Page 234 - Mechanical Engineers' Handbook (Volume 2)

5 Data Storage 223

Data can be stored around trigger in three ways:

Beginning trigger: Window can be Ending trigger: Window can be stored with
stored with trigger occurring at earliest trigger occurring at latest point of stored
point of stored interval. interval.
Benefits: Can aid in detecting what Benefits: Can aid in detecting what happened
happened after the trigger event. leading up to the trigger event.
Disadvantages: Can’t tell what led up to Disadvantages: Can’t tell what happened after
the triggered event. the triggered event.

Beginning of End of
moving moving
window window

Timeline
This was the
This was the Centered trigger: Window can be stored with most recent
earliest data data acquired
trigger occurring at middle point of stored
acquired in in the moving
interval.
the moving Benefits: Can aid in detecting what happened window
window
before and after the trigger event.
Disadvantages: None, other than still have to
predetermine what constitutes a trigger.
Figure 5 Relationship between trigger and moving window.

This has the advantage of allowing monitoring and analysis of high-speed events and
still capturing some data to enable determining some data relationships. This is most useful
if the engineer has some idea of what events may yield valuable relationships. It is much
less useful when events, triggers, or relationships are unknown or unexpected. Sampling data
at slower intervals may serve to allow accidental capture and identification of useful rela-
tionships, but the work required to find those relationships is much higher and of questionable
probability of success.
As an example, in the sawmill many variables are changing state at high speed. For
diagnostic purposes, it is valuable to see a high-speed snapshot of states of photo eyes
compared to saw drops, gate openings, and grade decisions. However, the volume of data is
normally too great for storage and later analysis. There are some events that are of more
importance than others, such as when a gate is opening early or a saw is failing to drop
consistently. These can often be recognized and the data captured in the window of time
before and after the event can be stored, allowing later analysis of what led to the event and
what happened shortly afterward. Some characteristic data can be summarized for each time
window, stored, and used later for analysis, such as the number of photo eye changes, number
of saw drops, and number of gate openings. More complex relationships can be tallied to
aid in diagnostics, such as number of gate openings for grade 2. The more complex the
relationship, the more difficult the programming task to ensure capturing the incidence to
storage. A typical pattern is to collect process variables that may be of interest, often from
a programmable logic controller. As a given problem begins to be identified, additional logic
can be added to examine relationships between process inputs and sequences of events,