88
A physical process can be described by the structure Ps and, the
undesired disturbances cLP, superimposed on itPs dP. The
disturbances dP consist of a regular component rp and of a random
component mp. This may be written symbolically as: dP
Vp -f~ Wlp.
Only abstract processes (e.g. algebraic, logical processes,
can be free of disturbances.
When applying a physical process to an imperfect input, a dis
torted output is obtained:
{Si Dt) x (Ps dP) S0 D„.
The meaning of symbol x is "the process applied to the input".
The testing procedures for comparison of similar instruments
should preferably be conducted for larger groups of competitive
systems. They should cover the operations and ranges which they
have in common.
When conducting the testing procedures, some thought should
be given to the succeeding stage the data processing.
The output of the test should facilitate an easy and consistent
processing of data. This requires a systematic arrangement of
tests, including some standardisation.
4.5. Data processing
The purpose of data processing in tests is to transform the output
—thus the raw informationinto a form suitable for the judgment
of the system performance. The output is obtained:
a) in a digital form, or
point by point restitution
b) in an analogue formI) single points,)
II) line figures,
III) photographic records,
IV) distances, angular values, light trans-
mittance, temperature, voltage (ob
tained with auxiliary devices).
For numerical data processing, the analogue output, representing
the raw information, must be converted into digits. The accuracy
of the method and execution of the conversion are essentially
important.
The sampling rate in this conversion process should be selected such
that no significant information will be lost. The equipment used
for measurements must be sensitive enough and calibrated.
The execution of a conversion is associated, as any other operation
with the procedural errors; they should be kept insignificant.
The accuracy of the analogue to digital conversion can be veri
fied by applying it to known representative samples. The equipment
and the technique of conversion should be more accurate than the
