Abstract
Input-output relation of a nonlinear dynamic system can generally be described by Volterra functional series. The Volterra functional series is a generalization of the convolution integral conventionally used in linear systems. However, when the system is not self-regulatory, the integration interval of the functionals must be infinite and thus these functionals are not suitable for practical use. Therefore, in this paper, the Volterra functional series are transformed into those with finite integration interval. The Volterra kernels in the transformed series are constructed so as to include the effect of the unknown input signal in the past. Application of deterministic signals to both the system and the transformed Volterra functional series model and evaluation of the responses lead to the determination of Volterra kernels which represent the dynamic characteristics of the system.
This method can be applied to the system whose topological structure is not a priori known. The degree of approximation of this method can be arbitrarily specified. This method is also valid for the systems with self-regulation.
