Abstract: To solve the problem of topology similarity calculation of process plant models,a similarity calculation method based on cumulative distribution function has been presented.Firstly,the method mines the underlying topology semantic from topology structure through HMM modeling and state sequences reference; then with the LDA model,the topology semantic is mapped to length fixed feature vector; finally,all the feature vectors are described by a probabilistic model and the similarity between two topology structures is calculated based on the cumulative distribution function.According to the specialized knowledge and experiences from process plant professionals of model designing and modeling,the accuracy and effectiveness of the presented method is verified.