On identification of Volterra kernels of nonlinear systems by separating kernel slices

作者最近发展了一种利用伪随机M序列获得非线性系统Volterra核的新方法。在该方法中，最重要的问题是如何分离出现在互相关中的重叠的核切片。研究了两种分离核切片的方法：一种是选择合适的M序列，另一种是幅度变化法。我们得到了以下结果：(1)考虑到当输入M序列的幅值变化时，n阶Volterra核输出的幅值变化，得到了分离重叠的Volterra核片的方法。(2)将该方法应用于一个化工过程，并于2002年6月在熊本举行的AdCONIP会议上公布了结果。(3)该方法也在2002年8月在中国大理举行的APCCM‘2000会议上发表。(4)2003年9月在英国考文垂举行的ICSE上介绍了非线性系统辨识的一般思想。(5)将该方法应用于非线性模型预测控制，结果在韩国庆州举行的ICCAS’03会议上公布。(6)该方法在非参数非线性模型预测控制中的应用发表在Krean J.Chem.Eng。2004年，该方法被认为在非线性工业控制系统中具有广泛的适用性。

The authors have recently developed a new method for obtaining Volterra kernels of nonlinear system by use of a pseudorandom M-sequence. In this method, the most important problem is how to separate overlapped kernel slices which appear in the crosscorrelation. The authors have investigated two methods for separating those kernel slices : one is the suitable selection of M-sequence, and the other is amplitude variation method. We have obtained the following results.(1)On considering the fact that when the amplitude of the input M-sequence is changed, the amplitude of the output from n'th order Volterra kernel varies, we have obtained the method for separating overlapped Volterra kernel slices. The simulation shows that this method is useful.(2)This method is applied to a chemical process, and the result was presented at AdCONIP held in Kumamoto in June 2002.(3)This method was also presented at APCCM'2000 held in Dali, China in August 2002.(4)The general idea of identification of nonlinear system is presented at ICSE held in Coventry, UK, in September 2003.(5)This method is applied to Nonlinear Model Predictive Control, and the results are presented at ICCAS'03 held in Gyoungju, Korea, in October 2003.(6)The application of this method to Nonparametric Nonlinear Model Predictive Control appeared in Krean J.Chem.Eng. in 2004This method is considered to be widely usable in nonlinear industrial control systems.