软基沉降的BP神经网络和灰色系统联合预测

BP Artificial Neural Network and Grey System United Model for Settlement Prediction of Soft Soil Foundation

使用BP神经网络插值方法对灰色数据进行了预处理,进而建立了预测软基沉降量的BP神经网络和灰色系统联合模型。实例分析表明,该模型短期沉降预测结果的最大相对误差小于2%,最终沉降预测结果的相对偏差小于5%,且灰色预测时取后期沉降数据的计算结果准确度高于取前期沉降数据的计算结果准确度。

The establishment of the BP artificial neural network and grey system united model and grey data pretreatment depends on BP artificial neural network interpolate. The analysis of illustration shows that the maximum opposite error of predicting short period settlement is less than 2%, and that of ultimate settlement is less than 5%. As for choosing settlement data aspect, the result accuracy of choosing late stage data is higher than that of choosing prophase data.