The screening of traditional antioxidant peptides often needs enzymatic hydrolysis， purification， separation， identification and in vitro experimental verification， which costs a lot of time and material resources. Based on bioinformatics， this study used computer virtual digestion， prediction and evaluation of related activities， pharmacokinetics and molecular simulation techniques to screen antioxidant peptides from bovine whey protein in turn， analyze and explain the possible mechanism of Keap1-Nrf2-ARE antioxidant pathway， and analyze the stability of conformation in the process of molecular reaction. Using the Peptideranker program to predict and score the biological activity， 137 peptides with a score greater than 0.4 were screened. At the same time， Toxinpred， Innovagen， Expasy-compute and Pepdraw programs were used to further predict and analyze the physical and chemical properties of the peptides. Through pharmacokinetic ADMET and TOPKAT analysis， 14 peptides with no toxicity， no sensitization， no mutagenicity， no carcinogenicity and good water solubility were screened. Then it was docked with Keap1 receptor， and the conformational mechanism of the polypeptide with the top four docking scores was analyzed with the optimal binding complex of Keap1. Finally， through the evaluation of the intermolecular force of the complex conformation， it was concluded that the CDEF sequence had the most antioxidant activity potential， and the CDEF-keap1 receptor complex was simulated by molecular dynamics. The results showed that the conformational change was relatively stable in the process of dynamic simulation. Therefore， CDEF antioxidant peptides could play a better role in antioxidant activity in human body. Compared with traditional enzymatic hydrolysis methods， this study could quickly and efficiently screen antioxidant peptides， and provide a new idea for the rapid screening of antioxidant peptides from natural food sources.