Abstract: An algorithm based on grey model to predict idle period for DPM is proposed in order to improve the accuracy of predicted results.The algorithm needs only a few samples for prediction and does not have strict requirements on the quantity and distribution characteristics of samples.A self-adaptive prediction is implemented to promote the accuracy of predicted results by adjusting parameters dynamically.Experimental results show that the predicted results by use of the algorithm approximate well to real ones and are more accurate compared to algorithms based on the exponential average method.