Optimization Algorithms, Nodari Vakhania, Editör, Open Text Corporation , London, ss.84-104, 2021
It has been recently revealed that particle
swarm optimization (PSO) is a modern global optimization method and it has been
used in many real-world engineering problems to estimate model parameters. PSO
has also led as a tremendous alternative method to conventional geophysical
modeling techniques that suffer from initial model dependence, linearization
problems, and being trapped at a local minimum. One area that is neglected in the
use of PSO is the joint modeling of geophysical datasets that have different
sensitivities, whereas this kind of modeling with multiobjective optimization
techniques has become an important issue to increase the uniqueness of the
model parameters. However, the use of subjective and unpredictable weighting to
objective functions can lead to a misleading solution in multiobjective
optimization. Multiobjective PSO (MOPSO) with the Pareto approach allows to
obtain a set of solutions that contains a joint optimal solution without
weighting requirements. This chapter begins with an overview of the PSO and Pareto-based
MOPSO and presents their mathematical formulation, algorithms, and alternative
approaches used in these methods. The chapter goes on to present a series of synthetic
modeling of seismological data, which is a type of geophysical data by using Pareto-based
multiobjective PSO. According to results matched perfectly, we believe that
multiobjective PSO is an innovative approach to joint modeling of such data.