ИСТИНА |
Войти в систему Регистрация |
|
ИСТИНА ИНХС РАН |
||
This article is devoted to automatic identification of main earthquake phases (wave types), especially to the detection of first precursors of P with a view to making use of these for operation with signals strongly contaminated with noise as recorded by individual earthquake warning devices that can be installed at end users. In that case, methods are urgently needed to cope with a great number of manmade seismic noises. The noise may be due to very diverse factors, while the noise itself may be either constant or impulsive. We suggest using artificial neural networks to deal with the discrimination of seismic signals, viz., an earthquake or noise. We show that, if the processing of seismic signals can be carried out very fast (during 4-5 seconds), then such earthquake warning systems can significantly reduce human and material losses.