This study addresses the distributed flow-shop scheduling problem (DFSP) and aims to minimize the makespan and the total processing time. Although many intelligent algorithms have been proposed to solve DFSP, the efficiency and quality of these solutions still need further improvement to meet higher production requirements. Therefore, a hybrid particle swarm optimization with enhanced directional search and Q-learning-based variable neighborhood search is proposed. The directional search quickly explores the particle swarm in multiple directions, which enhances the convergence ability in different areas of Pareto front. The Q-learning-based variable neighborhood local search prevents the proposed algorithm from falling into a local optimum. The comparative results and statistical analysis of the experiments demonstrate the superior convergence and distribution performance of the proposed algorithm.