Detection of weak periodical signals of fault merged in quasi-random telegram

Two problems are presented, one is quasi-random telegram (QT) and the other is cycle-correlation. When signals are so strong that their amplitudes are limited to a certain extent due to the saturation characteristic of devices or testing instruments, then the signals should be changed into QT; there are differences between QT and classic random telegram (RT). Cycle-correlation can be used to detect weak periodic signals merged in noise but classical auto-correlation cannot be used in this case. The difference between cycle-correlation and classical auto-correlation is also given. A computer simulation extracting periodical weak fault signals from the background of QT was carried out. Conclusions for the stationary random process are also provided.