时滞和滤波联合辨识的递推最小二乘算法

Recursive least squres Algorithm for Joint Delay-Filter Identification

时滞和滤波联合辨识问题既是经典时滞估计问题的推广，又是自适应系统建模和时灌估计两方面的交叉。本文针对先滤波后时滞的系统模型，从方法上改进BOUDREAU＆KABAL提出的基于快速横向滤波器的递谁最小二乘滤波算法，使其时间复杂度由O［19ρ］下降为0［7ρ］，便于实时在线应用。我们以低通滤波器与线性时滞串联系统的辨识为例，表现该算法对变化时滞的跟踪能力及联合辨识性能。

The problem of jonit delay-filter identification is concerned with conventional time-delay es-timation, modeling of adaptive spstem and time-delay estimation This paper deals with system model of first filterthen time-delay, improved the algorithm of recursive least squares filter baased on fast across filter form Boudrfau& Kabel, the time complexity is redused from 0[19ρ] to 0[7ρ]. This method will convenient be used in real-timeenvironment. In this paper the authors gave example on spstem identification of low pass filter and lineartime-delay serial spstem, so that exhibits track abilirty for variance of time delay and property for jointidentification of this algorithm.