Based on replicate results of liquid chromatography-mass spectrometry(LC-MS),a statistical learning model,which combined time difference and peak shape similarity,was proposed in this paper to solve the problems of low matching accuracy and low coverage of peptide chain alignment.A time difference statistical model was built,which focused on the statistical characteristics of time shift.However,only based on the time feature,the error of alignment could not be eliminated completely.Besides the time,peak shape feature was also introduced in this paper under the hypothesis that the same peptide chain would produce similar LC peaks in repeated experiments.This model was also tested by testing peptide sequences signal.Results showed that the accuracy of the proposed method could reach 98.3%.The coverage for the union of the two datasets could achieve 91.0%.The peak shape similarity model could improve the final result of the time model,helping to confirm the corresponding peak pair in LC-MS replicates data.