Optimal low order model identification of fractional dynamic systems

This article deals with modeling, simulation, identification and model reduction of non-integer systems in time domain. The fractional order system simulation is based on a fractional integrator operating on a limited spectral range. It allows approximating the fractional system by an integer state-space representation of high dimension. An output error method is used to perform the model parameters identification including the fractional order. Non-integer order systems known to exhibit long memory behavior require high dimensional models to represent them. In this paper, an iterative non-linear programming method is used to perform the approximation reduction and numerical simulations illustrate the efficiency of the reduced model to recover the fractional order system characteristics.