![background image](http://html.mh-extra.com/html/gassonic/observer-i/observer-i_instruction-manual_568581008.webp)
GASSONIC OBSERVER-i
8
2.0 General Description and Features
The Gassonic Observer-i detects leaks from pressurized gas systems by sensing the airborne
ultrasound produced by gas escaping from the leak. This detection method is omni-directional,
functional in extreme weather conditions, and ideal for fast monitoring leaks from valves and flanges in
complex pipeline systems, both onshore and offshore.
The main advantage using ultrasonic gas leak detectors is that the detector does not need to wait for
gas accumulation, instead it responds instantly, in distances up to 28 meters, when a gas leak occurs.
The Gassonic Observer-i is suitable in all pressurized gas installations, from 2 BAR (29 psi) or more,
where the escaping gas is in a gaseous state when it leaks.
The Gassonic Observer-i
can be configured to operate in Enhanced or Classic Modes. In Classic
Mode, reminiscent of the legacy Observer-H UGLD, the alarm decision is based on the user-adjustable
SPL threshold. The Classic Mode makes it possible to retrofit the Gassonic Observer-i into Gassonic
Observer and Gassonic Observer-H installations. In Enhanced Mode, the detection method is based
on the intelligent ANN algorithm that can distinguish gas leaks from background noise.
The Gassonic Observer-i is certified to ATEX, IECEx, FM, CSA, HART, and IEC 61508 standards. The
detector housing is cast AISI 316L, acid-proof stainless steel and the ingress protection is IP66 with a
NEMA rating of Type 4X. The performance of the Gassonic Observer-i as a safety device is not
covered by the ATEX certificate.
2.1.
Artificial Neural Network (ANN)
An essential performance parameter for an ultrasonic gas leak detector is to ensure high
acoustic sensitivity to real gas leaks while at the same time minimizing the influence from
background noise sources, unrelated to gas leaks. To ensure this very important feature, the
Gassonic Observer-i is the first ultrasonic gas leak detector to use multi-spectral Artificial
Neural Network (ANN) algorithms in the detector’s advanced acoustic sound processing design
to distinguish real gas leaks from false alarms.
The ANN is a mathematical algorithm and is used to search for
familiarity
in a large and
complex set of data. ANN works very similar to how the human brain handles the constant flow
of information received through its senses - the eyes, ears, nose, and mouth. For example,
when we have seen a face of a person in their youth and heard their voice, we are often able
to recognize the person 20-30 years later, even if the person has changed over time. The
reason we can recognize the person is that our brain is not programmed to search for an exact
match or pattern, instead the brain looks for a combination of familiar similarities which the
brain has been trained to compare and then make a decision. If the human brain did not look
for familiarity when we met another person but instead looked only for an exact match as we
remembered the person, we would only recognize that person if they had not changed and
looked exactly the same.
An ultrasonic gas leak detector does not have to recognize people at different ages; instead it
needs to effectively recognize the sound signature from gas leaks while at the same time also
rejecting sound signatures from acoustic background noise unrelated to gas leaks. The
Gassonic Observer-i uses advanced ANN algorithms to enhance and optimize the detector’s
ability to distinguish between normal background noise not related to leaking gas and real gas
leaks. Due to the use of ANN technology, the Gassonic Observer-i is able to constantly record
and analyze the flow of complex acoustic sound generated in very noisy plant environments,
and instantly raise an alarm if the specific noise from a gas leak is recognized.