Mohd Fairusham Ghazali, Muhammad Hanafi Yusop, Erdiwansyah, Semin, Hadi Suyono
Leak detection in water distribution pipelines remains challenging because transient pressure signals are nonlinear, non-stationary, and highly contaminated by background noise. Although the HHT provides adaptive time–frequency analysis for the interpretation of transient signals, its practical implementation is limited by the subjective selection of IMFs, which often leads to inconsistent leak identification and reduced reproducibility. To address this limitation, this study proposes a hybrid HHT–I-Kaz framework that establishes an automated analytical sequence consisting of signal decomposition, statistical IMF ranking, adaptive mode selection, and transient spectral localisation. The proposed I-Kaz criterion evaluates the kurtosis-derived energy distribution of each IMF to identify components containing impulsive non-Gaussian leak reflections, thereby eliminating manual IMF interpretation and improving robustness against noise contamination. Experimental validation was conducted on a 67.9 m MDPE pipeline equipped with a calibrated 3 mm leak located 19.7 m from the inlet under operating pressures of 1–2 bar. The proposed framework achieved 96.8 % classification accuracy, an AUC of 0.97, improved the signal-to-noise ratio by approximately 18 %, reduced computation time by nearly 20 %, and localised leaks with a maximum error below 3 %. The I-Kaz criterion consistently identified the most informative IMF, with IMF3 yielding the highest coefficient of 0.0118, confirming its effectiveness as an objective, physically interpretable IMF selection mechanism. The results demonstrate that integrating higher-order statistical analysis with adaptive HHT decomposition provides a reliable, noise-resilient, and fully automated solution for real-time pipeline leak detection and localisation in water distribution and industrial pipeline systems. © 2026 Elsevier Ltd
Faculty of Mechanical and Automotive Engineering Technology, Universiti Malaysia Pahang Al-Sultan Abdullah, Pahang, Malaysia; Centre for Advanced Mobility and Aerospace, Universiti Malaysia Pahang Al-Sultan Abdullah, 26600, Malaysia; Department of Natural Resources and Environmental Management, Universitas Serambi Mekkah, Banda Aceh, 23245, Indonesia; Department of Marine Engineering, Institut Teknologi Sepuluh Nopember, Surabaya, 60111, Indonesia; Department of Electrical Engineering, Faculty of Engineering, Universitas Brawijaya, Malang, Indonesia
