Abstract: A dynamic multi-row facility layout optimization model and algorithmic framework is developed to address the limited handling of dynamic factors and unstable optimization performance in manufacturing workshops. A fuzzy stochastic material flow is generated by introducing normal distribution perturbations to capture the uncertainty of material demand. Based on this, a multi-period mixed-integer programming model is formulated considering material handling cost, facility rearrangement cost, and land occupation cost. An improved particle swarm optimization (PSO) algorithm is proposed, where chaos-based initialization enhances search diversity, dynamic adjustment of inertia weight and learning factors balances global and local search, and an improved neighborhood update mechanism based on layout structure increases convergence accuracy. Simulation results show that the improved PSO achieves higher stability and faster convergence, reducing the average number of iterations by about 30%. A case study confirms the feasibility and effectiveness of the model and algorithm, with the total cost reduced by approximately 10.8% compared with static layouts. The method demonstrates better global search capability and computational efficiency than existing metaheuristic algorithms, providing an effective approach for dynamic facility layout optimization in manufacturing systems.