Abstract: A fuzzy neural network model based on entropy clustering is applied to analyze shaft parts.The model utilizes the clustering method to partition the fuzzy input space and to extract the fuzzy IF-THEN rules.The gradient descent algorithm is used to classify and refine various characteristic parameters of shaft parts.Experimental results show that the presented method is feasible and effective to analyze shaft parts clustering,and is more adaptive than the common clustering methods to group shaft parts with many varieties and complex types.