摘要
作为时间序列模型的一种 ,AR模型由于参数估计和定阶简单而广泛应用于系统辨识。在多维AR序列的最小二乘建模的基础上 ,结合卡尔曼滤波算法 ,推导了应用卡尔曼滤波技术的多维AR序列参数估计方法以及加入衰减因子后的卡尔曼滤波算法。该算法不需要保存历史数据 ,在得到新的“观测”数据后可以对AR模型的估计参数进行实时改正。在确定AR模型阶数时 ,提出了快速F检验法 ,大大减少了建模过程中的计算工作量 。
AR series, as one of time series models,is applied broadly in system identification because its parameter estimation and rank decision are simple.On the basis of multi dimension AR series modeled by least square criterion and the Kalman filtering technique,the method for estimation of parameters of multi dimension AR series by Kalman filter is developed in this paper. The method with attenuation foctor is also derived. It is not necessary to keep the historical data for these methods.Thanks to application of Kalman fileter, the estimated parameters of AR series can be updated real time by newly observed data.The Fast F test method proposed in this paper can decrease much modeling calculation work in the decision of rank of AR series and have an applicable value.
出处
《大地测量与地球动力学》
CSCD
2003年第2期92-95,共4页
Journal of Geodesy and Geodynamics
基金
教育部地球空间环境与大地测量重点实验室基金资助项目 ( 0 2 -0 9-0 1)
