To deal with the complexity and uncertainty of wind turbine fault information， a vagueness method based on the belief measure and plausibility measure of random set is proposed. Firstly， it makes the sampled mode and detected mode vagueness， and matches the sample mode with detected mode through belief measure and plausibility measure to realize data fusion. Then the matched result is used as belief function and plausibility function of evidence theory to realize feature fusion. Finally， the fusion result is used as evidence to be fused in the decision-making step， thus it can give a precise diagnosis result. The wind turbine example verifies the rationality and effectiveness of the proposed method.