As to the issue of predicting real-time aerial target track， grey prediction model is an effective method in the context of limited data.But the accuracy of grey model prediction often fails to meet the requirements in practical engineering applications.Through in-depth study on the modeling mechanism of grey mode， it was found that the selections of initial value and grey parameters are not optimal.Then， by using ant colony optimization for the initial value and grey parameters， the accuracy of track prediction was improved significantly.The effectiveness and precision of the method was verified through case calculation.