Industrial fluid pipeline leak detection and localization based on a multiscale Mann-Whitney test and acoustic emission event tracking

Abstract

This paper proposes a novel technique for leak detection and localization in industrial fluid pipelines. Artificial intelligence-based supervised methods are popular adoption pipeline identification. However, techniques need prior knowledge about failure training purposes. To address this challenge, new method independent of is proposed. First, acoustic emission signals from the acquired. Then, multiscale Mann-Whitney test developed, output statistics used as state indicator. After detecting leak, proposed localizes by using newly developed called event tracking. A major challenge false alarms that generated poor identification leak-related events. The tracking presented work precisely determines first detects hits variability index constant alarm rate algorithm. short-term energy calculated hit perceived windows. high-energy events separated into an bank. Leak-related filtered out bank theory wave propagation. obtained elucidate leaks thus reduce error localization. results outperformed reference terms accuracy under variable pressure conditions.