Levy Flight Dream Optimization Algorithm: A Modified Version of Dream Optimization Algorithm

Abstract

This manuscript elucidates a novel Modified Dream Optimization Algorithm (mDOA). The foundational framework of the Dream Optimization Algorithm (DOA) is informed by the cognitive phenomena associated with human dreaming. These cognitive mechanisms (memory retention, forgetfulness, supplementation strategies, and dream sharing processes) are systematically encoded as an optimization agent designed to tackle global optimization dilemmas. The DOA is afflicted by the challenge of an imbalance between exploration and exploitation, exhibiting a higher propensity for exploration than for exploitation, which results in an elevated likelihood of becoming ensnared in local optima. The enhancement of mDOA was achieved through the integration of a Levy flight variant to boost the exploitation phase. The efficacy of mDOA is evaluated against six prominent metaheuristics utilizing ten benchmark test functions (Schwefel, Ackley, Michalewicz, Griewank, Pathologic, Rastrigrin, Rosenbrock, Schaffer, Sphere, and Bohachevsky1), it demonstrated 85% enhancement in its convergence towards global optima. From the simulation results obtained, it shows that the mDOA succeeded in attaining the optimal global solution in 7 out of 10 cases, constituting 70.0% of the benchmark functions. Conversely, the 0ther algorithms used achieved 3 out of 10 cases, representing 30.0% of the benchmark functions. These shows an improvement in the mDOA.