CIDO: Chaotically Initialized Dandelion Optimization for Global Optimization

Abstract

Metaheuristic algorithms are widely used for problems in many fields such as security, health, engineering. No metaheuristic algorithm can achieve the optimum solution for all optimization problems. For this, new metaheuristic methods are constantly being proposed and existing ones are being developed. Dandelion Optimizer, one of the most recent metaheuristic algorithms, is biology-based. Inspired by the wind-dependent long-distance flight of the ripening seed of the dandelion plant. It consists of three phases: ascending phase, descending phase and landing phase. In this study, the chaos-based version of Chaotically Initialized Dandelion Optimizer is proposed for the first time in order to prevent Dandelion Optimizer from getting stuck in local solutions and to increase its success in global search. In this way, it is aimed to increase global convergence and to prevent sticking to a local solution. While creating the initial population of the algorithm, six different Chaotically Initialized Dandelion Optimizer algorithms were presented using the Circle, Singer, Chebyshev, Gauss/Mouse, Iterative and Logistic chaotic maps. Two unimodal (Sphere and Schwefel 2.22), two multimodal (Schwefel and Rastrigin) and two fixed-dimension multimodal (Foxholes and Kowalik) quality test functions were used to compare the performances of the algorithms. When the experimental results were analyzed, it was seen that the Chaotically Initialized Dandelion Optimizer algorithms gave successful results compared to the classical Dandelion Optimizer.