Establishment of leakage detection model for oil and gas pipeline based on VMD-MD-1DCNN

Abstract

In order to solve the problem of recognition error caused by noise interference in oil and gas pipeline signal and traditional pipeline leakage detection relies on expert experience to extract features manually, an oil and gas pipeline leakage detection model based on deep learning is proposed in this paper. The model consists of data preprocessing part and pattern recognition part. Firstly, a signal denoising algorithm based on variational mode decomposition (VMD) and Manhattan distance (MD) is proposed to reduce the error caused by data quality in the subsequent pattern recognition process. Secondly, combined with the one-dimensional and temporal characteristics of pipeline signal, the signal denoised by VMD-MD algorithm is used as the input of one-dimensional convolution neural network (1DCNN) in deep learning, then the data features are learned independently through the characteristics of network structure. Finally, the network structure and parameters are optimized and analyzed to build the optimal pipeline leakage detection model according to the experiment. The experimental results show that, compared with other existing models, the ensemble VMD-MD-1DCNN model has a better improvement in each evaluation index, which verifies the effectiveness of its application in pipeline leakage detection.