有色噪声补偿下降阶系统状态的最优估计

在大型空间结构的控制系统设计中,降阶系统的状态估计是非常重要的环节,它直接关系到控制器的性能。针对B.Friedland的工作,本文提出了有色噪声补偿模型,在恒等变换下显式地消去了有色噪声项,利用正交投影定理仅对感兴趣的状态分量导出了无偏、最小方差估计。这组递推估计式与采用增广状态方法相比,不仅在处理容量、计算效率上有很大改进,而且是标准Kalman滤波方程的推广。注意到原系统与降阶系统之间的内在联系,本文采用随机系统分析方法,给出了在高斯、马尔可夫假设下,有色噪声模型特征参数的在线识别技术。数字仿真表明,对偏差补偿起主要作用的是有色噪声补偿模型中的驱动白噪声项,这是对B.Friedland工作的发展。

Optimal State Estimate of Reduced Order System through Colored-Noise Conpensation Model Approach

It is shown that the state estimate of reduced order system is playing an important role in control system design of large space structure .which influences the control per- formance sharply. A colored-noise compensation model is proposed with respect to B, Friedland's paper. The colored-noise item is apparently eliminated by the identity transformation,and the unbiased minimum variance estimation towards a part of the state components is obtained by means of the orthogonal projection theorem. Compared with the state argument method ,the iterative equations are not only an improvement on capacity and efficiency,but also a generalization of Kalman filter equations. Noticing the coherent relationship between the reduced order system and the initial system,the on-line realization technique of characteristic parameters is also described through stochastic analysis. Numerical simulation shows that the driving white noise of compensation model plays a main role,and this is a development of B. Friedland's work.