Cycle detection

Abstract

After measuring a variable during some finite time interval T, we want to determine the possible regular cycle variations. To do this, we give the summary of a method described earlier. This method consists in determining a set of successive discrete Fourier Transforms. This allows us to study first the phase coherence between the different terms that have been obtained, secondly the mean values. This paper principally aims at discussing the criteria that can corroborate or deny the presence of a quasi‐periodical signal. Special attention has been devoted to non‐stationary phenomena. Some examples (varves, dendrochronology, biological phenomena) have been examined.