Particle swarm optimization (PSO) is a classic algorithm in the field of swarm intelligence. Despite its widespread use, PSO faces challenges in complex optimization scenarios, particularly its propensity for falling into local optima and slow convergence. Comprehensive learning particle swarm optimization (CLPSO) has enhanced PSO’s global search capabilities by introducing a comprehensive learning mechanism. This study presents further enhancements by proposing the dual population adaptive strategy comprehensive learning particle swarm optimization (DPAS-CLPSO), which combines a dual population framework with an adaptive learning paradigm to increase population heterogeneity and algorithm search efficiency. Experimental results on the CEC2017 standard benchmark tests have shown that DPAS-CLPSO significantly outperforms several well-known algorithms in both 30D and 50D problem spaces. Lastly, the effectiveness of the algorithm in solving complex optimization problems is also underscored by statistical validation through rank sum tests.