摘要
基于格形滤波器的正交特性，利用稳态观测过程在赫氏（Ｈｉｌｂｅｒｔ）空间的投影，运用最小二乘式递归跟踪，本文建造了Ｄ步预测器的模型，这是对常用的１步预测最小二乘式格形滤波器模型的发展。
The model for D-step predictor is constructed based on the orthogonalizing property of the lattice filter.This model is built by using projections in Hilbert space for a stationary observation process and by implementing least square regression tracking.It is a generalized expansion of the model of least squares lattice filter for general one-step prediction.
