Abstract:

To predict the settlement of turfy soil subgrade with the method of traditional non-equal interval models, there are a few shortages such as cumbersome calculation and low accuracy for short-term forecast Especially when there is a slope mutation point in the subsidence curve, it is difficult to meet the prediction accuracy requirements. For the above shortcomings, average step length conversion is used for non-equal interval sequence to obtain quasi-equal interval sequence; the improved quasi-equal interval QGM(1,1) forecasting models are obtained by amending the original settlement data with the method of linear interpolation; and the subsidence curve is divided into two parts to carry out the prediction at the slope mutation point. As is illustrated in the examples, when the improved quasi-equal interval QGM(1,1) models are used to predict turfy soil subgrade settlement, the average errors between predicted and measured values are 2.99% and 0.25% respectively, far less than the results obtained by means of traditional non-equal interval models. The quasi-equal interval QGM(1,1) model with pretty high accuracy for short-term prediction can serve as a reliable reference to settlement control in project.

Key words: quasi-equal interval QGM(1,1) model, average step length conversion, time segment forecasting, turfy soil, settlement prediction