Corrosion under insulation (CUI) poses a significant threat to the integrity and safety of long-distance thermal pipelines. However, conventional non-destructive testing (NDT) methods struggle to achieve early-stage detection of CUI while effectively balancing detection reliability, operational safety, and low cost. This paper presents a novel curvature-adaptive flexible array capacitive sensing system for the early identification of insulation defects, a critical precursor to CUI. The proposed system features a flexible capacitive probe with coplanar array electrodes on an adjustable substrate, allowing it to conform to various pipeline curvatures to enhance signal-to-noise ratio. A key innovation lies in the proposal of a novel approach for determining the depth of internal defects in thermal insulated pipelines, which leverages the stratified electric field generated by the electrode array and realizes defect depth estimation by analyzing the relative capacitance deviations across different electrode pairs. Furthermore, a feature decoupling method based on optimizing the sensor’s lift-off height is introduced to distinguish between internal insulation defects and external protective jacket topography. To automate the process, an enhanced one-dimensional convolutional neural network is developed, achieving high-precision automated classification of jacket surface defects. Experimental results demonstrate the system’s capability to detect simulated 'air–water void’ defects within insulation layers up to 50 mm in depth, with a measurement sensitivity at the 10⁻4 pF level. The integrated system offers a robust, cost-effective, and non-invasive solution for the early warning of CUI, addressing critical limitations of existing NDT techniques and paving the way for intelligent pipeline integrity management.
Zeng et al. (Tue,) studied this question.