Neuro controller design for aircraft auto-landing under actuator failure and severe wind condition

Abstract

In this paper, a robust Neural Network (neuro) aided H2 control scheme is developed for a nonlinear F16 aircraft in auto-landing phase under actuator failure and severe wind conditions. In this scheme, a dynamic Radial Basis Function network called Minimal Resource Allocating Network (MRAN) that incorporates a growing and pruning strategy, is utilized to aid H2 controller using a feedback-error-learning mechanism. Specifically, the performance of this neuro-H2 controller for an aircraft auto-landing is studied and compared with H2 control schemes. Simulation studies show that the performance obtained by the neuro-H2 controller scheme is better than H2 controller.