Enhanced Multi-objective Artificial Physics Optimization Algorithms for Solution Set Distributions

Abstract

To optimize the distribution of solution sets in multi-objective optimization algorithms, this study takes the artificial physics optimization algorithm as an example, and introduces the elite learning, inverse learning, bi-directional speed, and chaotic mutation strategies to guide the evolution and search of individuals. This study guides the algorithm to carry out multi-objective solving from three aspects: external archive set, optimal guided individual selection, and step size optimization. The experimental outcomes denote that the improved artificial physics optimization algorithm designed in this study has the best minimum convergence value, standard deviation, optimization times, and convergence performance on the test function. The population fitness curve performs well. At the same time, the super volume index is 0.958, and the solution running time on different test functions is 6.13s, 7.61s, 8.46s, 9.68s, and 10.77s, respectively, with the smallest value. The improved multi-objective artificial physics optimization algorithm achieves the lowest Generative Distance value of 0.126, the lowest Inverted Generative Distance value of 0.171, the lowest extensiveness evaluation value of 0.210, and the maximum distributiveness evaluation value of 0.989. The actual solution coverage is high. This study expands the solution ideas and methods for multi-objective optimization problems, significantly improving the performance of artificial physics optimization algorithms in solving multi-objective optimization problems.