The invention belongs to the technical field of nuclear
logging data processing, and particularly relates to a
neutron detector pulse
signal intelligent discrimination method and
system based on
deep learning, and the method comprises the following steps: collecting a pulse
signal outputted by a
detector, and when the rising edge of the pulse
signal is detected to exceed a preset
ground noise threshold value, determining the pulse signal; intercepting a
time sequence containing the complete pulse signal, and carrying out normalization
processing on the
time sequence to obtain a normalized dimensionless
voltage sequence; and calculating a first-order difference absolute value of adjacent sampling points in the dimensionless
voltage sequence, and carrying out weighted calculation in combination with the
time distance from the current sampling point to the peak moment to obtain a waveform oscillation entropy index representing the waveform microcosmic oscillation degree. According to the method, more than 98% of
neutron tube ignition interference can be effectively eliminated, the high-energy part of the
energy spectrum is remarkably purified, and the problem that data of the aged
neutron tube are unavailable is solved.