AERODYNAMIC LOAD ESTIMATION OF A MANEUVERING FLEXIBLE WING FROM STRAIN DATA

1. SUMMARY OF THESIS/PROJECT:

The project involves estimating aerodynamic loads of a maneuvering flexible wing from strain data. The whole project consists of two major parts. First, a three-dimensional finite element wing model will be constructed based on a rotating reference system (“co-rotational” formulation). This formulation will be extended to deal with 3-dimension rotational and translational motion of the wing structure. Loads-strain relationship can then be established from this finite element model, which will be used in the next part of the project. In the second part, a strategy incorporating intelligent agents will then be developed to estimate aerodynamic loads of the maneuvering flexible wing structure from strain data. The project will provide a new and more efficient approach to the estimation of aerodynamic load of a maneuvering flexible wing. In addition, the project will also produce a better understanding of wing load spectrum during maneuvers, which leads to better wing design, better fatigue and flutter assessments.

2. INTRODUCTION

Primary loads acting on an aircraft are mainly aerodynamic, structural and inertia loads. These primary loads are typically obtained from several different sources such as; wind tunnel test, flight test, computer modeling. Aircraft load estimation is a complex process, involving many assumptions and simplifications in the handling of non-linearities (geometric, material properties, etc.) and the interrelation between the forces acting on the aircraft. Alternatively, the aircraft load can also be estimated from structural strain information..

The problem of estimating load from its strain data is categorized as an inverse problem. That is, the problem of deducing system’s input or boundaries from given output responses. In performing the inverse problem of estimating aerodynamic load from strain data, several issues need to be addressed. First is the inertia effect during a maneuver. The inertia contribution to the strain becomes very significant for a flexible structure in large maneuver. In this regard, the mathematical model will be highly non-linear due to large displacements. To overcome this non-linearity, a co-rotational (Ref. 7,8,9) (rotating reference system) formulation will be developed for a maneuvering flexible wing. A full 3-D representation of the wing structure, which involves a complex rotational and translation displacement, will also be developed.

Another factor to consider is in the analysis of the inverse problem approach. Normally the strain data is directly translated to load data, based solely on ground strain calibration. However, the obtained strain, e.g. from flight test may contain significant noise, which includes both inherently random inputs, as well as deterministic but unobservable inputs. This input noise leads to potentially non-unique solutions. By incorporating prior knowledge of the system to the analysis of strain data, will increase the likelihood of getting a solution. In this project, a computational intelligent agent, specifically FUGENE (Fuzzy, Genetic and Neural Network) will be utilized.

Several authors have demonstrated the usefulness of intelligent agent in structural identification. Neural networks have been used to estimate beam loads from its strain data (ref.4,6), to estimate structural dynamic model (ref.5,1), and to estimate natural modes (ref.2). Fuzzy parameters have also been used in the analysis of flexible multi-body systems (ref.20). Genetic Algorithm has been used to solve inverse optimization problem succesfully (ref.3,14,15,19). This project will then extend the ability of intelligent agents to estimate the aerodynamic load of an aircraft wing. In particular, the research will consider ways of using the fusion of these three agents (FUGENE) to overcome the aforementioned problems of non-linearity, noise and unobservable states.

3. OBJECTIVES

To prepare a finite element co-rotational wing model in maneuvering flight. The model will be able to handle the non-linearity associated with large displacements, with little loss of accuracy or approximation. This model will be required in generating data for the development of the inverse method.
To develop an inverse method to estimate aerodynamic wing loads from strain data. Intelligent agents will be developed and used as the basis for improving the analysis of the inverse problem.

4. SCOPES

The project will cover a maneuvering flexible wing structure in subsonic flight.
The inverse approach will effectively utilize the fusion of three intelligent agents fuzzy, genetic and neural networks.
The developed model wing model will be generic and hence applicable to variety aircraft wing configuration.
Long-term aerodynamic and structural changes will be neglected.
The developed inverse approach will then be demonstrated on the PC-9 aircraft.

5. RESEARCH QUESTIONS

1. How can a co-rotational formulation be used in modeling a geometrically complex flexible wing in maneuvering flight?

How can a simplified finite element beam in planar motion be formulated and simulated to model a flexible wing in rotational and translational motion?
How can the planar model be extended to 3-D frame model?
How can a 3-D aircraft motion be included in the co-rotational formulation for a 3-D frame model?
How can different aerodynamic load spectrum be included in the formulation?
What behavior can be learnt from the derived model regarding aerodynamic load and strain relationship?

2. How can intelligent agents be incorporated to perform the inverse problem of estimating aerodynamic loads from strain data?

How can a fusion of three intelligent agents FUGENE (Fuzzy, Genetic and Neural Network) be used in the inverse problem?
How can FUGENE be applied to perform inverse problem for a planar flexible wing model?
How can FUGENE be effectively used in estimating aerodynamic load spectrum of a 3-D flexible wing undergoing a 3-D maneuver in subsonic flight?

6. RATIONALES

This research will contribute a new approach in handling strain measurements to estimate aerodynamic loads. The proposed approach will consider inertial and maneuvering effects, as well as their uncertainties, which are normally ignored in a conventional flight data analysis of this type.

The Aerospace Department at RMIT, collaborating with AMRL (Aeronautical and Maritime Research Laboratory) organizes a center of expertise (CoE) in aerodynamic loading. The center is involved in developing procedures to analyze strain data obtained from flight tests. Extensive flight test data of Pilatus PC-9 and FA-18 have been accumulated by AMRL. The flight data is abundant and hence requires an efficient and accurate system to process this data.

The proposed method will investigate and develop an approach to estimate aerodynamic load from strain data. For cases where an aircraft has a flexible wing structures experiencing significant maneuver, the inertia load on the wing becomes very significant. This has to be accounted in the post-flight data analysis. The FUGENE system, which combines flight data and other a priori knowledge, will produce a more efficient flight data analysis.

The proposed approach will then:

Provide a new and more efficient approach, in estimating aerodynamic load from strain data. The approach will account for inertia load, structural load, and non-linearity due to large rotational maneuver.
Provide a better understanding of the aerodynamic loading spectrum on a maneuvering flexible wing, which will lead to a better wing design, better structural static and dynamic (e.g. fatigue) assessments.
In regard to flight testing, the approach will be able to analyze both steady state and maneuvering strain data more efficiently, thus avoiding a more time consuming, complicated and expensive procedures.

7. METHODS

The project will be divided into two major components; forward model and inverse model developments. In the forward model development, a finite element wing model based on a co-rotational formulation will be developed. The co-rotational formulation provides a simple structural framework to be used in the computation where the rotational effect is significant (ref.7,8,9,18). MATLAB, C++, MSC-PATRAN-NASTRAN, STRAND-6 software will be used to develop and validate the model. The Aerospace Engineering Department at RMIT has licenses for all these software.

In the inverse method development, an approach will be developed to estimate aerodynamic load from strain data. Intelligent agents will be used to assist in the estimation process (ref2,10,11,12,16). The FUGENE system will be able to facilitate inferencing, learning, and optimizing strain data in estimating the aerodynamic load (ref.17). MATLAB, Neural Network toolbox, Fuzzy logic Toolbox, System Identification toolbox, MMLE3 toolbox, Signal Processing toolbox, C++ will be used for the analysis.

8. BIBLIOGRAPHIES

[1] Abhijit Mukherjee S. Rajasekaran, G. A. Vijayalakshmi Pai, (1998), Self-Organizing Neural Network For Identification Of Natural Modes, Journal of Computing in Civil Engineering, July 1998, volume 12, issue 3, p.163-164
[2] Adeli Hojjat, Shin-ling Hung, (1995), Machine Learning; Neural Networks, Genetic Algorithms, And Fuzzy Systems, John Wiley & Sons, Inc.1995
[3] Belegundu, A.D. Murthy, P.L.N., (1996), A New Genetic Algorithm For Multiobjective Optimization, AIAA-Paper 96-4180 (A96-38879), AIAA/NASA/ISSMO, Symposium on Multidisciplinary Analysis and Optimization, 6th, Bellevue, WA, Sept. 4-6, 1996
[4] Cao X., Sugiyama Y., Mitsui Y.,(1998), Application Of Artificial Neural Networks To Load Identification, Neural Network, Computers and Structures, vol.69, 1998, p.63-78
[5] Chen H.M., Yang J.C.S., Aminni F., (1995), Neural Network For Structural Dynamic Model Identification, Journal of Engineering Mechanics, volume 121, no. 12, December 1995, p. 1377-1381
[6] Chen S., Billings S.A., (1992), Neural Network For Nonlinear Dynamic System Modeling And Identification, International Journal Of Control, volume 56, issue 2, p.319-346
[7] Crisfield M.A, (1990), A Consistent Co-Rotational Formulation For Non-Linear, Three Dimensional, Beam Elements, Computer Methods In Applied Mechanics And Engineering, volume 81, 1990, p.131-150
[8] Ellkaranshawy H.A., Dokainish M.A., (1995), Co-Rotational Finite Element Analysis Of Planar Flexible Multi-Body Systems, Computers and Structures, volume 54, no.5, 1995, p.881-890
[9] Hsiao K.M., Yang R.T., (1995), A Co-Rotational Formulation For Nonlinear Dynamic Analysis Of Curved Euler Beam, Computers and Structures, Volume 54, issue 6, 1995, p.1091-1097
[10] Jadid M.N., Fairbairn D.R., (1996), Neural-Network Applications In Predicting Moment-Curvature Parameters From Experimental Data, Engineering Applications of Artificial Intelligence Volume 9, issue 3, 1996, p.309-319
[11] Johansen T.A., (1996), Identification Of Non-Linear Systems Using Empirical Data And Prior Knowledge--An Optimization Approach, Automatica, volume 32, issue 3, 1996, p.337-356
[12] Kemna A.H., Mellichamp D.A., (1995), Identification Of Combined Physical And Empirical Models Using Nonlinear A Priori Knowledge, Control Engineering Practice, volume 3, issue 3, 1995, p.375-382
[13] Komatsu S., Kobayashi H., (1978), Experimental Identification Of Aerodynamic Forces, Journal of the Engineering Mechanics Division, Volume 104, no. 4, July/August 1978, p.921-938
[14] Krishnakumar K., Swaminathan R., (1995), Solving Large Parameter Optimization Problems Using Genetic Algorithms, AIAA-Paper 95-3223 (A95-39654), AIAA Guidance, Navigation and Control Conference, Baltimore, MD, Aug. 7-10, 1995
[15] Obayashi, Shigeru, Takanashi, Susumu, (1995), Genetic Optimization Of Target Pressure Distributions For Inverse Design Methods, AIAA-Paper 95-1649 (A95-36505), AIAA Computational Fluid Dynamics Conference, 12th, San Diego, CA, June 19-22,1995, p.33-42
[16] Rovatti Riccardo, Guerrieri Roberto, (1996), Fuzzy Sets Of Rules For System Identification, IEEE Transaction On Fuzzy Systems, volume 4, no.2, May 1996
[17] Russo Marco, (1998), FuGeNeSys- A Fuzzy Genetic Neural System For Fuzzy Modeling, IEEE Transaction on Fuzzy Systems, volume 6, no.3, August 1998
[18] Ryu J., Kim H., Semyung W., (1998), A Method For Improving Dynamic Solutions In Flexible Multi-Body Dynamics, Computers & Structures, volume 66, issue 6, 1998, p.765– 776
[19] Wang Q.J., (1997), Using Genetic Algorithms To Optimize Model Parameters, Environmental Modeling and Software, volume 12, issue 1, 1997, p.27-34
[20] Wasfy T.M., Noor A.K., (1998), Finite Element Analysis Of Flexible Multi-Body Systems With Fuzzy Parameters, Computer Methods in Applied Mechanics and Engineering, Volume 160, issue 3-4, 1998, p.223-243
3.5 Research Methods subject to be taken as part of the program:

9. SENIOR SUPERVISOR'S CURRICULUM VITAE

NAME : A/Prof Pavel TRIVAILO
PRESENT POSITION/INSTITUTION : Associate Professor of Aerospace Engineering; Course Co-ordinator; Research Leader (Dynamics and Control), Department of Aerospace Engineering, RMIT.
ADDRESS (Postal): Department of Aerospace Engineering, RMIT, GPO Box 2476V, Melbourne, VIC 3001

QUALIFICATIONS :

BEng(First Class Honours) & MEngSc - Ukrainian National Polytechnical University, Kiev;
PhD (1980) - Research Institute for Problems of Strength of the Ukrainian Academy of Sciences, Kiev;
Tenure Rank of Senior Scientific Research Fellow (1991) - Supreme Accreditation Board at the Council of Ministers of the USSR, Moscow;
State Scientific and Technical Expert Examination of Inventions - The USSR National Scientific and Technical Institute of State Patent Expertise, The State Committee on Inventions and Discoveries.

MEMBERSHIP OF PROFESSIONAL ORGANISATIONS : Member of The IEAust (since 1994), CPEng

TEACHING EXPERIENCE : since 1988.
Currently lecturer responsible for the following subjects:? Vibration Theory; ? Dynamics; ? Aircraft Dynamics-1; ? Aircraft Dynamics-2; ? Aircraft Mechanisms; ? Aerospace Workshop.

PROFESSIONAL EXPERIENCE : Research and Teaching: Research Institute for Strength Problems of the Ukrainian Academy of Sciences, Kiev; Ukrainian National Polytechnical University, Kiev; Moscow "K.E.Tsiolkovsky" University of Aviation Technology; RMIT.

RESEARCH EXPERIENCE : since 1975.

EXPERIENCE IN SUPERVISION OF RESEARCH CANDIDATES: Australia: Senior supervisor of 2 PhD, 5 Master degree (1 completed) and Second supervisor of 3 PhD and 1 Master degree candidates (all current).

PUBLICATIONS, EXHIBITIONS OR PROFESSIONAL WORKS
Dr.P.Trivailo has been an author and co-author of more than 140 papers, research reports and publications, including 80 inventions on mechanisms, devices and testing machines, defended by USSR patents.
The examples of the latest publications are:

Trivailo P.M., Plotnikova L. and Wood L.A. Enhanced Parameter Identification for Damage Detection and Structural Integrity Assessment Using "Twin" Structures. (Invited Lecture). - Fifth International Congress on Sound and Vibration, University of Adelaide, SA, 15-18 December 1997. – pp.1733-1741.
Trivailo P.M., Plotnikova L. Global Damage Identification in Aerospace Structures Using "Twin" Structures Modal Method. - Proceedings of The International Modal Analysis Conference IMAC-XV Japan, Chuo University, Tokyo, Japan, 1-4 September 1997. - pp 667-675.
Trivailo P.M., Plotnikova L. A Method of Damage Identification in Vibrating Aerospace Flexible Structures: "Twin" Structures Approach. - Proceedings of The International Aerospace Congress 1997. Sydney, Australia, 25-27 February 1997. - pp. 767-783
Blanksby C., Trivailo P.M. Deployment/Retrieval, Earth's Magnetic Field Interaction and Tether Severance Modelling for the Tethered Satellite System. - Proceedings of The International Aerospace Congress 1997. Sydney, Australia, 25-27 February 1997. - pp. 81-96
Trivailo P.M. Vibrations: Theory & Aerospace Applications. Volumes 1 & 2. (The textbook for senior undergraduate and graduate aerospace students). - Melbourne: RMIT Publisher. - Feb 1998. - 595 pp.
Trivailo P.M. New MATLAB Toolbox for Aircraft Flight Dynamics. (New package of programs to analyze, simulate and animate in 3D the flight of different type aircraft). - Proceedings of the 1996 Australian MATLAB Conference. - Melbourne, January 1996.
May R.L., Trivailo P.M., Bourmistrov A.V., Connell H.J., Wood L.A. Modelling Towed Cable-Body Systems. - Proceedings of the International Aerospace Congress PICAST 2 - AAC 6 (Second Pacific International Conference on Aerospace Science and Technology & Sixth Australian Aeronautical Conference). - Melbourne, Australia, 20-23 March 1995. - pp. 145-152.
May R.L., Trivailo P.M., Connell H.J. Modification of CBAS Code to Model Deployment or Retrieval of a Towed Body. - Report No.CR 93/08, The Sir Lawrence Wackett Centre for Aerospace Design Technology, RMIT. (Prepared for ERL, Salisbury, Department of Defence Contract). - Jan. 1994. - 32 pp.

10. SECOND SUPERVISOR'S CURRICULUM VITAE

NAME : John Matthew Wharington
PRESENT POSITION/INSTITUTION : Professional Officer class 2, Maritime Platforms Division, Aeronautical and Maritime Research Laboratory (Defence Science Technology Organisation), Melbourne.
ADDRESS (Postal): 1/18 Fulton St, East St Kilda, Melbourne, VIC 3183

QUALIFICATIONS :

BEng (Hons 1) RMIT, 1992
PhD RMIT, 1998

MEMBERSHIP OF PROFESSIONAL ORGANISATIONS : None

TEACHING EXPERIENCE : Tutor to PhD students, Dept of Aerospace Engineering RMIT, 1996-1998
Tutor to undergraduate and graduate students, Dept of Business Computing RMIT, 1998

PROFESSIONAL EXPERIENCE : Assistant engineer, IPEC Aviation, 1991-1992
Engineer and researcher, Sir Lawrence Wacket Centre for Aerospace Design Technology, RMIT, 1996-1999

RESEARCH EXPERIENCE : Full-time since 1992, in fields of intelligent control, dynamics, simulation.

EXPERIENCE IN SUPERVISION OF RESEARCH CANDIDATES: None.

PUBLICATIONS, EXHIBITIONS OR PROFESSIONAL WORKS

J. Wharington and R.C. Dengate, An Algorithm for Indirect Adaptive Flight Control using Neural Networks, Fifth Australian Aeronautical Conference, September 1993.
P.W. Blythe, J. Wharington and I. Herszberg, The Development of Neural Network Techniques for the System Identification of Aircraft Dynamics, Fifth Australian Aeronautical Conference, September 1993.
J. Wharington and I. Herszberg, Autonomous Flight Vehicle Control using Associative Reinforcement Learning, contract report CR-94-20, RMIT Wackett Aerospace Centre, November 1994.
J. Wharington and I. Herszberg, Software Tools for the Development of Autonomous Control Systems, Sixth Australian Aeronautical Conference, March 1995.
J. Wharington and I. Herszberg, A Behavioural Conditioning Approach to Autonomous Flight Control Design, First Australasian Congress on Applied Mechanics, March 1996.
J. Wharington and I. Herszberg, An Embedded Technique for Discovering Implicit Trajectories of Autonomous Flight Vehicles, International Aerospace Congress 1997, February 1997.
J. Wharington and I. Herszberg, Control of a High Endurance Unmanned Aircraft, 21st Congress of the International Council of Aeronautical Sciences, September 1998.
J. Wharington and I. Herszberg, Dual Adaptive Heuristic Critic Applied to Dolphin Soaring, Second Australasian Congress on Applied Mechanics, 1999 (to appear).
L. Drack, H.S. Zadeh, J. Wharington, I. Herszberg and L. Wood, Optimal Design using Simulated Annealing in Matlab, Second Australasian Congress on Applied Mechanics, 1999 (to appear).
J. Wharington, Autonomous Control of Soaring Aircraft by Reinforcement Learning, PhD thesis, RMIT Department of Aerospace Engineering, 1998.
J. Wharington, I. Herszberg, Finite Difference Vehicular Path Planning, Journal of Navigation, Cambridge University Press, (submitted).
J. Wharington and I. Herszberg, Fast Solution of Dynamic Soaring Trajectories, Technical Soaring, (submitted).