论文部分内容阅读
应用时域上的现代时间序列分析方法 ,基于自回归滑动平均 (ARMA)新息模型和白噪声估值器 ,应用控制理论中的极点配置原理 ,对线性离散时间广义随机系统提出了极点配置广义稳态Kalman估值器 .它们具有全局渐近稳定性 ,且通过配置估值器的极点可按指数衰减速率使初始状态估值的影响快速消失 .它们可在统一框架下处理滤波、平滑和预报问题 .它们避免了Ric cati方程和最优初始状态估值的计算 ,因而可减小计算负担 .一个仿真例子说明了它们的有效性
Based on the ARMA model and the white noise estimator and the principle of pole assignment in control theory, a generalized definition of the pole configuration is proposed for the linear discrete-time generalized stochastic systems by using the method of modern time series analysis in time domain. Steady-state Kalman estimators, which have global asymptotic stability and can rapidly disappear the impact of initial state estimates at exponential decay rates by configuring the poles of the estimator, which can handle filtering, smoothing and prediction in a unified framework Problem, which avoids the computation of the Riccati equation and the optimal initial state estimate, thus reducing the computational burden. A simulation example illustrates their validity