Chaotic Lévy Whale Optimization Algorithm with Simulated Annealing and Differential Evolution

Abstract

Typical Whale Optimization Algorithm (WOA) suffers from slow convergence, early trapping into local optima, and weak exploration ability. This work aims to solve these problems by combining Differential Evolution (DE), chaos theory, Lévy flight, and Simulated Annealing (SA). This work designs a Chaotic Differential Whale optimization based on Simulated annealing and Levy flight (CDWSL). CDWSL improves randomness of solutions, and reduces the possibility of falling into local optima. CDWSL is evaluated with ten typical functions, two composite ones and a real-world one of computation offloading in vehicular edge computing. In addition, experiments on higher- dimension problems are also conducted to evaluate the performance of CDWSL. The comparison of experimental results shows that CDWSL achieves superior performance over its typical state-of-the-art peers in solving both low-dimension and high- dimension simple benchmark problems. Furthermore, CDWSL yields better results for both composite benchmark functions and the real-world computation offloading problem than some typical algorithms.