论文部分内容阅读
位移变形分析与预测可为基坑工程安全趋势评定提供量化依据。在论述回归分析、灰色模型与人工神经网络3种变形分析模型的基础上,以残差平方和最小为准则,提出了加权组合模型的建立思路。以南京滨江公寓基坑变形监测为例,分别将单一模型与组合模型应用于沉降监测数据处理过程中,采用最大绝对值、平均绝对值与标准差等残差指标对各方法建模精度及预测精度进行对比分析,通过3种模型特点与实用范围的总结,验证了加权组合模型的可行性与优越性。
Displacement analysis and prediction can provide quantitative basis for safety assessment of foundation pit engineering. Based on the three deformation analysis models of regression analysis, gray model and artificial neural network, the idea of weighted combination model is proposed based on the minimum sum of residuals. Taking the deformation monitoring of foundation pit of Nanjing Binjiang Apartment as an example, the single model and combined model are respectively applied to the data processing of settlement monitoring. The precision, prediction and accuracy of each method are evaluated by using residual indexes such as maximum absolute value, average absolute value and standard deviation. Accuracy analysis. Through the summary of three kinds of model characteristics and practical range, the feasibility and the superiority of the weighted combination model are verified.