相关量测噪声下的修正动态随机逼近算法

A Modified Dynamic Stochastic Approximation Algorithms under Correlated Case with ARMA Measurement Noise

将Ruppert 提出的模型推广到带有 ARMA 量测噪声的多维情形。提出了修正动态随机逼近算法,并证明它在 C?aro 意义下的(a,?)收敛性。最后还考虑了特殊情形下算法的(a,?)收敛性。

In this paper,the model given by Ruppert is extended to multidimensional case with ARMA measurement noise. A modified dynamic stochastic approximation algorithm is presented and which Convergence in Cesaro average sense is proved.