论文部分内容阅读
结合1/f噪声信号功率谱随频率成反比变化的关系,以及稀疏分解可以根据信号灵活构造原子库的特点,提出一种基于稀疏分解估计大功率半导体激光器1/f噪声的新方法,构造了具备1/f噪声特点的过完备库.在该过完备库中通过Matching Pursuit算法完成了白噪声与1/f噪声混叠信号的稀疏分解.实验结果显示:该方法估计出淹没在白噪声环境中1/f噪声的γ参数,与频谱分析仪的测量结果有较好的一致性,通过对比不同的过完备库证明了所构造的过完备库的优越性.
Combining with the relationship between the power spectrum of the 1 / f noise signal and the inversely varying frequency of the 1 / f noise signal and the sparse decomposition, a new method of estimating the 1 / f noise of high power semiconductor laser based on the sparse decomposition can be proposed. An overcomplete library with 1 / f noise characteristics.The sparse decomposition of the white noise and 1 / f noise aliasing signal is completed by the Matching Pursuit algorithm in the overcomplete library.The experimental results show that this method estimates the submergence in the white noise environment The γ parameter of 1 / f noise is in good agreement with that of the spectrum analyzer. The superiority of the constructed overcomplete library is proved by comparing different overcomplete libraries.