An adaptive gravitational search algorithm for optimizing mechanical engineering design and machining problems

The gravitational search algorithm (GSA) is widely used for solving optimization problems because it performs in a superior manner as compared to various competing evolutionary and swarm-based metaheuristics. However, GSA frequently gets trapped in local optima due to a lack of solution diversity. Although chaotic gravitational search algorithm (CGSA) can resolve this issue to some extent but its degraded exploitation rate and convergence speed may not result in desired outcome. To this end, diversity-based chaotic GSA (DCGSA) has exhibited its capability to resolve the issues encountered with GSA and CGSA to a certain extent for unconstrained problems. DCGSA achieves this characteristic through the use of an adaptive gravitational constant, which varies according to the diversity values of the population. Since most real-world problems are subjected to some constraints, it is prudent to improve the performance of GSA in solving constrained optimization problems. The present study integrates the enhanced search capability of DCGSA with a generalized constraint handling mechanism to improve the performance of GSA in solving constrained problems. It is observed that DCGSA significantly outperforms GSA on both CEC (Congress on evolutionary computation) 2006, 2010 and 2017 functions and competes strongly with CGSA. Diversity analysis shows that the capability to balance exploration and exploitation rates is enhanced using CGSA and DCGSA. Furthermore, DCGSA algorithms outperform GSA and CGSA on real-world machining and CEC 2020 mechanical design problems. Comparison with state-of-the-art algorithms is made to analyze the performance of the algorithms from a larger perspective.