Improved autolanding controller for aircraft encountering unknown actuator failures

The authors have assessed the capability of various neural-aided classical feedback controllers that have been designed for autolanding of a typical modern high performance fighter aircraft under unknown actuator failures and external wind disturbances. Analysis of the fault tolerance envelopes of these neural-aided controllers revealed that position and rate saturation of the healthy actuators resulted in loss of control and failure to complete the autolanding task. Therefore, the over-all fault-tolerance region was not simply connected and exhibited gaps. In this paper we have successfully overcome the problem of gaps in the fault-tolerance envelope of the basic feedback controller by a judicious choice of feedback variables and developing a strategy for optimal gain selection to enlarge the failure tolerance envelopes in the presence of severe winds. The controller is motivated by Nonlinear Dynamic Inversion (NDI) approach and is able to handle six different types of single / double control surface failures. This is achieved by exploiting the full capability of control allocation inherent in the redundant control surfaces. The autolanding controller discussed in this paper is the most robust controller designed so far for the benchmark autolanding problem chosen for study.