1 / 13

Non-linear least square curve fitting method

Non-linear least square curve fitting method. Amey Modak, Yan zhu University of Washington. Introduction to least squares. Fitting a mathematical relationship to observations Approximate solution to over determined systems Least squares – the most commonly used method

taipa
Télécharger la présentation

Non-linear least square curve fitting method

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Non-linear least square curve fitting method Amey Modak, Yan zhu University of Washington

  2. Introduction to least squares • Fitting a mathematical relationship to observations • Approximate solution to over determined systems • Least squares – the most commonly used method • Minimizes sum of residuals

  3. Linear and Non-linear least squares • Depends on model function. • Linear least squares has residuals that are linear in all unknowns. • NLS have dependent on independent variable and the parameters. gradient equations do not have a closed solution. • This makes NLS an iterative method

  4. Gauss-Newton method • Function • Sum of squares of Residual • Minimization: • have dependent on x and β. • This requires initial guess

  5. Gauss-Newton Cond… • Improve parameter matrix iteratively • Linearize model after every iteration • Gradient equation: • Rearranging:

  6. Problem definition • Material: copper • Dimensions = 1m X 1m X 0.01m • Thermal conductivity = 400 W/m-k • Rho = 8960 kg/m3 • Specific heat = 386 J/kg-k • Steffen Boltzmann constant = 5.670373e-8 • Convective heat coefficient = 1 • Ambient temperature = 300 K • Emissivity = 0.1

  7. Data Generation

  8. Gauss-Newton Solution • Model Function • Residual • Jacobians for each step:

  9. Gauss-Newton Solution Cond… • Initial guess from Matlab lsqcurvfit [500; 4; 500; 0.1] • Use improved parameter matrix to iterate further • Continue iterations until (with prescribed error) • [528.17, 3.988, 469.16681, 0.07617905]

  10. Outliners Correction

  11. Comparison with linear model

  12. Conclusion • incredibly useful tool for analyzing sets of data • Converges faster if initial guesses are in the range • Robust method is useful to eliminate significant errors • Linear extrapolation for non-linear models gives significant errors • will not converge if the initial guesses are not in a suitable range

  13. Thank You…

More Related