Health Monitoring of Ultra-low Temperature Valves Based on Complex Shearlet Domain Dynamic Threshold

Ultra low temperature valves are widely used in liquefied natural gas (LNG) pipelines. However, valve leakage may cause major disasters due to the particularity of LNG. To facilitate the efficient monitoring of ultra-low temperature valves’ health, we introduce a dynamic threshold valve health monitoring algorithm based on the complex Shearlet domain. The proposed algorithm first performs complex Shearlet transform on the obtained infrared images, and then denoises the transformed coefficients. Secondly, the de-noised infrared images are segmented by dynamic threshold. The valve will be assumed to be leaking when the image changes in a large range, thus serving the purpose of early warning of a possible major disaster. The experiments verified that the proposed method can quickly and effectively realize the health monitoring of ultra-low temperature valves which further verified the effectiveness of the algorithm.