The possibility of constructing invariant feature space of seismic detection signals in the time and frequency domains

The authors solves the problem of finding an invariant feature space in signals of seismic detection systems. Considered indicative of time and frequency domain. By adjusting the parameters of the classifier authors were able to ensure the invariance feature space in the time domain and establish so-called «seasonal scale factors». Proved that the impact of seasonal factors on the classification results in the frequency domain is small compared with the main factor - the presence of the intruder.