摘要
本文首先将传统意义下的不相关及其与协方差函数之间的关系推广到n(n≥2)阶不相关及其与高阶累积函数之间的关系。以此为基础,讨论了非最小相位线性随机系统的可辨识性及辨识方法。将可辨识性条件从独立同分布放宽到n(n>2)阶不相关。对用熵函数进行辨识的方法做了较详细的研究并给出了可辨识性定理。
In this paper the identifiability and identification algorithms for the non-minimum phase linear system with unmeasurable stationary input are discussed. The uncorrelation fill n-th order called n-th order uncorrelation is defined, where n is greater than or equal to two. Its retation with commulants till n-th order is proven. This is the generalization of the relation between the traditional uncorrelation and covariance. Based on these definitions and the relations, the original contraint of the input of mutual independence and identical distribution is removed for the identifiability of the system. The entropy is introduced and the identifiability theorem is given when the entropy function is used to identify the system.
Simulation is made and it verifies the correctness of theory and algorithms given in this paper.
出处
《自动化学报》
EI
CSCD
北大核心
1993年第4期404-412,共9页
Acta Automatica Sinica
关键词
系统辨识
断续系统
平稳输入
Stochastic systems
system identification
deconvolution.
