An algorithm for parameter estimation of twin-rotor multi-input multi-output system using trust region optimization methods

Abstract

The twin-rotor, multi-input, multi-output system is a typical lab-scale helicopter system, since it resembles behavior of helicopter systems. This experimental complex aerodynamic test rig needs precise modeling to assure satisfactory control performance. Therefore, this article is intended to provide a gray-box approach for the modeling of the twin-rotor, multi-input, multi-output system. An algorithm is proposed by dividing the system into independent subsystems that can be estimated for an excitation input or a characteristic response. The proposed algorithm uses trust region optimization methods for successive estimations. The iterative optimization methods of Levenberg–Marquardt, gradient descent active set, and interior point are compared with respect to trust region methods. The model is validated with various sets of scenarios: different inputs, different types of motion, and different operating points. Effect of operating point selection on the parameter estimation process is investigated. Time domain response comparison of the single-degree-of-freedom vertical, single-degree-of-freedom horizontal, and two-degrees-of-freedom models are acceptable.