Abstract: Due to the non-equivalent processing capacity and working time among multiple heterogeneous mixed-model assembly lines (MMALS) in the bus manufacturing process, a parallel non-equivalent production planning model for flexible customization is proposed. By analyzing the interactions between the order decomposition and production sequencing, a leader-follower joint optimization model that takes the order decomposition as the leader and production sequencing as the follower is proposed, in order to optimize the product urgency, matching degree and production line load. A bi-level interactive genetic algorithm with Pareto front-line solution is developed to solve the model. To improve the performance of the genetic algorithm, an adaptive adjustment method is introduced to guarantee the crossover and mutation probabilities and niche technology to ensure the diversity of population. The proposed model is verified using cases in MMALS of buses and compared with other two methods, namely, the multistage genetic algorithm and the actual scheme of the enterprise. The proposed model that uses the bi-level interactive genetic algorithm can represent the actual situation of the enterprise and maximize the efficiency of MMALS.

Key words: order decomposition, production sequencing, flexible customization, leader-follower joint optimization, non-equivalence