Scribd
Upload
111 views50 pages

EMT 3200 Group #4 Curve Fitting

This document discusses nonlinear curve fitting. It begins by introducing the topic and defining nonlinear regression as fitting a curve to data using a nonlinear equation of the form Y= f(X, β) + E. It then lists the group members and references used. The rest of the document covers various methods for nonlinear curve fitting, including linearization, nonlinear least squares, the Gauss-Newton algorithm, and the Levenberg-Marquardt algorithm. It provides examples and applications in fields like electrical, mechanical, and civil engineering.

Uploaded by

elio
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
111 views50 pages

EMT 3200 Group #4 Curve Fitting

This document discusses nonlinear curve fitting. It begins by introducing the topic and defining nonlinear regression as fitting a curve to data using a nonlinear equation of the form Y= f(X, β) + E. It then lists the group members and references used. The rest of the document covers various methods for nonlinear curve fitting, including linearization, nonlinear least squares, the Gauss-Newton algorithm, and the Levenberg-Marquardt algorithm. It provides examples and applications in fields like electrical, mechanical, and civil engineering.

Uploaded by

elio
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 50

1

Faculty of Engineering and Technology


NON LINEAR CURVE FITTING
ENGINEERING MATHEMATICS V
EMT 3200
LECTURER: MS. E. TRIM
GROUP # 4
GROUP MEMBERS 2

Names: Usi #:
 Terrence De Peana  1013445 (Electrical)
 Carlos Moore  1026905 (Electrical)
 Hardeo Paramsook  1027131 (Electrical)
 Godwin Cameron  1007719 (Electrical)
 Oral Torrington  1005749 (Electrical)
Reference 53
 

 Sematech. “NON LINEAR LEAST SQUARES REGRESSION”. Retrieved from:

http://www.itl.nist.gov/div898/handbook/pmd/section1/pmd142.htm

on 2019/02/15

 Gurley. Numerical Methods Lecture 5 – Curve fitting Techniques. CGN 3421 Lecture Notes. Retrieved

From:

http://web.iitd.ac.in/~pmvs/courses/mel705/curvefitting.pdf on 2019/02/12

 KaleidaGraph. The KaleidaGraph Guide to Curve Fitting. Retrieved From:

http://www.snergy.com/Tools/curvefitting.pdf on 2019/02/20
Outline 4

• INTRODUCTION

• AIM

• CURVE FITTING

• METHODS

• APPLICATIONS

• CONCLUSION
Introduction 5

 CURVE FITTING IS THE PROCESS OF SPECIFYING THE


MODEL THAT PROVIDES THE BEST FIT FOR A SERIES
OF DATA POINTS.
 CURVE FITTING CAN BE DONE EITHER BY LINEAR AND
NONLINEAR REGRESSION; LINEAR TAKING THE FORM
Y = A + BX, WHILE NONLINEAR TAKES THE FORM Y = F
(X, Β) + E.
Aim 8

 TOEDUCATE YOU THE STUDENTS OF EMT


3200 ON “CURVE FITTING; THE NONLINEAR
APPROACH” AND THE METHOD USED TO
OBTAIN THESE NONLINEAR FITS.
Nonlinear Curve Fitting 9

 Nonlinear regression uses nonlinear regression


equation, which takes the form:

Y= f (X, β) + E
What is it used for 10

 
 FIND THE GENERIC EQUATION THAT REPRESENTS
EVERY POINT ON THE CURVED DATA SET.

 SMOOTH THE DATA OBTAINED.

 IMPROVE THE APPEARANCE OF YOUR PLOT.


Methods Used in Non‐linear 21

Curve Fitting
 LINEARIZATION

 NON‐LINEAR LEAST SQUARES


 GAUSS NEWTONIAN ALGORITHM

 LEVENBERG‐MARQUARDT ALGORITHM
Linerization 12

 
THREE COMMON EXPRESSIONS ARE FOUND IN
REAL TIME:

 Exponential: y = a ebx
 Power: y = axb
 Saturation growth:
Linearization 13

  following manipulations must first be made to the


The

models in order to linearize the data:


 Exponential: ln y = ln a + bx
 Power: ln y = ln a + blnx
 Saturation growth: = +
Linearization of Exponential 24

  Consider

a function to fit data of the general
exponential form

STEP 1: Take logarithm of both sides to linearize


the equation
Linearization of Exponential 15

 STEP 2: The data Z now takes the form

Z = a0 + a1 W

 STEP 3: Use linear regression formula method to


find the constants of regression
Linearization of Power Expression 16


  Considera function to fit data of the general
exponential form

STEP 1: Take logarithm of both sides to linearize the


equation
Linearization of Power Expression 17

 STEP 2: The data Z now takes the form

Z = a0 + a1 W

 STEP 3: Use linear regression formula method to


find the constants of regression
Linearization of Saturation Growth 18


  Considera function to fit data of the general
exponential form
Y=

STEP 1: Take reciprocal of both sides to linearize the


equation
= +
Linearization of Saturation Growth 19

 STEP 2: The data Z now takes the form

Z = a0 + a1 W

 STEP 3: Use linear regression formula method to


find the constants of regression
29
Linearization
Linearization 30

 
Example:

Fit an exponential curve of the form


to the following data points:

             

X 0.4 0.8 1.2 1.6 2.0 2.3

             

Y 800 975 1500 1950 2900 3600


Linearization 31

 Solution:

             

W 0.4 0.8 1.2 1.6 2.0 2.3

             

Z 6.685 6.882 7.313 7.576 7.972 8.189


Calculations for linear regression method 23

N = 6 (number of data points)


∑ w = 0.4 + 0.8 + 1.2 + 1.6 + 2.0 + 2.3 = 8.3
∑ z = 6.685 + 6.882 + 7.313 + 7.576 + 7.972 + 8.819 =
45.247
∑ (w. z) = (0.4) (6.685) + (0.8)(6.662) + (1.2)(7.313) +
(1.6 )(7.576) + (2.0 )(7.972) + (2.3 )(8.189) = 63.856
∑ w2 = 0.42 + 0.82 + 1.22 + 1.62 + 2.02 + 2.32 = 14.09
ŵ = 8.3/ 6 = 1.383
Ź = 45.247 / 6 = 7.541
Calculations for linear regression method 24
    a1=

 a0 = Ź - a 1 ŵ

 a1=

 a1 = 0.485
 a0 = 7.541 – 0.485 (1.383) = 6.870
 Therefore:
Z = 6.87 + 0.485 W
 ln a = e6.87
 Therefore:
 y = 962.949 e0.485X
41
Least Squares Method
 Consider the general form for Least Square
 

Method:
Curve Fitting Polynomials with Least 42
Squares
  

 Consider
 for the data set

Time 1 2 4 5 8
Years 3 4 6 11 20
Curve Fitting Polynomials with 27

Least Squares
   =
Curve Fitting Polynomials with Least 28
Squares

  The formula for the least square objective function is

 The gradient of is
Curve Fitting Polynomials with 29

Least Squares

  This can be Written as

 Hence
Curve Fitting Polynomials with Least 30
Squares
  The

Hessian matrix is

 If
 Therefore,
Newton-Gauss Algorithm 49

  
 The Newton-Gauss algorithm is used to
complete least squares problem that are
difficult to solve.

 The iterative formula for the Gauss-Newton


method is given below:
Newton-Gauss Algorithm 32

  

When the value of p is computed it is added to


the initial guess of the x values and the system
of equations are repeated until there recurrence
or the value of the error is almost 0.
Newton-Gauss Algorithm 33

Works well if
 the initial guess is close to the actual value
 The model fits the data to represented well

Gives an approximation of the value. If this does


not work another algorithm is needed
Levenberg‐Marquardt Algorithm 50

The Levenberg‐Marquardt algorithm is a


iterative procedure which combines the Gauss-
Newton method and the steepest descent
method. The algorithm works well for most
cases and become the standard of nonlinear
least square routines
Levenberg‐Marquardt Algorithm
50

•  Gauss

Newton:
• Fast convergence but sensitive to the starting
location.

• Gradient Descent: The opposite.


Levenberg‐Marquardt Algorithm
50
Levenberg‐Marquardt Algorithm
50
Levenberg‐Marquardt Algorithm
50
Levenberg‐Marquardt Algorithm 39


  Inthis method the second term of the Hessian is matrix
is also considered.

 Where is a scalar. The search direction is obtained by


solving the linear system.
(
Applications 40

 Electrical Engineering – Modelling


performance of semi-conductors, Model
designs and Data transfer.
Applications (Electrical) 41

 Develop a second order polynomial curve for


values of the Amplitude and Noise in a Data
transfer system.

Amplitude 0.0 0.0 1.0 1.5 2.0 2.5


(V)
Noise 0.0674 -.09156 1.6253 3.0377 3.3535 7.9409
Applications 42
Applications 43
Applications 44
Applications 45
Curve Fitting Polynomials32
with Least Squares
Applications 47

 Mechanical Engineering - Prediction of


equipment’s ability to penetrate rocks during
boring.
Applications 48

 Hydraulics – Modelling of streamflow in a channel


48
Application (Mechanical)
  The

torque, T newton metres, required to
rotate shafts of different diameters, D
millimetres, on a machine is shown below. If
the law is , where and are constants, draw a
suitable graph and hence determine the
values of a and n.
D 7 10 18 25 40

T 0.974 1.71 4.33 7.28 15.3


Application (Civil) 50

The
   pressure P and volume V of a mass of gas
in a container are related by the law , where n
and c are constants. From the values given
below plot a suitable graph and hence
determine the values of n and c.

V 4.60 7.20 10.1 15.3 20.4 30.0

P 14.2 7.59 4.74 2.66 1.78 1.04


51
Conclusion

 Nonlinear regression uses nonlinear regression equation, which


takes the form:
Y= f (X, β) + E

 Curve fitting nonlinear techniques will produce an equation that


can be used to find points anywhere along the curve. It is also
used to smooth the curve and improve the appearance of the
plot.
Conclusion 52

 Curve fitting techniques such as Linearization,


Nonlinear Least Squares Method or Levenberg –
Marquardt Algorithm are used to obtain the best fit
curve model for a set of data points collected.

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy