Performance enhancement of a self-scheduled longitudinal flight control system via multi-objective optimization

Hugo Lhachemi (1) , D. Saussie (1) , G. Zhu (1)

1 EPM - École Polytechnique de Montréal

The present work aims at improving the performance of robust and smooth self-scheduled controllers in the framework of structured H-infinity design. The developed procedure exploits multi-model and multi-channel capabilities of a MATLAB-based tool hinfstruct in the design of robust and self-scheduled flight control systems. In this approach, both controller and gain-scheduling architectures are defined a priori and are cast into the structured H-infinity synthesis framework. By formulating the considered problem in the multi-objective optimization framework, the control design amounts then to computing weakly Pareto optimal solutions in which weighting coefficients can be determined based on physical considerations or preliminary designs. The tuning of weighting coefficients can be performed by normalizing the minimization effort over the operating domain or assigning more weight to a specific sub-domain. The proposed procedure is applied to the design of a robust and self-scheduled longitudinal flight control system. Numerical analysis and simulation show an important performance improvement.

Domaines: Systèmes et contrôle [cs.SY]

Hugo Lhachemi, D. Saussie, G. Zhu. Performance enhancement of a self-scheduled longitudinal flight control system via multi-objective optimization. 2014 American Control Conference, Jun 2014, Portland, United States. pp.1377-1383, ⟨10.1109/ACC.2014.6859147⟩. ⟨hal-03133681⟩