Abstract: Focus on the problem of multi-source fault diagnosis,and complex decision-making in the fault diagnosis process of aircraft integrated drive generator,from the view of data mining,the variable precision rough sets is applied to process the diagnosis information system,and the system decision is reduced.The decision theorem of the approximate coordination set of β under variable precision rough set is given using binary relation and dependency space defined by object set,with determining the approximate classification quality γ to determine the range of β values and further construct the approximate distribution table,and the fault diagnosis method for aircraft integrated drive generator is proposed,which is based on β lower approximation attribute reduction algorithm.Through the actual generator output fault signal diagnosis process to reduce the decision-making system,the validity of the method and the suitability of the decision method which obtain from a small sample are verified by the accuracy and coverage of the right boundary region of the rough set.