摘要
应用离散子波变换原理 ,对信号进行谱估计 ,提出了基于离散子波变换的特征提取方法 ,并对实测数据进行特征提取 ,获得了稳定有效的线谱和连续谱 。
Feature extraction is crucial to exact reconstruction of underwater radiated noise. We present a new method for feature extraction. Our method is based on discrete wavelet transform. We use equations (9) and (10) to perform spectral estimation. With these equations we obtain higher frequency resolution than Fourier transform. It can also effectively remove the random part of real measurement data through statistic average and improve the precision of line spectral feature extraction. According to the recognition principle of line spectrum, the location, amplitude and width of line spectrum can be decided. Then we put forward a method for spectral separation, which is described in subsection 2.2. With such a method, we can separate line spectrum from continuous spectra. Simulation result of spectral separation (Fig.4) shows that our method is effective. Our method has met with preliminary success when applied to the research of underwater acoustic synthetic environment.
出处
《西北工业大学学报》
EI
CAS
CSCD
北大核心
2000年第2期250-253,共4页
Journal of Northwestern Polytechnical University
基金
国家教委博士点基金! ( 97CJ0 3 0 1)
国家重点实验室基金资助! ( STL -6-98)
关键词
辐射噪声
谱估计
子波变换
水声设备
radiated noise, feature extraction, spectral estimation, wavelet transform
