Levy噪声中EMD降噪的随机共振研究

发布时间：2018-07-14 16:37

【摘要】：经验模态分解(empirical mode decomposition,EMD)降低噪声的同时也削弱信号能量,并会产生虚假信号,导致信号检测存在缺陷,针对这一问题,提出Levy噪声环境下经验模态分解随机共振检测方法。通过将含噪信号进行EMD分解,对分解后信号进行叠加取平均二次采样等处理方法,使其满足随机共振要求,利用自适应算法优化系统参数,进而使处理后信号能够在双稳系统中产生随机共振,达到精确检测的目的。理论分析及实验证明在Levy噪声下,此方法能实现同一特征指数下单频信号与多频信号检测,实验表明在单频信号信噪比为-28 dB情况下能有14 dB的提高,特征指数为1.8下多频信号5 Hz频谱幅值从311.8增加到724,10 Hz频谱幅值由138.9增加到143.2。此方法对在复杂噪声环境中降低剩余噪声能量同时,提高信号能量,减少虚假信号,相对于仅仅进行EMD分解无法判断信号成分,能更好的达到检测效果。
[Abstract]:Empirical mode decomposition (EMD) reduces the noise while also weakens the signal energy, and produces false signals, which leads to the defect of signal detection. In view of this problem, a random resonance detection method of empirical mode decomposition in Levy noise environment is proposed. Through EMD decomposition of noisy signals, the signal is carried out after the decomposition of the signal. The average two sampling and other processing methods are superposed to satisfy the stochastic resonance requirement. The adaptive algorithm is used to optimize the system parameters, and then the processed signal can produce random resonance in a bistable system and achieve the purpose of accurate detection. The theory analysis and experiment prove that the method can realize the single frequency under the same characteristic exponent under the Levy noise. Signal and multi frequency signal detection, experiments show that the signal to noise ratio of single frequency signal to -28 dB can be increased by 14 dB, characteristic index is 1.8 lower multi frequency signal 5 Hz spectrum amplitude from 311.8 to 724,10 Hz spectrum amplitude from 138.9 to 143.2., the method can reduce residual noise energy in complex noise environment and improve signal energy, Reducing false signal can achieve better detection results than EMD decomposition alone.
【作者单位】：重庆邮电大学通信与信息工程学院;
【基金】：国家自然科学基金(61371164) 重庆市杰出青年基金(CSTC2011jjjq40002) 重庆市教育委员会科研项目(KJ130524)资助
【分类号】：TN911.23

【相似文献】