基于扩展置信规则库的基坑变形预测方法

Prediction of Wall Deflections in Deep Excavations based on Extended Belief Rule-Base

为了能够在初步设计阶段对基坑围护结构的最大水平位移进行合理预测,提出一种基于扩展置信规则库的基坑变形预测方法.方法将影响基坑围护结构变形的因素作为规则前件,将最大变形位移作为规则后件,以既有的基坑工程案例为基础构建扩展置信规则库;以待预测基坑的输入信息为规则激活条件,激活规则库中的相应规则;通过证据推理算法融合激活规则,得到基坑不同变形等级的组合信度分配值;将所有变形等级效用值与其对应的组合信度分配值加权求和,得到目标基坑的变形预测值.通过与多元回归分析法、神经网络法、支持向量机法的对比分析表明,介绍的方法不仅能够对影响基坑变形的定性因素进行处理,且不受数据信息不完整的限制,具有预测精度更高、预测机理明晰的特点.

In order to estimate the maximum horizontal displacement of foundation pit enclosure structure in preliminary design phase,a prediction method of wall deflections in deep excavations based on extended belief rule-base is proposed.The implementation of this method is divided into two steps:first,the factors that affect the deformation of foundation pit enclosure structure are taken as the front of the rule,and the maximum deformation value is taken as the back part of the rule,and the extended belief rule-base is constructed on the basis of 35 completed foundation pits;Secondly,Taking the input information of the foundation pit to be predicted as the condition of rule activation,the corresponding rules in the rule base are activated,and the activation rules are fused by the evidential reasoning algorithm,and the combined belief distribution values of different deformation levels of the foundation pit are obtained.The deformation prediction of the target foundation pit can be realized by weighted summation of the utility values of all deformation grades and their corresponding combined belief assignment values.The comparison with the three traditional prediction methods:multiple regression analysis,neutral network method,and SVM,shows that the method introduced in this paper can not only address the qualitative factor influencing the deformation of the pit but also have no limitation of incomplete data.Meanwhile,the proposed method has the characteristics of higher prediction accuracy and much clearer prediction mechanism.