+ All Categories
Home > Documents > FVSysID ShortCourse 8 Software

FVSysID ShortCourse 8 Software

Date post: 07-Mar-2016
Category:
Upload: anonymous-ry7aem
View: 130 times
Download: 10 times
Share this document with a friend
Description:
fly vehicle identification

of 18

Transcript
  • AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006 Software/1Dr. Ravindra Jategaonkar

    Software for Flight Vehicle System Identification

    ESTIMA / FITLAB / FVSysIDat DLR

  • AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006 Software/2Dr. Ravindra Jategaonkar

    This page is left intentionally blank.

  • AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006 Software/3Dr. Ravindra Jategaonkar

    MMLE3: Modified Maximum Likelihood Estimation: A general Fortran program for Maximum Likelihood Parameter Estimation (1980)NASA Dryden, Edwards, CA (Richard Maine and Ken Iliff)

    pEst: Parameter Estimation (1987)interactive parameter estimation for nonlinear dynamic systems.NASA, Ames/Dryden (James Murray and Richard Maine)

    ESTIMA: Integrated Software tool for parameter estimation (1983 / 1998) and simulation of dynamic systems.DLR Institute of Flight Systems, Braunschweig, Germany (Ravindra Jategaonkar)

    IDEAS: Integrated Data Evaluation and Analysis System (1998)Science Applications International Corporation, California, MD (Dennis Linse)

    Overview of Software Tools (1)

    Software packages used at Aerospace industries are not included:Boeing, Airbus, Embraer, ..

  • AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006 Software/4Dr. Ravindra Jategaonkar

    MATLAB based software packages:SIDPAC: System Identification Programs for AirCraft (2002)

    NASA, Langley, VA (Gene Morelli)

    FITLAB Tools box for Linear and Nonlinear parameter estimation (2001)DLR Institute of Flight Systems, Braunschweig, Germany (Susanne Weiss)

    FVSysID Flight Vehicle System IDentification: A Time Domain Methodology.(Author: Ravindra Jategaonkar) 2006, AIAA Book

    Frequency DomainCIFER: Comprehensive Identification from FrEquency Responses

    Identification of transfer-function models, Spectral signal analysis,HQ analysis.Mark Tischler, NASA Ames

    DIVA MIMO Analysis and Transfer function approximationsDLR Institute of Flight Systems, Braunschweig, Germany

    FITLAB Matlab Tools box for parameter estimation in frequency domain Linear systemsDLR Institute of Flight Systems, Braunschweig, Germany (Susanne Weiss)

    Overview of Software Tools (2)

  • AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006 Software/5Dr. Ravindra Jategaonkar

    FVSysID: Matlab based software (1)

    Part of AIAA Book: Flight Vehicle System Identification: A Time Domain Methodology

    Specially tailored software for system identification in time domain- Developed using basic Matlab (No special Toolbox required)

    - Provides source codes (m-files of all the methods / functions)

    - Provides sample flight data

    - Chapter wise sorted

    - Applicable to any general linear / Nonlinear dynamic system- Multiple run evaluation

    - Off-line algorithms:- Output Error Method (Linear / Nonlinear systems)

    - Filter Error Method (Linear / Nonlinear systems)

    - Recursive algorithms:- Recursive least squares (RLS)

    - Fourier transform regression (FTR)

    - Extended / Unscented Kalman filters (EKF / UKF)

  • AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006 Software/6Dr. Ravindra Jategaonkar

    FVSysID: Matlab based software (2)

    AIAA Book: Flight Vehicle System Identification: A Time Domain Methodology

    - Unstable aircraft:- LS, TLS, FEM, EKF, UKF- Stabilized OEM- Eigenvalue transformation- Equation Decoupling

    - Utility programs- Integration method (Runge-Kutta 2nd, 3rd, 4th order)- Differentiation of measured variables (multiple run)- Smoothing / Filtering of measured data (multiple runs)

    USER Interface

    - mDefCasexx.m Definition of model, flight data, initial values etc.- state_eq function to code rhs of state equations- obser_eq function to code observation equations

    ]),(),([)( xtutxftx =&],),(),([)( yxtutxgty =

  • AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006 Software/7Dr. Ravindra Jategaonkar

    FVSysID: Matlab based software (3)

    AIAA Book: Flight Vehicle System Identification: A Time Domain Methodology

    Example: Estimation of lift, drag, pitching moment Derivatives

    )cossin(

    )sin()cos(

    )cos()sin(

    TtzTtxy

    em

    y

    Te

    L

    Te

    D

    IF

    CIcSqq

    qmVF

    VgqC

    mVSq

    mF

    gCmSqV

    ll&

    &

    &

    &

    ++==

    +++=

    +++=

    eemo

    mqmmVmm

    LLVLL

    DDVDD

    CVcqCC

    VVCCC

    CVVCCC

    CVVCCC

    ++++=

    ++=

    ++=

    200

    00

    00

    Te

    Zzm

    Te

    Xxm

    TtzTtxy

    em

    ym

    m

    m

    m

    m

    mFC

    mSqa

    mFC

    mSqa

    IFC

    IcSqq

    qq

    VV

    sin

    cos

    )cossin(

    =

    +=

    ++=====

    ll&

    sincoscossin

    DLZ

    DLXCCC

    CCC=

    =

    State equations

    Observation equations

  • AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006 Software/8Dr. Ravindra Jategaonkar

    FVSysID: Matlab based software (4)

    AIAA Book: Flight Vehicle System Identification: A Time Domain Methodology

    Example: Estimation of lift, drag, pitching moment Derivatives

    The unknown parameter vector

    TemmqmmVmLLVLDDVD CCCCCCCCCCC ][ 000 =

    qV ,,, zx aaqqV ,,,,,, &

    ee F,

    States: State equation: xdot_TC04_hfb_lon

    Observations: Obs. Equations: obs_TC04_hfb_lon

    Inputs:

    test_case = 4; % Integer flag for the test casestate_eq = xdot_TC04_hfb_lon; % function for state equations obser_eq = obs_TC04_hfb_lon; % function for observation equationsNx = 4; % Number of states Ny = 7; % Number of observation variablesNu = 2; % Number of input (control) variablesNparSys = 11; % Number of system parameters

    See m-files

  • AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006 Software/9Dr. Ravindra Jategaonkar

    FVSysID: Matlab based software (5)

    AIAA Book: Flight Vehicle System Identification: A Time Domain Methodology

    Example: Estimation of lift, drag, pitching moment Derivatives

    Run the test case:Output Error method:/FVSysID/chapter04/ml_oem: test_case = 4

    Filter Error method/FVSysID/chapter05/ml_fem: test_case = 4

    Recursive estimation methods (EKF, UKF, UKFaug)/FVSysID/chapter07/mainRPE: test_case = 4

    Several test cases and sample flight data provided

  • AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006 Software/10Dr. Ravindra Jategaonkar

    Integrated Software Tool forSystem Identification and Simulation of Dynamic Systems

    Standardmodels(FPR,Lon,Lat,6DOF,2nd)

    ModelingUsermodels

    (completely flexible)

    Commonenvironment

    Sys Id andsimulation7Optimization methods6IntegrationformulasLargescale systems.....

    Nucleus

    Results

    Modeling, estimation and simulation

    Onlineinteractivemode withgraphicaldisplay

    Flight DataCDFor

    ASCII

    Logfile:estimates,correlation,standard deviation,...

    Timehistories:cdf,Ascii

    Convergence

    Linearized system

    Eigenvalues/vectors

    Scripts:Utility,SysID,Plotprograms(Commontoallplatforms)

    CompletelyconfigurableUnix/X-windowsbased

    "xfiles"

    Front end and command proceduresAvailability:Fourplatforms-IBM-AIX,SG-IRIX,SunOS-Solaris,Linux

    Graphic:Gnuplot/GVorGINO

    Licensed software:-No.ofconcurrentuser-Industry,Researchorg./Univs.

    Support:Demoprograms,Initialfamiliarization,shortcourses

    Commercial aspectsCONPLOT:Convergence plots

    (time and frequency domain)

    EIGPLOT:Eigenvalues/vectors BODEPLOT:Frequency response

    THPLOT:Timehistory plots CRSPLOT:Crossplots

    Post processing

    PSDPLOT:Powerspectral density

    ESTIMA (1)

  • AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006 Software/11Dr. Ravindra Jategaonkar

    Front End

    Scripts:Utility,SysID,Plotprograms(Commontoallplatforms)

    CompletelyconfigurableUnix/X-windowsbased

    "xfiles"

    Command procedures

    ESTIMA (2)

  • AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006 Software/12Dr. Ravindra Jategaonkar

    Standardmodels(FPR,Lon,Lat,6DOF,2nd)

    ModelingUsermodels

    (completely flexible)

    Commonenvironment

    Sys Id andsimulation7Optimization methods6IntegrationformulasLargescale systems.....

    Nucleus

    Results

    Modeling, estimation and simulation

    Onlineinteractivemode withgraphicaldisplay

    Flight DataCDFor

    ASCII

    Logfile:estimates,correlation,standard deviation,...

    Timehistories:cdf,Ascii

    Convergence

    Linearized system

    Eigenvalues/vectors

    Core

    ESTIMA (3)

  • AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006 Software/13Dr. Ravindra Jategaonkar

    User Interface

    ModelingUser models

    (completelyflexible)

    Models

    Standard models(FPR,Lon,Lat,6DOF,2nd)

    Statevariables:p, r,, , vControlinputs:a, r, 3, 4, f, ....Observations: ay,,p,r,, , ayc

    Keyword: mod_sub = lat

    44330 lClCrrlCaalCrlrCplpClClClC +++++++=

    44330 nCnCrrnCaanCnCrnrCpnpCnCnCnC ++++++++= &&

    44330 YCYCrrYCaaYCrYrCpYpCYCYCYC +++++++=

    Fixed structureControl via Input file

    Program control fileChannelNos.forinputandoutputvariables

    A/cmasscharacteristics

    Aerodynamicparameters

    ESTIMA (4)

  • AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006 Software/14Dr. Ravindra Jategaonkar

    Output error method

    Optimization MethodsUnconstrainedGauss-NewtonBounded-VariableGauss-NewtonLevenberg-Marquardt

    Jacob'sHeuristicsearchEXTREMConjugatedirectionsPOWELL

    SIMPLEX(Nelder andMead)SUBPLEX(ROWAN)

    Integration MethodsEulerRunge-Kutta 2ndorderRunge-Kutta 3rdorderRunge-Kutta 4thorderRunge-Kutta-Fehlberg withstepsizecontrolBackwarddifferentiationformula

    SysIdandsimulation7Optimizationmethods6IntegrationformulasLargescalesystems.....

    Nucleus

    HybridizationofoptimizationandIntegrationmethods

    Large scale systems30States,60outputs,60inputs,80timeslices,80000datapoints

    ESTIMA (5)

    Algorithmic Features

  • AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006 Software/15Dr. Ravindra Jategaonkar

    FITLAB: Toolbox for Sys ID under MATLAB (1)

    Matlab Toolbox

    Identification of nonlinear models m-Files Simulink-Models S-Functions

    Identification of linear models

    time and frequency domain

    startup algorithm

    Maximum-Likelihood cost function

    Optimization using

    bounded Gauss-Newton method

    subspace-simplex search method

    Multiple maneuvers can be evaluated together

    Graphical user interface

    FITLAB Features & Functionality

  • AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006 Software/16Dr. Ravindra Jategaonkar

    FITLAB: Toolbox for Sys ID under MATLAB (2)

    Menu-driven system identification

    Operation via mouse

    Standard plots quicklook plots for

    measured data report plots for results

    Restart option

    Generation of batch files for repeating tasks

    FITLAB GUI Graphical User Interface

  • AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006 Software/17Dr. Ravindra Jategaonkar

    ReferencesJategaonkar, R. V., Flight Vehicle System Identification: A Time Domain Methodology,Volume 216, AIAA Progress in Astronautics and Aeronautics SeriesPublished by AIAA Reston, VA, Aug. 2006, ISBN: 1-56347-836-6http://www.aiaa.org/content.cfm?pageid=360&id=1447

    Jategaonkar, R. V., ESTIMA: A Modular and Integrated Software Tool for Parameter Estimation andSimulation of Dynamic Systems - Users Manual, Version 1.0, DLR IB 111-2001/29, July 2001.

    Jategaonkar, R. V. and Thielecke, F., ESTIMA - an Integrated Software Tool for Nonlinear ParameterEstimation, Journal of Aerospace Science and Technology, Vol. 6, Issue 8, Dec. 2002, pp.565-578.

    Linse, D. J., Aircraft System Identification Using Integrated Software Tools, RTO-MP-11, Paper 11, March 1999.

    Maine, R. E., and Iliff, K. W., Users Manual for MMLE3, a general FORTRAN Program for Maximum Likelihood Estimation, NASA TP-1563, Nov. 1980.

    Murray, J. E. and Maine, R. E., The pEst Version 2.1 Users Manual, NASA TM-88280, Sept. 1987.

    Morelli, E. A., System Identification Programs for AirCraft (SIDPAC), AIAA Paper 2002-4704,August 2002.

    Tischler, M. B., CIFER version 2.2 Comprehensive Identification from Frequency Responses,NASA CP-3226, Sept. 1994.

    Weiss, S.

  • AIAA Short Course: Flight Vehicle System Identification in Time Domain, Aug. 2006 Software/18Dr. Ravindra Jategaonkar

    This page is left intentionally blank.


Recommended