An improved particle swarm optimization for history-matched reservoir parameters

In this work, we propose an improved PSO algorithm that uses an averaging filter to reduce the number of function evaluations required to locate an optimal point. The method is applied to estimate history-matched reservoir parameters on two test problems and comparison is made with the standard PSO and the distance-weighted exponential natural evolutionary strategies (dxNES) (Fukushima et al. 2011). Results from the comparisons show that the proposed algorithm outperforms the other algorithms.