A Model-Based Bayesian Framework for Pipeline Leakage Enumeration and Location Estimation

Leak detection in pipelines is an important issue, because leakages pose financial losses, environmental pollution and even health risks. The paper considers the problem of detecting multiple leaks based on transient wave theory for a reservoir pipeline valve system. The given measured data under consideration may contain one or multiple leaks originating from different locations when estimating the leak locations. This leads to two problems to be solved: first determine the correct number of leaks, and then identify the actual location of each leak. Thus, a probabilistic method of model-based Bayesian analysis is applied to this paper. This work employs a model to describe various scenes, individually defined by a specific number of leaks and their locations. Bayesian inference is used to select which model that is the most appropriate to fit the measured data. Through the process, the number of leaks is first estimated, and then the leak locations are extracted from the model that the measured data prefers. This paper presents different experimental setups and scenarios to demonstrate the availability of the proposed method, demonstrating that this model-based Bayesian analysis is an accurate tool for leakage enumeration and location estimation.