非等间距灰色优化模型及其在基坑沉降预测中的应用

传统非等间距灰色模型通常采用非等间隔数据进行分段线性插值,从而求得等间隔序列;但是实际基坑沉降并不是线性变化的,利用此法生成的等间隔序列较实际数据存在较大误差.针对传统非等间距灰色模型缺陷,分别利用RBF神经网络插值与三次样条插值生成等间距序列进而求得模型参数,并运用优化的非等间距灰色模型对某基坑的沉降量进行分析预测.计算结果表明：优化后的模型具有更高的精度,故本模型可作为基坑沉降预测的一种新方法.

Non-equidistant gray optimization model and its application to the foundation settlement prediction

Traditional non-equidistant gray model usually uses non-equal interval data fo piecewise linear interpolation in order to obtain equally-spaced sequence. Because the actual foundation settlement is not linear variation, the equally-spaced squence exists with greater errors than the actual data. In view of the traditional non-equidistant gray model defects, the RBF neural network interpolation and cubic spline interpolation are used to enerate equidistant sequence to obtain the model parameters. An optimized non- equidistant gray model is established for analysis prediction of a foundation settlement. The calculation results show that the optimized model has higher accuracy which can be used as a new method for the foundation settlement prediction.