Enhanced Leakage Detection and Estimation via a Hybrid Genetic Algorithm and High-Order Sliding Modes Observer Approach

This paper introduces a hybrid approach designed for both detecting and estimating the magnitude of leaks in oil pipelines. The method integrates a High Order Sliding Mode Observer with a Super Twisting Algorithm to serve as an observer for state estimation of the system. A parameterized model based on momentum and mass balance equations with discretization is used, where the parameters are the location and magnitude of the leakage. To find these parameters, it incorporates the Genetic Algorithm to solve an optimization problem that relies on a function cost related to the error norm between measurements and states estimation from HOSMO in order to measure the difference between the model with an assumed leakage and the real leakage. The solution of the minimization problem represent the leak position and magnitude. The feasibility and effectiveness of this method is evaluated using a simulation model representing a 306 km sector of the North-Peruvian Oil Pipeline. The results demonstrate its robustness against noise, showcasing a precision of ±250 m in pinpointing the location of leaks.