Abstract
In this paper, we propose a gray-box modeling method that incorporates physical characteristics of a dynamical system. The characteristics of the system are expressed in terms of the time-moment of the transfer function. We obtain the moment information from the pre-identification experiments. Explicitly using the moment information as equality constraints, we propose a constrained least squares method. We can construct a transfer function model that includes the desired moment information at a specified frequency. Moreover, we extend the proposed identification method to an uncertain moment-constrained case relaxing the equality constraint to inequality constraints.
