GASSONIC Observer-i, Инструкция по эксплуатации

Страница: 9 / 89

GASSONIC OBSERVER-i
9
The Artificial Neural Network (ANN) makes it possible to analyze the incoming acoustic sound,
based on the frequency domain instead of the sound level domain (dB levels) in individual
frequency bands. As such, the Gassonic Observer-i detects only the sound from gas leaks,
even if the gas leak sounds are occurring at much lower sound levels than the background
noise. In reality this means that the ANN is extremely immune to false alarms from unwanted
background noise sources, but at the same time extremely sensitive to gas leaks regardless of
size.
The Gassonic Observer-i analyzes acoustic noise from as low as 12 kHz where other
ultrasonic gas leak detectors have to filter out noise below 20 kHz to prevent interference from
compressors or other man-made acoustic noise unrelated to gas leaks. With the ANN multi-
spectral technology, the frequency band works down to 12 kHz, allowing more sound energy
from potential gas leaks to be picked up and analyzed which provides a greater detection
range of gas leaks.
With the ANN technology, the Gassonic Observer-i comes with pre-trained neural network
algorithms from the factory and does not require complicated onsite training procedures to
adapt to specific acoustic plant conditions; instead it is ready to operate in all kinds of
acoustical environments, immediately after installation.