摘要
高斯白色噪声序列在科学研究与工程领域得到广泛的应用,如系统辨识与仿真,电气与通信工程、生物医学工程等。然而,在已有的商业软件包中难以找到确实能产生标准高斯白噪声序列的程序段。该文提出了在计算机上产生标准高斯白噪声序列的新方法,可有效地产生N(0,1)分布并具有良好白色性能的高斯随机序列,其计算方法主要有两个部分组成:首先应用改进的Marsaglia-Bray方法产生标准正态分布的随机序列;然后,在均方误差最小的准则下,应用双随机交换最小化方法对序列进行白化处理。所得序列的幅值分布与给定的理论正态分布相一致,且序列具有优良的白色化随机性能。
Gaussian white noise sequences are extensively used in the scientific research and the fields of engineering such as system identification and simulation, electrical and communication engineering, and biomedical engineering, etc.. However, it is hardly to find such a piece of program in known commercial software packages really suitable to generate standard Gaussian white noise sequences. A novel method for generating standard Gaussian white noise sequences on computers is presented in this paper. It can be effectively used to generate Gaussian stochastic sequences with distribution and excellent white characteristics. The presented computational method mainly consists of two steps, namely a modified Masarglia-Bray algorithm which is used for generating a standard normally distributed stochastic sequence and Secondly, a double stochastic interchange minimization algorithm which is used to whiten the Gaussian sequence under the criterion of minimum mean-square error. With this method, as a result, the sequence has an amplitude distribution which is highly consistent with the specified theoretic normal distribution and excellent whitened stochastic characteristics.
出处
《中国电机工程学报》
EI
CSCD
北大核心
2004年第12期207-211,共5页
Proceedings of the CSEE