Decagon Devices KD2 Pro Operator's Manual page 63

KD2 Pro
sought which minimize the difference between model and measured
values of temperature rise. Temperature of the needle is measured,
so ∆ is first computed by subtracting the initial temperature from
all readings. Temperature rise is further scaled by multiplying by 4
and dividing by . If we call this new temperature variable T , then
we find the values of
error.
where the are the measured values and the
i
with Equation 9. Equation 11 shows the standard error of estimate
for the measurements.
where is the number of measurements. The units of
It can be made dimensionless if it is multiplied by the thermal con-
ductivity, . This dimensionless value gives the error in fitting the
model to data as a value independent of heater current or the ther-
mal conductivity of the medium. The KD2 Pro computes err as in
equation 12.
It is a dimensionless measure of the goodness of fit of the model to
the data. It can be converted to temperature by dividing by
multiplying by .
Note: The Err term is not a rigorous statistical indicator of mea-
surement quality, but it serves as a qualitative quality indicator.
and that minimize the sum of squares of
= (
i
=
yx
=
yx
59
7 KD2 PRO THEORY
2
)
i
are values modeled
i
are
yx
(10)
(11)
.
(12)
and