Ece 2306 Notes Vii Jan2014

Techniques of least square adjustment
a) Method of observation equations, also adjustment of indirect observations, also method of

variation of coordinates, also the parametric method.
b) Condition equation method, also adjustment of observations only
c) General or combined case
Parametric case
Also called: adjustment of indirect observations, method of variation of coordinates
It has the following characteristics:
1. Each observation equation contains only one observation with a unit coefficient.
2. The number of equations is exactly the same as the observations
3. The equation include both observations and unknown parameters
NB: Parameters are the unknown quantities in the adjustment whose estimates are being sought
Both the condition equations and parametric models should give same results.
The observation equation in the parametric case has the general form:
+ = 1 , 2 , . . ,
For the parametric adjustment model we normally write an observation equation relating both the
observation and its residue with the unknown parameters.
In each case we need one equation for each observation and for unique solution of the unknown, the no
of equations = no of unknowns.
Usually, however, in surveying there will always be more observations than number of unknowns, which
permit us to determine the most probable value using least techniques.
Example 1
Recall the example of the distance being measured twice: l1= 15.12 m and l2= 15.14m
1 + 1
=
2 + 2
- Notice one equation for each observation l1 and l2

- The parameter is
Since the observation L are known, we rearrange eqn 4 to get:
Page | 1
1
1
=
, ..
2
2
1
1
+
2
1
1
2
. .
+ =
For n observations and u parameters, the above matrices become:

: :
: 1 :
+
( 1)
1
1
=
( 1)
( 1)
Normally L, contains known numerical values.

Since we want to carry out the adjustment on the basis of the least squares we set the condition:
=
From equation, =
So that (i) becomes =
Expanding we get:
= =
= +
To minimize , we take partial derivatives with respect to the parameters (x) and equate them to zero:
=

+

= 0

=
=
= + = 2
Combining all the terms:
= 0 + 2 = 0
= 2 + 2 = 0
2 = 2 ,
=
Page | 2
Transposing on both sides

= ,
Matrix A(nu) consist of known coefficients and is called a design matrix or configuration matrix. The
product = is referred to as Normal Equation matrix. It is a square symmetric matrix of
dimensions uu.
Example 2:
Recall the straight line example, we can put it in the form of + = , i.e y-ax-b=0
Point
3.2
4.0
5.0
1 + 1 1 = 0
1 + 3.2 2
1 = 2 + 3.2
2 + 2 2 = 0
2 + 4 4
2 = 4 + 4
3 + 3 3 = 0
3 + 5 6
1
2 1
2 + 4 1
3
6 1
2 1
4 1
6 1
3.2
4.0
5.0
3 = 6 + 5
3.2
= 4.0
5.0
2 4 6
1 1 1
W = identity since all measurements are assumed to be of equal precision.

To solve for =
, =
56 12
12 3
1
3
12
0.125 0.5
=
0.5 2.333
24 12 56
Page | 3
=
=
52.4
12.2
0.125 0.5
0.5 2.333
52.04
0.45
=
2.67
12.2
= 0.45
= 2.27
Using matrix to form normal equations

For large systems of equations it is always convenient to deal with matrices when carrying out least
squares. Let a system of normal equations be represented by the matrix notation: Ax=L+ V, where lvector of observations, V-vector of residuals, X-vector of unknowns
11
21
:
1
12
22
:
2
: 1
: 2
:
:
:
1
2
:
1
2
:
1
2
:
3
The normal equations are then formed as follows: ( ) =

The matrix( ) is always square and symmetric = N. It is always invertible (has an inverse)
=
Example 3
+ = 3.0 + 1
2 = 1.5 + 2
= 0.2 + 3
1
1 1
3.0
2 1 = 1.5 + 2
3
1 1
0.2
= +
To form the normal matrix equations:
( ) =
1
1
1
2
1
2
1 1
1
1
1
1 =
1
1
3.0
2
1
1.5
1 1
0.2
Page | 4
6 2
6.2
=
2 3
1.3
6 2
= 2 3
1 21.2
6.2
=
1.3
14 20.2
1 3 2 6.2
1 21.2
=
=
14 2 6 1.3
14 20.2
1.507
= 1.443
Example 4
= 5137 40"
= 7118 45"
= 5703 50"
The observation equations for each angle

5137 40" + 1 = 1
7118 45" + 2 = 2
5703 50" + 3 = 180 (1 + 2 )
1 = 1 5137 40"
2 = 2 7118 45"
3 = 180 1 + 2 5703 50"
1
1 0
5137 40"
1
2 + 0 1 = 7118 45"
2
3
1
1
12256 10"
+ =
( )
( )
5137 40"
7118 45"
180
12256 10"
1 0 1
0 1 1
1 0
1
1
0 1 =
0
2
1
1
=
0.901074707
0 1
1.244637923
1 1
2.145639908
2 1
1 2
0.666667 0.33333
0.33333 0.666667
=
=
( ) =
3.046714615
3.390277831
0.666667 0.33333 3.046714615

0.33333 0.666667 3.390277831
1
0.901041455
2 = 1.244635139
Page | 5
1
51 37 33.14"
=
2
7118 44.43"
3 = 180 1 + 2 = 5703' 42.43"
1 = 1 5137 40"
2 = 2 7118 45"
3 = 180 1 + 2 5703 50"
1 = 51 37 33.14 5137' 40 = 00' 6.86"
2 = 7118 44.43-7118' 45 = 00' 0.57"
3 = 180 51 37 33.14" + 7118 44.43" 5703 50" = 00' 7.57"
Condition equation adjustment model

These are also referred to as the method of correlates or the adjustment of observations only.
Conditions are enforced upon observations and their residues. The observations are adjusted by
removing random errors so that they satisfy a certain condition. No parameters are included in the
condition equations. The condition equations = the redundancy.
In this case the adjustment model takes advantage of the priori conditions that are imposed on the
observations.
Examples of condition equation adjustment:
a) Sum of angles in a polygon (n-2)180 where n = number of sides. E.g. the sum of the angles in a
plane triangle should equal 180 = + + only after removing random errors.
b) For closed traverse, E + N = 0
If you take a traverse between any two fixed stations, then the sum of the difference in coordinates =
difference in fixed points.
i
2 1 =
ii
2 1 =
c) The difference in elevation in a closed leveling loop should be equal to zero. H =0

In the condition equations model, a mathematical model is set up for each observation such that the
adjusted observation fulfill a certain condition.
Page | 6

Ece 2306 Notes Vii Jan2014

Uploaded by

Copyright:

Available Formats

Ece 2306 Notes Vii Jan2014

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ece 2306 Notes Vii Jan2014

Uploaded by

Copyright:

Available Formats

Techniques of least square adjustment

a) Method of observation equations, also adjustment of indirect observations, also method of

- Notice one equation for each observation l1 and l2

For n observations and u parameters, the above matrices become:

Normally L, contains known numerical values.

Combining all the terms:

Transposing on both sides

W = identity since all measurements are assumed to be of equal precision.

Using matrix to form normal equations

The normal equations are then formed as follows: ( ) =

The observation equations for each angle

5703 50" + 3 = 180 (1 + 2 )

0.666667 0.33333 3.046714615

Condition equation adjustment model

c) The difference in elevation in a closed leveling loop should be equal to zero. H =0

You might also like