Optical fiber sensor vibration signal decomposition and extraction method

The invention discloses an optical fiber sensor vibration signal decomposition and extraction method, which comprises the following steps of: firstly, segmenting an original vibration signal into equal-length fragments according to a time sequence, and reserving partial overlapped data between adjacent frames; multiplying each frame of signal by a specific periodic signal to form a periodic signal; performing Fourier transform on each frame of signal to convert a time domain characteristic signal into a frequency domain characteristic signal so as to obtain a group of specific frequency characteristics; thirdly, selecting and determining a threshold value of a characteristic frequency interval based on a Raida criterion method; and finally, extracting different modal components of the original vibration signal through band-pass filtering from the interval energy blocks from high to low. According to the method, the frequency range of signal energy abrupt change is found through the Raida criterion, the origin