Comparison of damped least-squares inversion and particle swarm optimization (PSO) on magnetic data

Modeling is one of the main methods used for the interpretation of physics. The method used is a literature review of the material relating to the method of geomagnetic and damped Least-Squares inversion and inversion discussion on the global approach, namely Particle Swarm Optimization (PSO). The parameters will be inverted magnetization is the value that will be used to obtain information susceptibilities. The modeled geometry is anomalies of horizontal crustal block continue. The data used in the form of synthetic data obtained from the forward modeling process. From the data, the inversion process is carried out to obtain the appropriate model parameters. The objective of this study is the comparison of the accuracy of the damped Least-Squares inversion method and PSO. The data from forward modeling is then varied with the addition of 20% noise in each data. In damped Least-Squares inversion, the value of certain λ varied so that the corresponding data and subsurface models are obtained. In PSO inversion, parameters value such as inertia weight (w), velocity coefficient (c), and number of particles are varied. The results obtained in this study is the damped Least-Squares inversion λ between 10E-08 to 10E-14 passably in determining subsurface model solutions, especially on the data without noise. PSO inversion results for both are used in the data without or with the addition of noise as it is a global approach, the number of particles required is 40-250 with iterations between 100-200, to the value of inertia weight (w, linear inertia weight and inertia constant weight) and velocity coefficient (c1 and c2) are respectively 0.4-09 and 0.3-1.4986 with c2 is larger than c1. Least-Squares inversion is good enough to be applied using attenuation in this study; the drawback is that when the data is the noise the Least-Squares inversion will be random, the accuracy depends on the value of λ used. Whereas the PSO inversion is very difficult to determine the values of the parameters because this PSO method is a global method inversion.