State estimation based optimal control and NARMA-L2 controllers of a scaled-model helicopter

Abstract

Scaled-model helicopters are highly nonlinear, coupled, and unstable machines. They have fast response and controlling them is very complicated and need high degree of precision. In this paper, a detailed nonlinear model is derived. An optimal controller that is based upon state estimation is designed and implemented for each of the control inputs using a linearized model around an unaccelerated hovering motion. The optimal linear controller was then applied to the nonlinear model. In addition, input/output measurements of the nonlinear model were used to train the Multi-Layer Neural Networks (MLNNs) of the Nonlinear AutoRegressive with Moving Average (NARMA-L2) controller. Then, NARMA-L2 was applied to the nonlinear model and the results were compared with both the classical, namely PI-D, and the optimal control.