Continuous-Time Dynamic Model Identification Using Binary-Valued Observations of Input and Output Signals

In the paper an approach to the identification of linear continuous-time dynamic systems based on acquired binary-valued observations of their input and output signals is presented. In the presented approach the binary-valued observations are interpreted as disturbed measurements of the corresponding original continuous-time input and output signals and system models are estimated using classical linear dynamic system estimation methods. The properties of such the approach are illustrated with three simulation examples. In these examples, frequency responses are identified using the empirical transfer function estimator, parametric models are recovered from the frequency responses identified, and finally, parametric models are identified directly from the binary-valued observations acquired.