Application of Global Optimization Using Simulated Annealing and Artificial Bee Colony Algorithm: A Comparative Study on the 2022 Mw 5.6 Cianjur Earthquake

Abstract Determining the hypocenter is vital to the process of analysing an earthquake. In the global optimisation problem to relocate hypocenters, the conventional algorithm often suffers from local minimum and fails to produce satisfactory results. As such, many algorithms have been devised to overcome this problem. This study compares the performance of the very fast simulated annealing (VFSA) algorithm to the artificial bee colony algorithm (ABC) using Cianjur’s 5.6 Mw 2022 earthquake data. The VFSA algorithm mimics the process of cristal-forming metal cooling from a high temperature while adding an exponential annealing schedule to ensure rapid convergence of errors. The ABC algorithm on the other hand models the optimisation process as a bee swarm’s behaviour in searching for food. The earthquake data is fetched from IRIS and synthetic arrival times on five stations are used. Forward modelling is done with homogenous velocity to simplify the process of producing synthetic data. Inversion is then carried out on the same model search space for both algorithms and the results are quantified on 500 simulation runs. It is found that while both results agreed on the same hypocenter that is accurate to the solution obtained from the IRIS data, the VFSA algorithm converges twice as fast as the ABC algorithm with significantly lower RMS error. However, the performance against local minimum is not evaluated, thus may result in different performance compared to other cases. Additionally, a comparison to other studies for both algorithms showed similar results, confirming the accuracy of the algorithms.