Nonlinear Worst-Case Analysis of an LPV Controller for Approach-Phase of a Re-Entry Vehicle

In this paper, a nonlinear robustness analysis of an LPV controller for the approachphase of a re-entry vehicle is presented. The nonlinear longitudinal equations of motion of the NASA-HL-20 atmospheric re-entry vehicle, a benchmark provided by Deimos Space as a representative of future re-entry vehicles, constitute the open loop model. The analysis is carried out using the optimization-based worst-case analysis tools developed at University of Leicester for Phase I of the European Space Agency (ESA) project – “Robust LPV Gain Scheduling Techniques for Space Applications”. The tools make up an analysis framework using several optimization methods such as local gradient based algorithms, global evolutionary algorithms, dividing rectangles algorithm, hybrid local / global evolutionary algorithms and multi-objective algorithms. In this paper, the worst-case deviations from a predefined re-entry profile due to simultaneous variations of multiple uncertain parameters are determined by two optimization methods hybrid differential evolution and hybrid dividing rectangles. The results demonstrate the flexibility, efficiency and reliability of the optimization-based worst-case analysis, and project it as a useful potential tool for complex controller validations in future space applications.