Article Details

Abstract

To compensate for the shortage of limited inspection range of inspection robots and UAVs in new energy power generation enterprises, this paper designs an audio and video fusion intelligent inspection system for the field of new energy power generation. On the station perception side, various types of cameras, audio sensors, and other perception equipment are comprehensively applied to collect audio and video multimodal data of wind turbines, power lines, photovoltaics, and booster stations in various areas. After the analysis of intelligent algorithm based on Deep Learning, the data is converted into equipment status recognition results. On the application side of the regional center, it deploys intelligent inspection application service to display the recognition results of equipment status on the station side, and builds intelligent inspection business functions such as inspection task management and alarm management, so as to facilitate the operation of equipment maintenance personnel. The perception scheme of audio and video fusion proposed in this paper, and the integrated application of algorithm frameworks such as YOLO, ResNet and DeepLab, expand the coverage of intelligent inspection points in new energy stations, reduce the frequency of personnel going to the site, and help to promote the implementation of the operation and maintenance management mode of new energy power unmanned stations.

Keywords

intelligent inspection; edge computing; perception equipment; multimodal data; Deep Learning

References

1. [1] Peng Mingzhi, Xu Yao, Hu Yongbo, etc Intelligent Inspection Technology for Secondary Equipment in Substations Based on Artificial Intelligence Technology [J]. High Voltage Technology, 2023, 49 (S1): 90-96
2. [2] Zhang Ji, Ma Ye, Zhang Ronghua, etc Framework Design and Key Technology Research of Digital Twin Substation [J]. Engineering Science and Technology, 2023, 55 (6): 15-30
3. [3] Yin Shengli Research on Artificial Intelligence based Substation Inspection Robot [J]. Automation Application, 2024, 65 (15): 42-44
4. [4] Wang Hao, Yan Hao, Ye Hairui, etc Intelligent Inspection of Photovoltaic Power Plants Based on Drones [J]. Infrared Technology, 2022, 44 (5): 537-542
5. [5] Yuan Tongqing, Xi Peng Quality evaluation algorithm based on multimodal audio and video fusion [J]. Journal of Shenyang University of Technology, 2022, 44 (3): 331-335
6. [6] Aston, Zachary, Li Mu, etc Hands on Deep Learning PyTorch Edition
7. [M] Translated by He Xiaoting and Rui Chao'er Beijing: People's Posts and Telecommunications Press, 2024: 572
8. [7] Xin Yuanyuan, Niu Jun, Xie Zhijun, etc Overview of Microservice Architecture Implementation Framework [J]. Computer Engineering and Applications, 2018, 54 (19): 10-17
9. [8] Pei Chuanfu, Cui Kai, successful Visual Recognition of Car Dumper Hook Removal Based on YOLO-V2 Algorithm [J]. China Equipment Engineering, 2024, 03 (3): 161-1623
10. [9] ZHANG R，XIE C，DENG L W. A Fine-Grained Object Detection Model for Aerial Images Based on YOLOv5 Deep Neural Network [J].Chinese Journal of Electronics，2023，32（1）：51-63.
11. [10] Hao Shuai, Ma Ruize, Zhao Xinsheng, etc Glass insulator self explosion fault detection algorithm based on super-resolution deep residual network [J]. High Voltage Technology, 2022, 48 (5): 1817-1825
12. [11] Chen Guanghua, Ge Mengying, Huang Baiyao, etc Tilt detection method for power towers based on binocular vision [J]. Journal of Beijing Jiaotong University, 2022, 46 (2): 128-137
13. [12] CHEN L C，PAPANDREOU G，KOKKINOS I，et al. Semantic Image Segmentation With Deep Convolutional Nets and Fully Connected CRFs [J/OL].arXiv:1412.7062 [cs.CV].（2014-12-22）.https://arxiv.org/abs/1412.7062.
14. [13] CHEN L C，PAPANDREOU G，KOKKINOS I，et al. DeepLab: Semantic Image Segmentation With Deep Convolutional Nets, Atrous Convolution, And fully Connected CRFs [J].IEEE Transactions on Pattern Analysis and Machine Intelligence，2018，40（4）：834-848.
15. [14] Qi Zihao, Tong Jie, Zhang Zhonghao, etc A fault voiceprint identification method for power transformers based on multi granularity knowledge features and Transformer networks [J/OL]. Chinese Journal of Electrical Engineering, 2023: 1-13 [2400-08-20] http://kns.cnki.net/kcms/detail/11.2107.TM.20231204.1422.004.html.

Funding

New Energy Intelligent Production Management Platform Project of China Huadian Group Co., Ltd. (CHDER/ZLJL1208-2022/A/0)

About the authors

Li Yang (1989-), male, Han ethnicity, from Baotou, Inner Mongolia, engineer, bachelor's degree, research direction: electrical engineering, intelligent inspection, power informatization; Qiu Jianpei (1992-), male, Han ethnicity, from Longyan, Fujian, engineer, bachelor's degree, research direction: automation, communication network technology; Song Kun (1983-), male, Han nationality, from Zibo, Shandong, is a senior engineer and master's student. His research interests include power informatization, big data, and artificial intelligence.