Adaptive Particle Swarm Optimization Assisted MCMC for Stochastic Inversion

MCMC is widely applied in Bayesian inversion theorem. The advantage of Markov Chain Monte Carlo (MCMC) is that it provides a stochastic optimization approach. The unknown parameters of impedance inversion are extremely high dimensional, ill-posed and non-uniqueness. Global optimization is one of the approaches to update optimization system and find the global minimum, which heavily depends on optimization algorithms. Compared with typical metaheuristic optimization methods, adaptive particle swarm optimization (APSO) has the ability to improve efficiency and can be applied in stochastic inversion issues. We propose APSO assisted MCMC method to solve seismic inversion and output the global optimum solutions. Besides, we test the algorithm on prestack data from the North Sea and compare with the calibrated well-log data, which proves that this method has a high performance on seismic inversion issue.