The invention discloses a method for detecting
weak pulse signals. The method includes the following steps that: one-dimensional multi-scale
wavelet decomposition is performed on measured signals, namely, the measured signals are decomposed into N
layers through one-dimensional
wavelet transform, and the high-frequency portions of the signals are reconstructed through utilizing a
wavelet decomposition coefficient, so that N high-frequency reconstruction signals dj and one low-frequency reconstruction
signal a are obtained; sliding kurtosis
processing is performed on the reconstruction signals dj, so that a corresponding sliding kurtosis
time sequence cj can be obtained; and superimposing synthesis and normalization
processing are performed on the sliding kurtosis
time sequence cj, so that a final sliding kurtosis
time sequence c can be obtained. After being subjected to the one-dimensional multi-scale
wavelet decomposition preprocessing, the measured signals embody different time-
frequency domain characteristics in different
decomposition scales, and when the sliding kurtosis
processing is performed on the signals, a detection effect of multi-
slide window synchronization action can be realized, and therefore,
differential analysis and processing on noises, interferences and pulses to be detected mixed in measured signals can be benefitted.