Computational source term estimation of the Gaussian puff dispersion

Abstract

The hazardous or toxic chemical releases have a detrimental impact on public safety. Estimating source parameters is of particular importance in aiding emergency response and post-assessment. Source term estimation from sensor measurements with a given Gaussian puff dispersion model is a typical inverse problem, which can be transformed into an optimization problem. In this paper, we employed the particle swarm optimization, the Nelder–Mead method, and their hybrid method to solve the optimization problem. Furthermore, we proposed a three-dimensional neighborhood topology which considerably improves performance of the particle swarm optimization. We implemented all these algorithms in JAVA on an embedded system to make a preliminary estimation of the accidental puff release. Numerical experiments with synthetic datasets show that the particle swarm optimization maintains a balance between computation time, accuracy, robustness, and implementation complexity. In contrast, the hybrid algorithm has an advantage in computation time at the expense of more sophisticated implementation.