In this paper, lot-splitting is adopted in the process of Seru production system formation for the inter-Seru imbalance problem caused by unreasonable batch size in Seru production. A multi-objective optimization model of line-Seru conversion is developed. Aiming to minimize the total throughput time and the total labor hours, by the model makes joint decisions on Seru formation and lot-splitting in the conversion process. Considering the characteristics of the multi-objective optimization problem and the complexity of the solution space, a local search strategy is added to improved the algorithm, the INSGA-II is proposed to solve the problem, and an encoding method based on vernier is designed to express the lot-splitting strategy to satisfy the constraint when Seru quantity changes. In evolution, the batch sizes and product types of each Seru are continuously optimized to balance the workload between Serus and shorten total throughput time. Finally, the effectiveness of the method is verified by numerical examples. The results show that lot-splitting can better balance the workload between Serus when the quantity of product has high fluctuation.