11, no. 15: 2332.
https://doi.org/10.3390/electronics11152332