Aiming at selecting appropriate parts in the product family architecture for innovating, a method for part analysis and selection in the renewal process of product family architecture is initially proposed. On the basis of parts analysis from the three aspects of cost effectiveness-commonality, performance sensitivity and demand-matching degree, a coupling spatial model CSM is established. With the fuzzy c-means clustering method, the parts were clustered, and the choice domain and the improvement direction were pointed out. Finally, a case study of part analysis and selection in the working device product family of the wheel loader was presented to illustrate the validity of the method.