Abstract: This paper presents a rule-driven structure evolution method to generate lightweight models.Firstly,each voxel with its neighborhood is defined as an intelligent agent,which make it possible to convert the structural assembly problem into an optimization problem of a multi-agent system;Secondly,in order to control the intelligent agents with incomplete knowledge and constraints,we introduced the defeasible logic to define rules;Finally,the optimized structure is generated by the group behavior.Experimental results showed that the proposed method can be used to generate target structure with different loads.In addition,unique designs can be achieved under the same load according to different rules.