In order to better solve the problem of slope subsidence prediction,this paper focuses on the overall change trend of slope,and uses multivariate optimization MGM (1,n) model to realize the prediction of sedimentation deformation of the monitoring points affecting the interconnected effects of slope. The MGM (1,n) optimization model of improved double value is established by filtering the initial value by means of the average relative residual index and reconstructing the background value calculation formula by introducing the background value coefficient. Based on the measured data of the A-section slope of the Central Shanxi Expressway,the optimized MGM (1,n) model,the traditional MGM (1,n) model and the GM (1,1) model were established,and the measured data were modeled and predicted. The results show that optimizing the MGM (1,n) model has higher prediction accuracy and some application prospects in slope subsidence prediction,which can provide reference for future research.