Abstract Submarine pipelines predominantly employ a double-layer insulation structure. Corrosion or damage to the outer pipe can lead to water ingress into the interlayer, ultimately resulting in the failure of the intermediate insulation layer due to submersion in seawater. Conventional detection methods (manual inspection, infrared thermal imaging, and fiber optic detection) are hindered by the influence of the complex marine environment and technical limitations, thereby failing to ensure the effectiveness of leakage detection for the outer pipe of submarine pipelines. This paper focuses on the intelligent acoustic internal detection technology. The innovation of this paper is proposing a novel leakage detection method for submarine pipeline outer pipes based on percussive acoustic signals. This method is supported by a detection experiment of submarine pipeline outer pipe leakage. Firstly, the percussive acoustic signals under two states of normal and outer pipe leakage (water ingress into the insulation layer) are collected in the experiment. Secondly, the acoustic signals are segmented and processed to construct a detection dataset. Finally, the time-domain and frequency-domain distributions of the processed signals are utilized as inputs to a dual-channel convolutional neural network (DC-CNN) for the leakage detection of the outer pipe of submarine pipelines. The experimental results demonstrate that the detection method based on percussive acoustic signals can achieve effective leakage diagnosis of the outer pipe of submarine pipelines with an accuracy rate of 95.83%, thereby providing a technical foundation for the subsequent development of intelligent acoustic internal detection equipment.
Liu et al. (Sun,) studied this question.