多策略融合改进的梦境优化算法OA

To address the limitations of the dream optimization algorithm(DOA)in initial population uniformity,exploration adaptability,and exploitation-stage global escape capability,this paper proposes a Multi-strategy fusion Improved Dream Optimization Algorithm(MIDOA).First,piecewise linear chaotic mapping combined with Quasi-Opposition-Based Learning is employed for population initialization to generate high-quality uniformly distributed individuals.Second,an adaptive search mechanism based on dream state classification is developed by monitoring stagnation degree to enable adaptive search intensity adjustment.Finally,an adaptive Lévy flight strategy is incorporated in the exploitation phase through a multi-strategy jumping mechanism to enhance escape ability from local optima.Experimental results on CEC2017 benchmarks show that MIDOA achieves optimal optimization accuracy and stability simultaneously on 6 functions.Compared with WOA,GWO,SCA,and SSA,MIDOA obtains optimal results on over 70％of test functions,with convergence accuracy improved by 1～2 orders of magnitude.The Wilcoxon rank-sum test validates statistical significance.The multi-strategy fusion approach effectively enhances convergence speed,optimization precision,and global search capability.