基于三阶累积量的闭环辨识.pdfVIP

35 8 上 海 交 通 大 学 学 报 Vol. 35 No. 8 2001 8 JOURNAL OF SHANGHAI JIAOT ONG UNIVERSIT Y Aug. 2001 : 1006-246 ( 2001) 08-1136-05 莫建林, 　张卫东, 　许晓鸣 ( , 200030) : 将高阶累积量对具有高斯分布特性 ( 白色或有色) 随机噪声 强烈抑制特性, 用于辨识开 环条件下 干扰, 一些文献中提出了基于高阶累积量 改进均方辨识准则( M SE) . 针对三阶累积 量 情形, 探讨了该准则在噪声干扰下闭环辨识中 适用性. 理论分析表明, 扩展到闭环情形下 M SE 辨识准则等价于无噪声干扰下对象处于开环情形时 均方误差辨识准则. 该方法可获得处于 闭环运行状态下对象 渐进无偏参数估计; 当对象采用ARMA 模型时, 可获得相应 线性递推辨 识算法. 仿真实验证明了该算法 有效性. : 累积量; 闭环辨识; 高斯噪声; 无偏参数估计 : T P 13: A Closed Loop Ident if icat ion Based on Three Order Cumulant - , - , - M O Jian lin ZH A N G Wei dong X U X iao min ( Dept. of Automation, Shanghai Jiaotong Univ . , Shanghai 200030, China) Abstract: Based on the fact that high order cumulant can strongly suppress the effect of such stochastic processes as ( perhaps colored) Gaussian noises and symmetrically distributed noises, a novel modified M SE criterion w as proposed in an open loop case w here input and output are contaminated by noises. Re- stricted in three order cumulant, the paper analysed the applicability of the criterion in noisy closed loop. It show s that the novel criterion used in the noisy closed loop enviroment equals to the effect of MSE esti- matiuon. Also, w hen ARMA model is adopted, the correspoding online recursive algorithm can be achieved. T he simulation results confirm the efficiency of the method. : ; ; ; Key words cumulant closed loop identification Gaussian noises unbiased parameter estimation , ( SNR) . [ 5, ] , , ( MSE) : [ 1, 2] . , ,