Abstract:Aiming at the accuracy of the grey model in the middle and late subsidence prediction, a dual-weight power function weakened buffer operator is proposed, and an improved variable-weight buffer GM(1,1) model is established combined with the variable-weight background value. It is proved that the new operator can improve the smoothness, reduce the scale deviation, realize the preprocessing of the original sedimentation data, and give the condition that the reduction error does not expand. The weights in the model are determined by the particle swarm optimization algorithm with the objective of minimizing the average relative error between the restored value and the original value. Finally, an airport high-fill foundation settlement data prediction is used as an example to verify the effectiveness of the proposed improved model. The results show that the new operator can effectively weaken the impact of the subsidence sequence, and the combination of the variable weight background value can significantly improve the accuracy of the gray model in the middle and late subsidence prediction. In addition, after the weight of the new operator is limited, two different variable weight weakening buffer operators can be obtained.