The invention belongs to the technical field of dynamic knowledge updating, and particularly relates to a dynamic knowledge updating method based on three decisions in a variable-precision rough set model, which comprises the following steps: 1, setting a threshold value, and selecting data precision in a rough set; 2, selecting a data relationship in the rough set according to a set precision threshold value, and carrying out data region judgment; 3, generating fuzzy set data in the rough set after judging that the data is finished, and setting an incommitment rule for the fuzzy set data; 4,selecting an approximate space, substituting the generated fuzzy set data into the approximate space, and judging the regional value of the data again; and 5, performing intra-region data comparison on the data generated after judgment, selectively discarding the generated repeated information to obtain updated data, and through three decisions, effectively reducing data updating vulnerabilities caused by two decisions, reducing repeated data of dynamic knowledge updating, and facilitating subsequent updating.