Modified Particle Swarm Optimisation (MPSO) for inversion of gravity field due to simple causative mass

Modified Particle Swarm Optimisation (MPSO) is an improved algorithm of Particle Swarm Optimisation (PSO), where the learning factors or acceleration coefficients (c1 and c2) and inertia weight (w) change during iteration as ability of finding the optimal solution can be enhanced. In the MPSO algorithm, a new concept of the velocity of the individual (particle) modification, the evolution of the particle best value (Pbest) and global best value or best value in the group (Gbest), is presented as an acceptable convergence in the MPSO algorithm solutions is found. An advantage of the MPSO over the PSO is that it does not stick: it does not stick to a local minimum giving, then, a premature convergence. We have tested the proficiency of the MPSO and PSO using the theoretical gravity caused by buried sources with simple geometry, such as spheres, horizontal cylinders, and vertical cylinders, with and without added random noise. In comparison with the PSO algorithm, the MPSO inversion gives the most satisfactory results for the noise-free and noise-corrupted theoretical gravity data. We have also applied the MPSO approach for inverse modelling of the five residual gravity anomalies due to various causative mass from the different parts of the world, as the estimated results are compared with other previous researches.