Continuous-Time System Identification Using Indirect Inference

Abstract

Identification of continuous-time systems typically present problems due to the facts that one cannot, in general, measure the time derivatives of the signals and, also, the sampled nature of the data. We utilise indirect inference as the underlying principle for continuous time system identification. Indirect inference has been widely used in the econometrics area for time series modeling. Here we adapt the indirect inference technique to include systems with an exogenous input and apply it to the problem of system identification. We use an example problem posed by Rao and Garnier to show the effectiveness of the indirect inference technique when contrasted to other continuous-time methods of identification.