A Pipeline Leakage Diagnosis Method Based on CNN-BiGRU Twin Network

Aiming at the problem of poor pipeline leakage detection, a leakage detection method based on the CNN-BiGRU twin network is proposed. Compared with the traditional deep neural network, the twin network adopts the method of small sample pair training. Under the same sample size, the effective training times of the network model are increased to improve the pipeline detection performance. This paper proposes that the convolutional neural network and the bidirectional gated recurrent unit together form the twin network structure, and compares them with other deep neural network models. The experimental results show that the CNN-BiGRU twin network method improves the recognition rate by 16.94% compared with CNN. It has achieved better detection performance and has certain engineering application value.