摘要:
P.M.Djuric等人(1992)用贝叶斯法对非平稳随机信号化为分段平稳随机信号的问题作了研究,导出了一个关于分段数、各段自回归(AR)模型阶数和各段之间分界点的优化方程,但未给出具体的解法.由于此方程非常繁琐,故本文对其解法进行了研究,推导出其中一些递归关系,使求解问题变得简化,大大减少了计算量,同时对非平稳随机信号的分段问题作了进一步讨论。
Abstract:
P. M. Djuric, et al.(1992) researched on the segmentation of nonstationary stochastic process into piecewise stationary stochastic process by Bayesian theory, and gave a dynamic equation about the number of segments, their boundaries and AR model orders for each segment, but did not give details of solution for the equation. Because the solution for the equation is very complex, this paper investigates the solution, derives some recursive relations, simplifies the problem, saves computation time and goes further into the segmentation of nonstationary stochastic process into piecewise stationary stochastic process.