Amir Houshang Ayati - Ph.D. Candidate, Faculty of Civil Engineering and Architecture, Shahid Chamran University of Ahvaz,Ahvaz, Iran
Ali Haghighi - Professor, Faculty of Civil Engineering and Architecture, Shahid Chamran University of Ahvaz,Ahvaz, Iran
Hamid Reza Ghafouri - Professor, Faculty of Civil Engineering and Architecture, Shahid Chamran University of Ahvaz,Ahvaz, Iran

This study aims at introducing a general framework for leak detection in pipelines by coupling machine learning to transient hydraulics. For this purpose, the Support Vector Machine (SVM) as a superior pattern recognition algorithm is applied to transient-based leak detection problems. First, a transient simulation model based on unsteady friction modeling and Method of Characteristics (MOC) is developed for the pipeline at hand. Then, the model is exploited to generate datasets containing the transient hydraulic responses at the measurement points. Afterward, the most efficient features and optimum SVM algorithm are selected through sensitivity analysis. To evaluate the performance of the proposed model, an experimental reservoir-pipe-valve system is constructed in the Hydraulics Lab of the Shahid Chamran University of Ahvaz. The model is finally applied to the case study, and the impact of applied kernels, size of datasets, and the length of the applied response signal are investigated. The results indicated that the model has high performance and could detect leaks accurately.

Leak detection, Support Vector Machine, Classification, Feature selection, Transient analysis