Data Fusion and State Estimation Using Belief Propagation in Gas Distribution Networks

This paper proposes a solution to the state estimation problem in gas networks using the distributed belief propagation (BP) algorithm. Power system identification applications require precise and robust state estimatiors as well as various sensor information. Compared to augmenting the power system with a very large number of sensors, a limited number of sensors and probabilistic graphical models can be used to infer the system state and reduce hardware investments. A novel BP algorithm propagates the pressure quantities at nodes in the gas network based on pressure manometer signals and applies a correction based on the information of neighboring nodes in the fusion step by using additional supporting sensors. Finally, the data fusion algorithm is demonstrated for a 14-node gas distribution network based on real data. This paper presents a novel algorithm aimed at tackling the traditional weighted least squares method to validate the developed novel approach in order to highlight the advantage of the distributed inference algorithm over traditional methods.