Genetic Algorithm-Based Model Order Reduction of Aeroservoelastic Systems with Consistent States.

Area of Science:

• Aerospace Engineering
• Control Systems
• Computational Mechanics

Background:

• Aeroservoelasticity analysis and control synthesis for flexible aircraft require accurate models across wide flight envelopes.
• Traditional model reduction techniques can be inefficient and lack accuracy for complex, parameter-varying systems.

Purpose of the Study:

• To develop a model order reduction (MOR) framework for constructing linear parameter-varying (LPV) reduced-order models (ROMs) of flexible aircraft.
• To enable efficient aeroservoelasticity analysis and robust control synthesis across broad two-dimensional flight parameter spaces.

Main Methods:

• Utilized genetic algorithms for automated physical state selection and ROM generation at parameter space grid points.
• Employed balanced truncation for unstable systems combined with congruence transformation for optimal realization and state consistency.
• Developed an interpolation strategy to obtain ROMs at any flight condition.

Main Results:

• Achieved over 12x reduction in states while maintaining accurate system response prediction across all input-output channels.
• Demonstrated superior efficiency and accuracy compared to manual or empirical state selection methods.
• Validated smooth pole transitions and continuously varying gains in interpolated ROMs, confirming consistent state representation.

Conclusions:

• The proposed MOR framework effectively generates accurate LPV-ROMs for flexible aircraft aeroservoelasticity.
• The genetic algorithm-guided approach enhances efficiency and accuracy in model reduction.
• The framework facilitates robust controller synthesis and aids in novel vehicle design by providing consistent state representations.