Robust modal filtering for control of flexible aircraft

Aeroservoelasticity;Distributed sensing;Modal filtering;Flutter;Shape control;X-56A;Simulation;Robust regression

Suh, Peter M. ; Mavris, Dimitri N. Aerospace Engineering Schrage, Daniel P. Brenner, Martin J. Burken, John J. Hodges, Dewey H. ; Mavris, Dimitri N.

University:Georgia Institute of Technology

Department:Aerospace Engineering

关键词: Aeroservoelasticity;    Distributed sensing;    Modal filtering;    Flutter;    Shape control;    X-56A;    Simulation;    Robust regression;   

Others  :  https://smartech.gatech.edu/bitstream/1853/51780/1/SUH-DISSERTATION-2014.pdf

美国|英语

来源: SMARTech Repository

【 摘 要 】

The work in this dissertation comprises aeroservoelastic simulation development, two modal filter design case studies and theoretical improvement of the modal filter. The modal filter is made robust to sensor bias. Studies have shown that the states estimated by the modal filter can be integrated into active structural control. The integration of modal filters into aircraft structural control systems is explored.Modal filters require distributed sensing to achieve accurate modal coordinate estimates. Distributed sensing technology has progressed to the point, where it is being tested on aircraft such as Ikhana and the upcoming X-56A. Previously, the modal filter was criticized for requiring too many sensors. It was never assessed for its potential benefits in aircraft control. Therefore it is of practical interest to reinvestigate the modal filter.The first case study shows that under conditions of sensor normality, the modal filter is a Gaussian efficient estimator in an aeroservoelastic environment. This is a fundamental experiment considering the fact that the modal filter has never been tested in the airflow.To perform this case study a linear aeroservoelastic code capable of modeling distributed sensing is developed and experimentally validated. From this code, a computational wing model is fitted with distributed sensing. A modal filtering design methodology is developed and applied. With distributed sensing and modal filtering feedback control is achieved. This is also compared and contrasted with a controller using state-of-the-art accelerometers. In addition, new methods of active shape control are introduced for warping an aeroelastic structure utilizing the modal filter and control surfaces. The next case study takes place in a realistic setting for an aircraft. Flexible aircraft bring challenges to the active control community. Increased gust loads, possibility of flutter, and off-design drag may detrimentally affect performance and safety. Aeroservoelastic tailoring, gust load alleviation (GLA) and active flutter suppression (AFS) may be required on future flexible air vehicles. It is found that modal filters can theoretically support these systems.The aircraft case study identifies additional steps required in the modal filtering design methodology. Distributed sensing, the modal filter and modal reference shape control are demonstrated on the X-56A flutter-unstable simulation model. It is shown that control of deformations at potentially millions of points on an aircraft vehicle can be achieved through control of a few modal coordinates. Finally modal filter robustness is theoretically improved and computationally verified. State-of-the-art modal filters have high bias sensitivity. In fact, this is so critical that state-of-the-art modal filters may never be certified for aircraft implementation. This is especially true within a flight critical control system. The solution to this problem is found through derivation of the robust modal filter. The filter combines good properties of concentration algorithms with robust re-descending M-estimation. A new trim criterion specific to the strain based modal sensing system is derived making the filter robust to asymmetric or leverage point outliers. Robust starts are introduced to improve convergence of the modal estimation system to the globally optimal solution in the presence of 100s of biased fiber optic sensors.