Construction method and application of noise reduction model based on deep learning

The invention relates to the technical field of pipeline leakage detection and noise reduction, and provides a construction method and application of a noise reduction model based on deep learning, and the method comprises the steps: carrying out the signal modulation of a pure signal and a noise signal according to different proportions, and obtaining noisy signals with different signal-to-noise ratios; performing short-time Fourier transform (STFT) on the pure signal to extract a frequency spectrum amplitude of the signal, and taking an absolute value of the amplitude as a target sample for model training; the method comprises the following steps: performing short-time Fourier transform STFT on a noisy signal to obtain a frequency spectrum amplitude of the noisy signal, taking an absolute value of the frequency spectrum amplitude, correcting the amplitude by using a PAS algorithm, and taking the corrected amplitude as a training sample; normalizing the amplitude of the training sample and the amplitude of t