Abstract: In order to explore the business strategy of multi-model product supply chain,this paper used Multi-Agent modeling and simulation method to establish a multi-model home vehicle supply chain network model.Aiming at maximizing the profit of manufacturer Agent,stimulus-response learning mechanism and the particle swarm learning mechanism are used to analyze the evolution of the model.The simulations show that,firstly,by using the stimulus-response learning mechanism,the sales volume of some models of the manufacturer will be greatly reduced.For manufacturers pursuing multiple models,it is not conducive to the promotion of multiple models,but the total sales volume and profits will increase.Secondly,by using the particle swarm learning mechanism,no matter what combination of C₁ and C₂ is,it is difficult to obtain the optimal sales and profit at the same time.For mass-consumer models,manufacturers can choose to strengthen C₂ and expand the market with “small profits but quick turnover” strategy to increase sales.For high-end consumer models,manufacturers can choose to strengthen C₁ and create high-quality products with a “benefit and profitable” strategy to increase profits.Relatively speaking,by using particle swarm learning mechanisms,manufacturers can quickly adjust strategies to cope with market changes,and the strategies are more stable after learning.This study has practical guidance for supply chain management with multiple models.

Key words: Supply chain, Multi-vehicle, Stimulus-response theory, Particle swarm optimization, Evolution analysis