A novel intelligent fault diagnosis model for flue gas turbine based on EMD (Empirical mode decomposition) and VPRS (Variable precision rough set) theories, is proposed in order to solve the difficult problems of knowledge information acquisition and improve fault diagnosis accuracy in practice. This model combines EMD and VPRS techniques. First EMD signal processing technique is employed to excavate the underlying fault information from dynamic signals. The features that reflect the equipment operation conditions from the EMD analysis of the dynamic original vibration signals are extracted and the series of IMFs (intrinsic mode function) feature sets are obtained. Then the energy features of the calculated IMFs are using as the condition attributes of the knowledge acquisition decision table while the fault modes are using as the decision attributes respectively. The decision table is deal with through the attributes' reduction, attributes' value reduction and the rules' reduction based on VPRS theory. The system fault diagnosis rules are extracted on the condition that the model classification ability remains and the redundancy information is removed. The model is applied for the flue gas turbine diagnosis knowledge acquisition and fault diagnosis in Yanshan. The desired diagnosis effect is obtained via the fault diagnosis model based on EMD and VPRS. Moreover, the application result also validates the power and the practice of the model.