Neural network based feedback error controller for helicopter

Purpose – This paper seeks to present a feedback error learning neuro-controller for an unstable research helicopter. Design/methodology/approach – Three neural-aided flight controllers are designed to satisfy the ADS-33 handling qualities specifications in pitch, roll and yaw axes. The proposed controller scheme is based on feedback error learning strategy in which the outer loop neural controller enhances the inner loop conventional controller by compensating for unknown non-linearity and parameter uncertainties. The basic building block of the neurocontroller is a nonlinear auto regressive exogenous (NARX) input neural network. For each neural controller, the parameter update rule is derived using Lyapunov-like synthesis. An offline finite time training is used to provide asymptotic stability and on-line learning strategy is employed to handle parameter uncertainty and nonlinearity. Findings – The theoretical results are validated using simulation studies based on a nonlinear six degree-of-freedom helicopter undergoing an agile maneuver. The neural controller performs well in disturbance rejection is the presence of gust and sensor noise. Practical implications – The neuro-control approach presented in this paper is well suited to unmanned and small-scale helicopters. Originality/value – The study shows that the neuro-controller meets the requirements of ADS-33 handling qualities specifications of a helicopter.