快速采样数据建模的最小二乘算法

由于传统的ＡＲ模型不适于快速采样数据的建模，提出了基于增量差分算子建模的递推最小二乘算法，讨论了这种模型与相应连续模型的关系．数据仿真表明这种模型较之于ＡＲ模型有较好的适用性．

A Least Square Algorithm for Modelimg Fast-Sampled Data

Fast-sampled data processing is becoming ever-increasingly important in modern ystemapplications,such as wideband communication and digital feedback control.The convention-al AR model is not suitable for fast-sampled data modeling. It has been proved that when thesampling interval is short enough,the estimates of the parameters in an AR model are de-pendent only on the order of the model and have nothing to do with the characteristics of thecorresponding continuous signal system. In order to overcome this shortcoming,an alterna-tive model based on incremental difference operator as well as the recursive least squaremethod is proposed.The relation between the model and the continuous signal model is dis-cussed.It is concluded that the parameters of the discrete model are a good approximation ofthose of the continuous signal model when the sampling interval is short enough.Numericalsimulation shows that the model proposed is more useful than the AR model.