Journal of Engineering Research

To address the safety inspection challenges of steam pipeline thermal leakage and wall thickness reduction, this study proposes a non-contact intelligent detection method based on infrared thermography. By integrating multi-source data with advanced machine learning algorithms, an association model linking thermal leakage intensity, temperature field distribution, and wall thickness reduction is constructed. Additionally, an adaptive algorithm is employed to compensate for environmental interference, significantly enhancing detection robustness. An innovative comprehensive evaluation system incorporating thermal pattern features, temperature gradients, and material attenuation is established to achieve precise identification and quantitative analysis of pipeline defects. Experiments demonstrate that this method performs exceptionally in early defect warning for core pipelines in power stations, with detection efficiency improved by over 40% and costs reduced by approximately 35% compared to traditional methods. The research findings not only provide novel technical solutions for steam pipeline safety inspection but also drive the development of industrial non-destructive testing towards intelligence, possessing significant engineering application value and academic expansion potential. igence, offering significant engineering application value and academic expansion potential.

Steam pipelines; Thermal leakage; Wall thickness reduction; Thermal imaging technology; Machine learning