对应于矩阵反演公式及 QR分解定理 ,加权最小二乘法 ( WRL S)存在两类不同的算法结构。基于 QR分解的结构中 ,建立于 Givens Rotation正交变换的算法优点在于 :并行算法结构易于实时在线计算 ;采用的信息融入与剔除形式易于扩展与调整。集元辨识 ( Set membership)法中的最优界椭球算法 ( Optimal bounded ellipsoid,OBE)采用优化策略与有效数据评价准则 ,实现了冗余数据的过滤并具有较理想的时变参数跟踪潜力。OBE辨识算法与 WRL S方法间存在紧密的联系 ,但在基于矩阵反演公式的加权最小二乘法上所获得的 OBE递推算法结构并不能充分利用其自适应跟踪潜力。本文在 Givens Rotation正交变换基础上建立相应的 OBE递推算法结构 ,并充分利用指示时变参数变化的监测因子 ,进一步引入自适应调整策略。仿真结果显示 ,该方法具有较佳的时变参数跟踪效果。 Corresponding to the matrix inversion formula and the QR decomposition theorem, there are two kinds of different algorithm structures for the weighted least square method (WRL S). Among the structures based on QR decomposition, the advantages of the algorithm based on Givens Rotation orthogonal transformation lie in that the structure of parallel algorithm is easy to calculate online in real time; the adopted information fusion and elimination form can be easily expanded and adjusted. Optimal bounded ellipsoid (OBE) in Set membership method adopts the optimization strategy and effective data evaluation criteria to achieve the filtering of redundant data and has better tracking potential of time-varying parameters. There is a close relationship between the OBE recognition algorithm and the WRL S method, but the OBE recursive algorithm structure obtained by the weighted least square method based on the matrix inversion formula can not fully utilize the potential of the adaptive tracking. Based on the Givens Rotation orthogonal transform, this paper constructs the corresponding OBE recursive algorithm structure, and makes full use of the monitoring factor which indicates the change of time-varying parameters to further introduce the adaptive adjustment strategy. Simulation results show that this method has a better time-varying parameter tracking effect.