Leakage in urban water supply systems results in huge financial loss. The World Bank estimates the worldwide monetary value of loss water at about 39 billion USD per year (Liemberger & Wyatt, 2019). This is direct water cost; leakage also gives rise to indirect socioeconomic costs as pipe failures may disrupt businesses, interrupt production processes, cause floods, and damage properties (Huang et al., 2020). In addition, leaks are potential entry points for contaminants and thus lead to risks to public health (Colombo et al., 2009;. Therefore, a fast, accurate and low-cost leak localization method is of desperate need.Fluid transient-based defect detection methodology is a promising general approach for estimating leaks and other defects and has been intensively investigated in the last three decades. This approach actively introduces hydraulic waves, measures pressure response at specified location(s), and analyzes the measured signals to identify and localize defects in a pipe system. This approach can be divided into two classes: (i) the model-data matching methods and (ii) the signal feature methods. The model-data matching methods estimate the concerned parameters (such as leak location and size) by matching pressure measurement with either (i-a) time-domain model (