Worst-case Analysis for Fixed-wing Aircraft Automated Landing based on Counter Optimization

Abstract Automated landing of fixed-wing aircraft is inevitably affected by wind disturbances. The evaluation of operational safety is often performed using Monte Carlo methods, yet they are inefficient at identifying the worst-case scenario rendering the resulting performance boundaries unreliable. To this end, this work directly identifies the worst-case wind disturbance against the safe landing based on counter optimization. Specifically, the three-dimensional wind disturbance is treated as the decision variable to maximize a set of error metrics that quantify landing performance. These metrics include lateral deviation, longitudinal deviation, airspeed error, and touchdown sink rate. Maximizing each metric formulates an optimization problem that is solved using Particle Swarm Optimization (PSO). The worst-case wind profile for each metric is identified for a given landing reference trajectory. A comprehensive understanding of the performance boundaries is therefore formed.