Table of Contents
Advertisement
phase
is
where new input
ived,
the necessary squares or
formed, and the appropriate
calculated
as required
by
the
output. The
rate
at
which
the
can be made, the input values
The Input
values are
summation
desired fina
completed
v{tithout
i
nterruption
dete
rmi
nes the
maximum r4te of execution (see Execution
Time).
The
Averagfng Period Processing occurs
whenever
tfte number of input samples entered
in Parametdr 7 is satisfied or whenever
an
Output
Intertval
occurs (i.e., whenever
the
Output Flag is set). Results from these
calculationslare stored sequentially in Input
Storage
locftions
starting with
the location
specified
in
lParameter
9.
The calculations
performed dre shown below,
where N
is
the
number
of
ittrput
samples
in
the averaging
period:
1.
Means:
=
E)UN
2.V
=
rx2lN
-
(DVNp
Deviations:
)
=
Y14trz
Y)
=
!XY/N
-
XX tY/N2
= CV(X,Y)/(SD(X)SD(Y))
SECTION
10.
PROCESSING INSTRUCTIONS
lf
a
fast execution interval is specified, it
is
possible that
the combined execution times of
the input and
averaging period processing may
exceed
the program table execution interval.
The occurrence of
an execution interval Overrun
(see Section
2.1 of 21X Operator's Manual)
is
noted by decimal points on either side
of
the
G
in LOG
(.0
MODE). This
results in
the omission
of
one input
sample.
The calculations
are not
affected, however, since the number of input
samples is incremented only when valid input
processing
occurs. Averaging
period processing
occurs only when
the number of input samples
specified in Parameter 7
is
accumulated.
Regardless of
whether all
of
the input samples
for the
averaging period (specified in Parameter
7) have occurred or not, averaging period
processing occurs whenever the
Output Flag
is
set.
This accommodates situations where the
Output Interval may not be an integer multiple of
the
averaging
period:
lf
for example a
30 minute
Output Interval is set by Instruction 92 and an
8
minute averaging period
is
specified by
Parameter 7, then three
I
minute and one
6
minute calculations
willoccur.
The proper
weighting
of
these
values in producing the final
output
is
described below.
The Output Processing occurs only
at
the
Output Intervaland involves averaging the
statistical results obtained at the averaging
periods.
These
final results are then stored
sequentially
in Input Storage beginning with the
location
specified in Parameter
9.
The Sample
Instruction
70 must be used to transfer the data
to Final Storage. All
but
the last averaging
period
in
the
Output Interual willcontain the
same number of input scans as specified by
Parameter
7.
To insure that results from the
averaging periods contribute to the
final result
proportionalto their averaging periods, the
output processing uses the
following equation:
RF
= (NR1 + NR2 + .... + N'RL)/NT
is
the
final result
R2, etc., are
the results from the averaging
period processing
is
the result from the last averaging period
in
the
Output Interval
is
the number of input samples
in
the
specified
averaging period (Parameter
7)
3.
4.
5.
RF
R1
BL
N
NOTE$: 1.
The square root
algorithfn
in
the
21X returns a result
of 0
for negative arguments.
2.
The
divide algorithm returns the
largest
floating point number possible
(ttOte
displayed
as
t99999) for
a
divide by
0.
3.
When computing the
variance of
a
constant
signal, round off error
producqs a
small negative result.
The 21X returns a
0
for the square
root of
4 negative number; therefore,
the staridard
deviation
is
set
to
0.
lf
the sig4al
is
also used
in
a
correlaton calculation, division by
0
returns an overrange value for
the
result.
10-13
Advertisement
Table of Contents
