Model identification of Time-Varying Diffusive Systems

This paper presents a characterization method based on diffusive representation for a class of linear or nonlinear time-varying diffusive systems. The system variation with time may come as a result of the own actuation over the device or as a result of an external disturbance. Experimental results for both cases are presented. This method has been tested on a prototype of the REMS thermal wind anemometer for Mars atmosphere in which the time variation is induced by wind changes in a wind tunnel. The same method is also applied to model the nonlinear charge trapping dynamics of a contactless MEMS capacitor, in which the system variation with time comes from the nonlinear dependence on the applied voltages. In both cases, the obtained state-space models are able to reproduce and predict the behavior of the devices under arbitrary excitations.