Robust multi-H2 output-feedback approach to aerial refuelling automation of large aircraft via linear matrix inequalities

Claase, Etienne H. (2013-03)

Thesis (MScEng)--Stellenbosch University, 2013.

Thesis

ENGLISH ABSTRACT: In recent years the aviation industry has shown an interest in the airborne refuelling of large transport aircraft to enable increased payload mass at take-off and to extend aircraft range. Due to the large volume of fuel to be transferred, a boom and receptacle refuelling system with a larger fuel transfer rate is employed. The refuelling operation is particularly difficult and strenuous for the pilot of the receiver aircraft, because the position of the receptacle relative to the tanker aircraft must be maintained within a narrow window for a relatively long period of time. The airborne refuelling of a large aircraft is typically much more difficult than that of a fighter aircraft, since the large aircraft is more sluggish, takes much longer to refuel, and has a relatively large distance between its refuelling receptacle and its centre of mass. These difficulties provide the motivation for developing flight control laws for Autonomous In-Flight Refuelling (AIFR) to alleviate the workload on the pilot. The objective of the research is to design a flight control system that can regulate the receptacle of a receiver aircraft to remain within the boom envelope of a tanker aircraft in light and medium turbulence. The flight control system must be robust to uncertainties in the aircraft dynamic model, and must obey actuator deflection and slew rate limits. Literature on AIFR shows a wide range of approaches, including Linear Quadratic Regulator (LQR), μ-synthesis and neural-network based adaptive control, none of which explicitly includes constraints on actuator amplitudes, actuator rates and regulation errors in the design/synthesis. A new approach to designing AIFR flight control laws is proposed, based on Linear Matrix Inequality (LMI) optimisation. The relatively new LMI technique enables optimised regulation of stochastic systems subject to time-varying uncertainties and coloured noise disturbance, while simultaneously constraining transient behaviour and multiple outputs and actuators to operate within their amplitude, saturation and slew rate limits. These constraints are achieved by directly formulating them as inequalities.

AFRIKAANSE OPSOMMING: Die lugvaart industrie toon huidiglik ’n belangstelling in die brandstof oordrag tussen twee groot vervoervliegtuie gedurende vlug, met die doel om die maksimum opstyggewig kapasiteit sowel as die maksimum ononderbroke vlugafstand vermoë van die hervulde vliegtuig te vermeerder. ’n Boom hervulling-stelsel word geïmplementeer om die hoë spoed van brandstof oordrag te voorsien. Die verrigting van vluggebonde hervulling van ’n groot, trae vliegtuig is moeiliker en meer veeleisend as bv. van ’n vegvliegtuig, veral vir die vlieënier van die hervulde vliegtuig, wat sy boom-skakel moet reguleer binne ’n relatiewe klein boom bewegingsruimte vir ’n relatiewe lang tydperk. Die kinematika betrokke speel ook ’n groter rol in ’n groot hervulde vliegtuig a.g.v. die langer afstand tussen die boom-skakel en die massa middelpunt/ draaipunt. Hierdie bied die motivering om ’n beheerstelsel te ontwikkel wat die taak outomaties uitvoer. Die doel van die navorsing is om ’n beheerstelsel te ontwerp wat die boom-skakel van die hervulde vliegtuig outomaties reguleer binne die bewegingsruimte van die boom, gedurende ligte en matige turbulensie. Daar word van die beheerder vereis om robuust te wees teen onsekerhede in die vliegtuig se meganika, sowel as om die beheer oppervlaktes en turbines van die vliegtuig binne hul defleksie-, wringkrag- en sleurtempo-perke te hou. Daar bestaan reeds ’n groot verskeidenheid van benaderings tot die outomatisering van luggebonde hervulling, onder andere LQR, μ-sintese en neurale-netwerk gebaseerde aanpasbare beheer, waarvan geeneen perke op aktueerders en regulasie foute direk in die ontwerp insluit nie. ’n Nuwe benadering word voorgestel wat gebaseer is op Linear Matrix Inequality (LMI) optimering. Die LMI tegniek is relatief nuut in die gebruik van beheerstelsel ontwerp. Dit stel die ontwerper in staat om ’n stogastiese stelsel, onderworpe aan tydvariante-stelsel-variasie en gekleurde ruis versteurings, optimaal te reguleer, terwyl aktueerders en stelsel gedrag direk beperk word.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/80195
This item appears in the following collections: