Sharp JX-9400 Technical Information Seite 173

  • Herunterladen
  • Zu meinen Handbüchern hinzufügen
  • Drucken
  • Seite
    / 211
  • Inhaltsverzeichnis
  • LESEZEICHEN
  • Bewertet. / 5. Basierend auf Kundenbewertungen
Seitenansicht 172
will be obtained, and a correlation between C and n will be found: the larger C
values correspond to the smaller n and vice versa. A good identification
technique should give the most likely couple of coefficients together with the
probability density fðC; nÞ. Such a technique exists (Tarantola, 1987) and is
summarized below.
Identification of the model parameters
Let us put in a vector, z, both measured data and model parameters that have to
be determ ined and assume that this vector is a random variable with a normal
distribution in k-fold space (k ¼ number of parameters and data):
fðzÞ¼A exp ½
1
2
ðz z
p
Þ
T
C
1
z
ðz z
p
Þ ð7:35Þ
where:
z
p
is the a priori vector, z, containing the measured values and reasonable
estimates of the parameters to be identified,
C
z
is the covariance matrix between the elements of z. Its diagonal elements
are the variances of the measured quantities and a priori estimated
variances of the parameters. These latter varian ces are generally large,
since the parameters are generally not known before the me asurements.
The components of the vector, z, are linked by a mathematical model or a set of
equations that can be written:
ðzÞ¼0 ð7:36Þ
For example, if a linear relationship is assumed between two measured vari-
ables, x and y, the set of equations:
y
i
¼ a þ bx
i
ð7:37Þ
can be written in a matrix form:
1 x
1
y
1
1 x
2
y
2
  
  
1 x
n
y
n
0
B
B
B
B
B
B
@
1
C
C
C
C
C
C
A
a
b
1
0
B
@
1
C
A
¼ 0 ð7:38Þ
Generally, the proposed model is not exact and it may be assumed that it has a
normal distribution:
gðzÞ¼B exp½
1
2
ðzÞ
T
C
1
T
ðzÞ ð7:39Þ
where C
T
is the covariance matrix of the model. If the model is exact, this
distribution is a Dirac distribution:
gðzÞ¼½ðzÞ ð7:40Þ
Combining the prior knowledge contained in the distribution fðzÞ with the
model described with the distribution gðzÞ gives a new distribution containing
152 Ventilation and Airflow in Buildings
Seitenansicht 172
1 2 ... 168 169 170 171 172 173 174 175 176 177 178 ... 210 211

Kommentare zu diesen Handbüchern

Keine Kommentare