FUZZY SETS OF QUASI-PERIODIC SIGNALS

A quasi-periodic signal is a signal that, having a repetitive recognizable periodic pattern, has variations in period and amplitude. Examples of this type of signals are, e.g., the electrocardiogram, the accelerations produced during the human gait, the vibration of a musical instrument, etc. Currently these signals are analyzed using, among others, statistical methods, e.g. Hidden Markov Models, or Soft Computing methods, e.g. Neural Networks. This paper contributes to this field by introducing a method for creating a Linguistic Model of a set of these signals based on a Fuzzy Finite States Machine. The obtained model is more intuitive than the existing approaches and its computational implementation is easy to interpret. A practical example of application is provided.