Pipeline leak detection method based on acoustic-pressure information fusion

Pipeline transportation is one of the essential transportation means, and accurate detection of pipeline leaks is of great significance for energy safety and environmental protection. However, most of the existing leak detection methods are based on a single type of signal, which generates a high false alarm rate in complex operating environments. Therefore, this paper proposes a pipeline leak detection method based on acoustic-pressure (A-P) information fusion and a noise reduction algorithm based on dual Pearson thresholds-ensemble empirical mode decomposition (DP-EEMD). Firstly, the original signal is decomposed using the EEMD algorithm. The signal is reconstructed by screening the practical components according to the similarity coefficients between the intrinsic mode functions (IMFs) and the original double-ended signals. Secondly, information fusion is performed at the data level between the acoustic signal and the pressure signal to form the fused signal. Finally, a one-dimensional convolutional neural network is established to extract the relevant features of the acoustic-pressure fusion signal to diagnose the pipeline leakage. The experimental results show that the classification accuracy of the leak detection method proposed in this paper reaches 98.3%, which is higher than that of the 1D-CNN network and BP network using a single type of signal as input. Meanwhile, the noise reduction effect of the DP-EEMD algorithm is also better than the wavelet noise reduction method and the standard EEMD noise reduction algorithm.