Gas pipeline leakage detection in the presence of parameter uncertainty using robust extended Kalman filter

Abstract

One of the most significant systems that can be expressed by partial differential equations (PDEs) is the transmission pipeline system. To avoid the accidents that originated from oil and gas pipeline leakage, the exact location and quantity of leakage are required to be recognized. The designed goal is a leakage diagnosis based on the system model and the use of real data provided by transmission line systems. Nonlinear equations of the system have been extracted employing continuity and momentum equations. In this paper, the extended Kalman filter (EKF) is used to detect and locate the leakage and to attenuate the negative effects of measurement and process noises. Besides, a robust extended Kalman filter (REKF) is applied to compensate for the effect of parameter uncertainty. The quantity and the location of the occurred leakage are estimated along the pipeline. Simulation results show that REKF has better estimations of the leak and its location as compared with that of EKF. This filter is robust against process noise, measurement noise, parameter uncertainties, and guarantees a higher limit for the covariance of state estimation error as well. It is remarkable that simulation results are evaluated by OLGA software.

An integral monitoring concept for data‐driven detection and localization of incipient leakages by fusion of process and environment data

“…One investigated method is an extended Kalman filter as a model‐based approach, which performs robust against noise and parameter uncertainties. [ 46 ] Further, a fusion of model‐based and model‐free approaches including simple limit checking as well as a Kalman filter for FDI has been proposed for incipient leakage detection. [ 47 ] Despite of some successful results applied on a two tank‐system, the detection of incipient leakages suffered from measurement noise and is therefore not suitable for an industrial application.…”

Robust extended Kalman filtering for nonlinear systems in the presence of unknown inputs and correlated noises

“…The leakage occurs at different distances and diameters for 20 minutes. To detect the leak and its location, the method presented in Reference 39 is used.…”

Locating leakage in pipelines based on the adjoint equation of inversion modeling

“…The leakage of the pipelines increases the risk of contaminant entry and further increases the projected expenditure of resources and costs. Pipeline leakage poses potential risks to public health and causes financial losses [ 1 , 2 ], while also burdening the environment by wasting energy [ [3] , [4] , [5] , [6] , [7] , [8] ]. These phenomena have gained worldwide attention.…”

“…Nowadays, there are many advanced and mature leak detection and location technologies for long straight pipelines (e.g., long oil pipelines, natural gas pipelines) [ [3] , [4] , [5] , 7 , 9 ], such as acoustic and infrared detection methods, negative pressure wave method, transient model method, distributed optical fiber method, etc. Among these technologies, the transient analysis method exhibits more advantages in leakage detection and pipeline location [ [10] , [11] , [12] , [13] , [14] , [15] , [16] ].…”