An Improved Differential Evolutionary Algorithm Based On Simulated Annealing And Levy Flights Mechanism

The Differential Evolutionary algorithm (DE) is a well-known and effective metaheuristic method for continuous optimization problems. However, due to contraction stagnation and early convergence problems, the performance of the DE algorithm can be influenced at the convergence speed and optimization accuracy. Therefore, this paper proposes an improved DE algorithm with the Levy flights mechanism and the Simulated Annealing algorithm (SA), namely DESA-LF algorithm, which effectively uses the advantages of them with generating the better candidate solutions to reduce the convergence speed for making the optimal solution towards the global optimal direction. Finally, we use six typical functions with different complexities and make the comparisons with the DE algorithm, the DE-LF algorithm and other meta-heuristic algorithms to verify the efficient of DESA-LF. From the results and convergence curves, the DESA-LF has the advantages of extending the convergence speed and enhancing the optimization accuracy. Therefore, it demonstrates that convergence speed and optimization accuracy of DESA-LF are highly efficient and superior.