Study on prediction of foundation settlement of Qingshan Power Plant in Wuhan

摘要

The equal time-interval original data is the precondition of the grey model establishment. As the acquired data in the practical using is usually unequal time-interval, there are some errors in the predicting results. On the basis of the grey model GM (1, 1), a new grey prediction model with time-dependent parameters is established. The parameters of model, which can be expressed in polynomial expression, are assumed to be continuous functions of time. The undetermined coefficients of polynomial expression can be solved by the principle of least square method based on the previous information. In the model, both time-varying property and the correcting effect of the forgetting factor are considered, so the prediction error is smaller. The improved model is applied to predict the foundation settlement of Qingshan power plant chimney in Wuhan. The predicted value by the model is compared with those by measuring and three-point method, the result shows that the value derived from three-point method is slight more than the actual settlement, and the value derived from improved GM(1,1) model is close to that from measuring, the predicted error is less than 2%. The example showed that the short-term predicted value from the model is quite exact and the predicted ultimate settlement value can be used as reference for the engineer. Based on the improved GM (1, 1) model, the foundation settlement is predicted from the observed data of the previous 468 days. The result shows that the predicted value from the model is very close to that from the three years observation.