Table of Contents
Advertisement
What about slowly drifting inputs?
Again – No Problem. The algorithm is processed in 'Real Time', i.e. as the reading is taken.
As it already knows the previous logged readings it can calculate where the next point
should be (assuming it's on a straight line) – if the measured value does not equal the pre-
dicted value, the point is logged as it no longer forms part of the straight line.
Fuzzy Logging, looks for straight lines – at any angle. Not just on the horizontal.
Do you have any examples?
Example 1.) Flow & Pressure Measurement of Mains Water Pressure
A recorder was installed, to monitor the flow of a mains water supply. At peak demand the
mains pressure had been subject to sharp drops in pressure and flow rate, and it was nec-
essary to find the cause of the problem.
The recorder had to have a fast scan rate, in order to capture the 'glitches'.
The recording period would be over many days if not weeks, so storage capacity was at a
premium.
A fast scan rate using the standard sampling method would result in a disk life of about a
day, which was not acceptable.
As this application consists of long periods of little activity (relatively constant flow rate), and
short periods of high activity (rapidly changing flow rate), it is ideally suited to Fuzzy Logging.
During the hours of stable flow where the flow rate remained more or less constant, the
Fuzzy Logging technique would give compression ratios up to 100 times. However, as soon
as a glitch appeared the fast sampling rate was able to capture and store all the points.
Example 2.) Cold Storage Temperature Measurement
A recorder was required to help track random and rapid temperature changes within the cold
storage rooms. Conventional sample recording had shown that temperature variations were
present, but was not of high enough resolution to pinpoint the cause.
Again as in example 1), the measured inputs would show long periods of stable constant
readings, interspersed with small sharp increases in temperature. In order to track the cause
of these variations, it was necessary to maximise the time resolution of the data. This appli-
cation was ideally suited to the Fuzzy Logging data storage technique, as the periods of in-
activity would result in compression rates of over 50 times.
Example 3.) Logged Data Example
The diagram below is a sample of actual logged data in both the Fuzzy Logging method (top)
and Sample Logging method (bottom), derived from the same analogue input.
43-TV-25-35 GLO Iss.4 Dec 06 UK
A5E01001767-04
257
Advertisement
Table of Contents
