Power quality compound disturbance identification based on knowledge distillation and RP-MobilenetV3

被引：0
|
作者：
He C. [1 ]
Li K. [1 ]
Dong Y. [1 ]
Song Z. [1 ]
Xiao X. [1 ]
Li B. [1 ]
Li X. [1 ]
机构：
[1] School of Electrical and Electronic Engineering, Huazhong University of Science and Technology, Wuhan
来源：
Dianli Xitong Baohu yu Kongzhi/Power System Protection and Control | 2023年 / 51卷 / 14期
基金：
中国国家自然科学基金;
关键词：
deep residual shrinkage network; image; knowledge distillation; MobileNetV3; power quality disturbances; recurrence plot;
D O I：
10.19783/j.cnki.pspc.221856
中图分类号：
学科分类号：
摘要：
To solve the problems of complex feature extraction, low recognition accuracy and model weight reduction in the recognition of complex power quality disturbance (PQD), a method of PQD signal visualization using recurrence plot (RP) and a model training method based on knowledge distillation are proposed. This method first mines the hidden features of PQD signals based on RP and builds image data sets. It then uses a deep residual shrinkage network (DRSN) to extract deeper features of image data sets and complete classification. Finally, based on knowledge distillation (KD), it lets the trained DRSN guide the training of the lightweight network MobilenetV3, and realizes cross network transmission of knowledge through distillation. The simulation experiment and hardware experiment show that MobileNetV3 trained by knowledge distillation can achieve high-precision and lightweight composite disturbance recognition, and the accuracy can be improved by 1.06% in a 30 dB noise environment. It has good recognition effect on actual disturbance signals and good noise robustness. © 2023 Power System Protection and Control Press. All rights reserved.
引用
收藏
页码：75 / 84
页数：9
相关论文
共 24 条
  • [1] YI Huijuan, GAO Yunpeng, ZHU Yanqing, Et al., Recognition of composite power quality disturbance based on improved incomplete S transform and LOO-KELM algorithm, Electric Power Automation Equipment, 42, 1, pp. 199-205, (2022)
  • [2] CAICEDO J E, DANIEL A M, EDWIN R T, Et al., A systematic review of real-time detection and classification of power quality disturbances, Protection and Control of Modern Power Systems, 8, 1, pp. 30-66, (2023)
  • [3] HE Guoqing, WANG Weisheng, LIU Chun, Et al., Study on technical standard of distributed resources grid integration, Electric Power, 53, 4, pp. 1-12, (2020)
  • [4] RUAN Zihang, XIAO Xianyong, HU Wenxi, Et al., Multiple power quality disturbance classification feature optimization based on multi-granularity feature selection and model fusion, Power System Protection and Control, 50, 14, pp. 1-10, (2022)
  • [5] NIU Gang, ZHANG Kai, GUO Xiangfu, Et al., Complex power quality disturbance identification based on Gramian angular field and graph convolution neural network, China Measurement, 48, 7, pp. 169-176, (2022)
  • [6] HE Caijun, LI Kaicheng, YANG Wangwang, Et al., Power quality compound disturbance identification based on dual channel GAF and depth residual network, Power System Technology, 47, 1, pp. 369-379, (2023)
  • [7] XU Liwu, LI Kaicheng, LUO Yi, Et al., Classification of complex power quality disturbances based on incomplete S-transform and gradient boosting decision tree, Power System Protection and Control, 47, 6, pp. 24-31, (2019)
  • [8] HAN Y, FENG Y, YANG P, Et al., An efficient algorithm for atomic decomposition of power quality disturbance signals using convolutional neural network, Electric Power Systems Research, (2022)
  • [9] WANG Yan, Review of research development in power quality disturbance detection, Power System Protection and Control, 49, 13, pp. 174-186, (2021)
  • [10] CHEN Zhengying, WANG Liming, YI Yong, Et al., Computation of radio interference excitation current of DC corona based on short-time Fourier transform, High Voltage Engineering, 45, 6, pp. 1866-1872, (2019)
123