(476d) The Square-Root Unscented Kalman Filter for Leak Detection and Location In Natural Gas Pipelines

Pipelines are the most efficient and economical way to transport and distribute natural gases all over the world. Leaks are the main problem of pipeline, which may cause excessive economical losses, or even worse, pose health hazards to communities. Therefore, designing a timely and reliable leak detection system is essential to maintain the normal operation of the pipeline company. In this work, a square-root unscented Kalman filter (SR-UKF) is proposed for leak detection and identifying location of the leak. The gas pipeline is modeled as a couple of quasi-linear hyperbolic partial differential equations (PDEs). Method of characteristics (MOC), which is suitable for large scale systems or systems with complex boundary conditions, is used to solve the quasi-linear hyperbolic PDEs. The proposed SR-UKF is applied to estimate the leak size and the location of the leak. The SR-UKF has the advantage of ease of implementation and has superior numerical stability. Comparative studies of extended Kalman filter (EKF) and the proposed SR-UKF are carried out to show the superiority of the proposed method.