Sharp JX-9400 Technical Information Seite 169

  • Herunterladen
  • Zu meinen Handbüchern hinzufügen
  • Drucken
  • Seite
    / 211
  • Inhaltsverzeichnis
  • LESEZEICHEN
  • Bewertet. / 5. Basierend auf Kundenbewertungen
Seitenansicht 168
Linear least square fit of the first kind
Such methods are used to find the coefficients of leakage models of Equations
4.1 or 4.2 in fa n pressurization (see Chapter 4, ‘The fan pressurization
method’).
The regression of the first kind assumes that the abscissa, x
i
, of each
measurement is well known and that the distribution of the ordinates around
the regression line is Gaussian with a constant standard deviation. This
method is very commonly used but it should be emphasized that the above
hypotheses are not verified in the case of permeability tests because the
values of x
i
are measured estimates.
The regression line of the first kind minimizes the sum of the square of the
residual ordinates (vertical distances):
SSR
y
¼
X
N
1
½y
i
ða þ nx
i
Þ
2
ð7:13Þ
Its coefficients can be calculated using the following relationships. First
compute the estimates of the averages:
xx ¼
1
N
X
N
i ¼1
x
i
yy ¼
1
N
X
N
i ¼1
y
i
ð7:14Þ
and the estimates of the variances:
s
2
x
¼
1
N 1
X
N
i ¼1
ðx
i
xxÞ
2
s
2
y
¼
1
N 1
X
N
i ¼1
ðy
i
yyÞ
2
s
xy
¼
1
N 1
X
N
i ¼1
ðx
i
xxÞðy
i
yyÞ
ð7:15Þ
Then the best es timates of the coefficients a and n, according the above
hypotheses, are:
n ¼
s
xy
s
2
x
a ¼
yy n
xx
ð7:16Þ
The slope given by Equation 7.16 is valid if the x
i
are exactly known, and the
minimized distance is the sum of the squar e of the vertical dist ances between
the measured points and the regression line.
148 Ventilation and Airflow in Buildings
Seitenansicht 168
1 2 ... 164 165 166 167 168 169 170 171 172 173 174 ... 210 211

Kommentare zu diesen Handbüchern

Keine Kommentare