Obtaining multivariable continuous-time models from sampled data

Abstract

While most physical systems or phenomena occur in continuous-time, identification methods based on discrete-time models are more widespread among practitioners and academic community, possibly due to the discrete-time nature of the data records. There has been a growing interest in estimating continuous-time (CT) models in the last decade. This work develops algorithms to estimate the parameters of multivariable state-space CT models from input-output samples using a method based on the recently developed MOLI-ZOFT approach. The performance of the algorithm is evaluated using real data from an industrial winding process. © 2017 American Automatic Control Council (AACC).