高压变电站噪声分离算法及其应用  

Separation Algorithm of the Noise in High Voltage Substation and Its Application

作　　者：彭任华 郑成诗[1,2] 丁茫[2] 李晓东[1,2] 

机构地区：[1]中国科学院声学研究所,北京100190 [2]中国科学院上海高等研究院,上海201210

出　　处：《高电压技术》2015年第5期1680-1686,共7页High Voltage Engineering

基　　金：国家自然科学基金(61201403;61302126)~~

摘　　要：为测量分析高压变电站同时存在的线路电晕噪声、电器(变压器、电抗器等)本体噪声和周边环境准平稳噪声,根据变电站噪声源的时频特性,提出一种基于梳状滤波器与小波变换相结合的高压变电站噪声分离算法。首先根据本体噪声的线谱特性,利用通带梳状滤波器滤波实现本体噪声的估计;其次利用高频小波系数构造电晕噪声检测信号实现电晕噪声与准平稳态噪声的分离。实验结果表明:噪声分离算法能有效估计出电晕噪声、本体噪声以及准平稳态噪声,噪声声压级以及A计权声压级估计误差<1 d B;不同区域的噪声组成差别较大,靠近电抗区区域电晕噪声、本体噪声、准平稳态噪声是变电站噪声的主要来源,而在变电器附近主要为本体噪声与准平稳态噪声,电晕噪声衰减较大。In order to measure and study the corona noise emitted by high voltage transmission line （HVTL）, the nou- menon noise generated by electric appliances （electric reactor, transformer, etc.） and the environmental quasi-stationary noise separately, on the basis of the time-frequency characters of the three types of noise signals, we proposed an algo- rithm for separating these noise components for high voltage substations by using the comb filter and the wavelet transformation. Firstly, a comb filter is used to extract the noumenon noise according to its line spectrum character. Se- condly, the high frequency wavelet coefficients are used to detect the corona noise, and then the corona noise and quasi-stationary noise are separated in time-frequency domain. Experimental results indicate that the algorithm is quite ef- ficient for estimating the corona noise, noumerion noise, and quasi-stationary noise, where the estimation error of the sound pressure level （SPL） and the A-weighted SPL are both less than 1 dB. The noise components are quite different in different measuring fields. The corona noise, the noumenon noise, and the quasi-stationary noise are the main noise com- ponents when the measurement point is close to the reactive zone. The corona noise is negligible when the measurement is close to the transformer due to its large attenuation in this situation.

关 键 词：小波变换 梳状滤波器 电晕噪声 本体噪声 准平稳态噪声 噪声分离 

分 类 号：TM63[电气工程—电力系统及自动化]