大观测噪声下双线性系统的一种小样本辨识方法-BSLKL

AN IDENTIFICATION METHOD OF BILINEAR SYSTEM WITH SMALL SAMPLES UNDER LARGE OBSERVATION NOISES

给出了在大观测噪声下双线性系统的一种小样本辨识方法-BSLKL。辨识步骤如下:首先从系统的零状态应响应推出外部描述模型,对该模型进行最小二乘估计,并实现成内部描述模型,然后通过卡尔曼平滑得到降低了噪声水平的输入-输出数据。最后,用这些数据再次进行最小二乘估计并进行实现。仿真表明了此法的有效性。

This paper describes an identification method of bilinear system with small sample under large observation noises—BSLKL. The identification is as following: at first the external description model is derived from the zero-state response of the systems, the parameters in the model are estimated through LS, and the internal description model is realized by using previous estimated parameters. Secondly, The input-output data are obtained in which the noise level are lowered through Kalman smoothing. Finally, the data are used to estimate the parameters, and realization is done again. Good results are shown by simalation examples.