HYBRID GLOBAL OPTIMIZATION TECHNIQUE FOR OUTPUT-ERROR IDENTIFICATION OF FRACTIONAL ORDER MODEL - APPLICATION TO AERO GAS TURBINE PLANT

Abstract

A hybrid global optimization technique for identifying an output-error fractional order model is proposed. The proposed technique combines local and interval global optimization methods. It is capable of estimating the fractional derivative orders and the parameters, with guaranteed convergence. The technique is applied to identify a fractional order model for a twin-spool aero gas turbine plant. The input variable is the fuel rate to the gas turbine and the output variable is the high pressure (HP) spool speed. The operating regime is at 90% HP spool speed demand. It is found that the fractional order model identified with the proposed technique gives a much better fit to the temporal data than higher integer order models.