欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
对于混入色噪声的混合信号，如果可以通过测量得到产生色噪声的白噪声，对白噪声进行非线性训练即可逼近色噪声，达到非线性滤波的目的。自适应模糊推理系统（adaptive nemo-fuzzy unference system，ANFIS）可以实现上述非线性逼近。文中在上述算法的基础上，提出一种EMD（empirical mode decomposition）．ANFIS的自适应色噪声消除方法，首先对混合信号进行EMD分解，得到各个内禀模态函数分量（intrinsic mode function，IMF），然后对分解得到的内禀模态分量进行ANFIS模糊消噪，最后对消噪后的各个分量信号进行叠加。由于所得内禀模态函数为近似平稳信号，且图形越来越趋于平缓，减小了ANFIS方法的逼近难度。在混合信号信噪比为2．8407dB时，经过EMD-ANTIS消噪后的估计误差比只经过ANFIS消噪后的估计误差减少11．74dB，证明EMD-ANFIS方法的有效性。
英文摘要：
Useful signal is contaminated by colored noise which is formed from white noise, if the white noise could be measured, the colored noise would be approached by nonlinear training of white noise. This is a method of noise cancellation, and ANFIS（adaptive neuro-fuzzy inference system） could meet the demand. An adaptive noise canceling method named EMD（empirical mode decomposition）-ANFIS was presented. The method disassembled the mixed signal into some IMFs（intrinsic mode function） firstly, then these IMFs were de-noised by ANFIS separately, and the results could be added to form estimation of useful signal. It made the nonlinear approximation become easier because the IMFs were stationary nearly and placid. A mixed signal whose SNR was 2. 8407 dB was dealt with by ANFIS method and EMD-ANFIS method, the estimation error of the latter was 11.74 dB less than the former. This shows the validity of EMD-ANFIS method.
同期刊论文项目
同项目期刊论文
期刊信息
位置：