摘要
介绍了噪声抵消的原理和从强噪声背景中自适应滤波提取有用信号的方法。利用模糊逻辑和RBF神经网络的等价性将模糊逻辑和神经网络有机的结合来构成模糊神经网络,并对BP神经网络、RBF神经网络和模糊神经网络三种基本自适应算法进行了对比研究。计算机模拟仿真结果表明,这几种算法都能通过有效抑制各种干扰来提高强噪声背景中的信号检测特性。相比之下,模糊神经网络算法具有良好的收敛性能,除收敛速度快于BP神经网络算法和RBF神经网络算法以及稳定性强外,而且具有更高的起始收敛速率,更小的权噪声,更大的抑噪能力。
The theory of noise canceling and the method for extracting desired signal from strong background noise using adaptive filtering are described. The configuration of fuzzy neutral network is also presented by combining fuzzy logic and neutral network based on their equivalence. And then BP neural network algorithm, RBF neural network algorithm and fuzzy neural network algorithm are compared. Computer simulation results show that all of these adaptive algorithms can improve the detection of weak signal in strong background noise. In comparison, the performance of fuzzy neural network algorithm is much better than those of BP neural network algorithm and RBF neural network algorithm. Besides much faster convergence speed, the behavior of fuzzy neural network filtering coefficients is much more stable; and fuzzy neural network has faster initial convergence rate, lower maladjustment noise and better robustness against noise and disturbance.
出处
《电子测量与仪器学报》
CSCD
2008年第4期57-62,共6页
Journal of Electronic Measurement and Instrumentation
基金
国家863课题资助项目(编号:2001AA602018-04)
关键词
噪声抵消
自适应滤波
模糊逻辑
模糊神经网络
noise cancellation, adaptive filtering, fuzzy logic, fuzzy neural network.
