This paper introduces a new algorithm, Adaptive Dual-Population Particle Swarm Optimization (ADP-PSO), which is designed to address complex, high-dimensional, and highly constrained optimization problems. ADP-PSO integrates a dual-population framework, where one subpopulation emphasizes convergence speed using a Von Neumann topology, and the other focuses on solution quality through stochastic learning mechanisms. A novel constraint-handling strategy combining boundary repair and adaptive penalty functions is also incorporated. The algorithm’s efficacy is validated on a real-world irregular flight recovery problem, exhibiting significant improvements in convergence speed, solution quality, and robustness compared to traditional PSO and recent variants like VPPSO.