학술논문
Feature Extraction of Electromagnetic Signals from Photovoltaic Modules of Black Piece Recognition Using SVD and Wavelet Packets
Document Type
Conference
Author
Source
2023 International Conference on Sensing, Measurement & Data Analytics in the era of Artificial Intelligence (ICSMD) Sensing, Measurement & Data Analytics in the era of Artificial Intelligence (ICSMD), 2023 International Conference on. :1-6 Nov, 2023
Subject
Language
Abstract
With the advancement of clean energy, silicon crystal photovoltaic (PV) modules have emerged as a major player of new energy generation due to their ability to convert sunlight into electricity. During the power generation process, PV panels emit low-frequency electromagnetic signals, which is worth paying attention to these parameters. In this paper, a method for extracting features from these low-frequency electromagnetic signals is proposed. Taking into consideration the sensor's sensitivity, acquisition frequency, and sampling time, this method employs Singular Value Decomposition (SVD) to enlarge the noise's Signal to Noise Ratio (SNR). During data analysis, the SNR of the original signal was improved from 13.58 dB to 24.53 dB. Furthermore, based on a thorough analysis of the low-frequency electromagnetic signals, this study also employs wavelet packet energy extraction to analyze the frequency domain signals. This technique effectively decomposes and reconstructs the low-frequency signals, allowing for the extraction of energy features from each frequency band. Experimental results indicates the effectiveness of this approach in accurately recognizing low-frequency electromagnetic waves. Additionally, the method proves proficient in determining the power generation status of PV modules.