具有测量时滞的不确定系统的递推鲁棒滤波

Recursive robust filtering for uncertain systems with delayed measurements

针对模型参数随机不确定性和测量时滞现象,提出了一种基于垂直投影定理和新息分析的递推鲁棒滤波方法。该方法首先将原系统转化成一个含有跨时相关性噪声的随机不确定性系统,接着重新计算了新息,然后用数学归纳法证明所得到的新息是一个互不相关的序列,最后用新息分析和垂直投影定理获得了所期望的递推鲁棒滤波器。该滤波器设计方法不需要将跨时相关性噪声进行状态增广,因而计算量小、精度高。仿真实验验证了该方法的有效性。

A recursive robust filtering method based on the orthogonal projection and an innovation analysis approach is proposed for stochastic uncertain systems with delayed measurements.First,the original system is transformed into a stochastic uncertain system with correlated measurement noises,and the innovations are recalculated.Then,by applying the mathematical induction,the newly obtained innovations are proofed to be uncorrelated with each other.Finally,the desired recursive robust filter is designed via an innovation analysis approach and the orthogonal projection theorem.Without resorting to state augmentation,the proposed filter treats the noise correlated across time directly.Therefore,the desired filter has less computational burden,but a high computational accuracy.Simulation results demonstrate the effectiveness of the proposed approach.