Abstract: To find models including the given partial structure from massive CAD models,a retrieval method based on topology approximation is proposed.Firstly,a uniform measurement method is presented to calculate the similarity of boundary faces of retrieval object and retrieval condition according to the given geometry and topology information of the boundary faces on the CAD model.Secondly,the breadth-first-search-based spanning tree algorithm is applied to find out the initial boundary matching between the retrieval object and the retrieval condition.The proposed topology approximation algorithm takes the initial boundary matching as the input to search for the optimal boundary matching.The similarity meeting the requirement of absolute error between the retrieval condition and the retrieval object is calculated on the basis of the optimal boundary matching.Finally,a partial structure retrieval tool based on UG is developed to validate the proposed method.The experimental results show that the method has higher efficiency and precision.