Real-Time Vision-Based Aircraft Vertical Tail Damage Detection and Parameter Estimation

Abstract

The need for reliable, fast, and accurate detection of fault is the key to fault-tolerant control reconfiguration of any flight control system. It is known that faults such as structural damage greatly modify the aerodynamic characteristics, often represented by stability and control derivatives. They also result in abrupt deterioration of flight performance and handling quality, which impairs safety. Since no detection technique is as reliable and fast as the Human Visual System (HVS), this paper demonstrates how popular vision algorithms such as edge detection and structural similarity index can be used for real-time detection of vertical tail damage of a fighter aircraft. A novel geometric method is proposed for fault identification using traditional mathematical geometry. Empirical techniques are used for the estimation of stability and control derivatives to generate a mathematical model of damaged aircraft instead of relying on expensive wind tunnel data. Flight mechanics analysis of the data generated confirms the effects of vertical damage in terms of reduction of static and dynamic directional stability and its control effectiveness. The parametrized state matrix thus obtained for the given flight condition thus leads to efficient real-time control reconfiguration of a damaged aircraft for safe recovery. The importance of this paper lies in using the established techniques of image processing for fault detection of damaged aircraft and parametrization for subsequent control reconfiguration, which is a novel application.