A novel improved chef-based optimization algorithm with Gaussian random walk-based diffusion process for global optimization and engineering problems

Abstract

The chef-based optimization algorithm (CBOA) is a human-based method inspired by the relationship between culinary students and chef instructors. The original CBOA does not have a process that separately controls population generation and updating. For this reason, in this study, a novel improved CBOA with diffusion process (DP) algorithm with the Gaussian walk (CBOADP) is proposed. In this way, it is aimed to transfer promising members to the next iteration. In the original CBOA algorithm, the selection of the chefs is performed randomly. The oppositional-based learning (OBL) approach aims to find the current best chef by updating the chef from the chef candidates selected from opposite regions. Because the selection of the chef plays an important role in the selection of students. To develop the method with optimal parameters, parameter analysis is performed as the DP’s walk and maximum diffusion parameters. The new method’s performance is evaluated through the CEC2019 and CEC2022 test suites and traditional and advanced methods. Results for real design problems are also reported. All results are supported by the Wilcoxon sign and Friedman rank statistical tests. As a result, in general, the best convergence has been achieved with CBOADP in about 75% of the benchmark test data, and statistically, it has first rank among other methods on average. In addition, CBOADP reached the best fitness value in five of the six compared engineering problems. Accordingly, it is seen that DP and OBL approaches contribute positively to the performance of CBOA and show a successful convergence trend among other compared methods.