随机系统多新息辨识在衰减激励下的性能

多新息辨识算法是估计系统参数的一种有效方法 ,其特点是可以克服坏数据对参数估计的影响 ,具有很强的鲁棒性。该文给出了衰减激励信号的定义 ,并在衰减激励条件下 ,利用随机过程理论 ,研究了随机系统多新息辨识算法的性能 ,给出了参数估计误差收敛时 ,衰减指数应满足的条件 ,以及算法中设计参变量的选择方法。分析表明 :若设计参变量选择为 r(t) =O(t2ε(lnt) c) ,(c>0 ) ,衰减指数满足 0≤ ε<1/ 4,则参数估计均方误差以 O 1(lnt) c 速度收敛于零

Performance of multi-innovation identification for attenuating excitation of stochastic systems

Multi innovation identification is an effective method of estimating the parameters of time invariant systems and time varying systems and can overcome the effect of bad data on parameter estimation and has strong robustness. The attenuating excitation signal is defined and the performance of multi innovation identification for stochastic systems, using stochastic process theory, is analyzed for the attenuating excitation condition. The results indicate that the convergence of the parameter estimation is guaranteed by choosing the design variable and that the mean squares parameter estimation error converges to zero when the attenuating index is less than 1/4.