Deterministic versus evolutionary optimisation methods for nonlinear robustness analysis of flight control laws

This paper considers the application of two different global optimisation approaches to the problem of analysing the robustness (flight clearance) of nonlinear flight control systems. The analysis employs a typical nonlinear clearance criterion used by the European aerospace industry together with a detailed simulation model of a high performance aircraft with a full authority control law. The deterministic optimisation algorithm used in the study is Dividing RECTangles (DIRECT), while the evolutionary algorithm is Differential Evolution. Both algorithms are hybridised with local gradient- based optimisation methods to improve convergence rates near the global solution. The reliability, computational complexity and efficiency of the two approaches are compared for this realistic engineering example, and the prospects for application of optimisation-based methods in the industrial flight clearance process are discussed.