A diagnosis and monitoring system with a multi-scale channel and spatial attention mechanism on residual structures for tubing leakage detection

In the context of the inherently dangerous oil and gas industry, it is important to closely monitor the condition of gas well tubing throughout the production process. However, the current major methods such as logging require on-site shutdown of production and moving the tubing column, which is costly and risky. In this paper, a diagnosis and monitoring system with multi-scale channel and spatial attention mechanism on residual structure for downhole tubing leakage is designed to judge the leakage status by the collected acoustic signals of annulus. Firstly, a simulation experimental device is constructed to obtain the leakage acoustic data under different leakage conditions; then a multi-scale channel and spatial attention convolution network based on residual structure is proposed to diagnose the leakage status above and below the annular liquid level, with an accuracy rate of 91.67 %; finally, we design a software and integrate the diagnostic model, sensors, acquisition card, and upper computer to construct a set of monitoring system, which can be remotely controlled to carry out real-time diagnosis of the leakage status, and process and upload the data. The system was tested on-site, and it enables high-precision leak localization and early leak detection. This work offers a reference and inspiration for future innovations in tubing leakage monitoring.