Energy Reports (Nov 2022)
Comparative study of two typical one-third octave algorithms in substation noise analysis
Abstract
The one-third octave band (1/3-octave-band) is the most important output item in noise spectrum analysis, which can reflect the frequency characteristics of the noise. At present, the most used 1/3-octave-band algorithms in engineering are the digital bandpass filter bank method and the fast Fourier transform (FFT) analysis method. To compare and evaluate the applicability of these two typical algorithms in power grid environmental noise dominated by low-frequency noise, this paper conducts detailed simulation test research from relative attenuation test, performance test, and prototype test at the AC UHV base of China Electric Power Research Institute. It is found that the FFT analysis method has advantages in suppressing spectrum leakage, thereby ensuring the accuracy of spectrum analysis, but the relative attenuation value of the spectrum at low frequency (31.5 Hz) and high frequency (5000 Hz) cannot meet IEC 61260-1-2014 required class 1 requirements, and when analyzing the noise emitted by the UHV base, the analysis results of some bands have a large deviation compared with BK connect software, and the maximum is -6.09 dB. In contrast, no matter whether the center frequency is low or high, the relative attenuation of the spectrum of the digital bandpass filter bank method can always meet the provisions of IEC, and when analyzing the noise emitted by the UHV base, the analysis result is consistent with the BK connect software as well, indicating that the digital bandpass filter bank method has better applicability when applied to the low-frequency noise scenes.
