Introduction   In order to predict settlement and calculate consolidation degree of preloading foundation treatment more accurately, this paper puts forward an Optimal Combined Prediction model on the basis of time series of original cumulative settlement and a method of evaluating the accuracy of the model.

    Method   Based on the traditional prediction model, such as Compertz model, grey G (1, 1) model and Pearl model, an Optimal Combined Prediction model (OCP model) was established and the accuracy of all single models and the Optimal Combined Prediction (OCP model) were tested with the monitoring data of foundation settlement at heat-power-cold energy generation system by using natural gas; Meanwhile, the result of the Optimal Combined Prediction model (OCP model) was compared with three-point method, Asaoka method, hyperbolic curves fitting method.

    Result   That shows: the accuracy of all single models and optimal combination model reaches the good precision level; Compared with three-point method, Asaoka method and hyperbolic curves fitting method, the calculated value of the Optimal Combined Prediction (OCP model) is closer to the real value.

    Conclusion   While predicting settlement and calculating the consolidation degree of preloading foundation treatment, different methods are used, reasonable value of evaluation index about foundation treatment can be obtained. The optimal combination prediction model (OCP model) can provide guidance for preloading foundation treatment construction and controlling process.