Paper N. On-line Dispersion Source Estimation using Adaptive Gaussian Mixture Filter. Edited version of the paper: M. F. Huber. On-line Dispersion Source Estimation using Adaptive GaussianMixture Filter. In Proceedings of the 19th IFACWorld Congress, pages 1059-1066, Cape Town, South Africa, August 2014

The reconstruction of environmental events has gained increased interest in the recent years. In this paper, the focus is on estimating the location and strength of a gas release from distributed measurements. The estimation is formulated as Bayesian inverse problem, which utilizes a Gaussian plume forward model. A novel recursive estimation algorithm based on statistical linearization and Gaussian mixture densities with adaptive component number selection is used in order to allow at the same time accurate and computationally efficient source estimation. The proposed solution is compared against state-of-the-art methods via simulations and a real-word experiment.