基于IAGA+M-SVR的岩土参数反分析方法及其工程应用

由于岩土工程的复杂性和不确定性致使岩土力学参数很难准确获得，提出了一种结合改进自适应遗传算法（IAGA）与多输出支持向量机（M-SVR）各自优点的基坑岩土参数位移反分析的方法。该方法充分利用遗传算法全局寻优以及支持向量机的小样本数据建模方面的优势，并且克服了传统遗传算法在复杂函数优化的寻优搜索中容易陷入局部极值、搜索效率低、不稳定以及利用单输出支持向量机建立多测点模型时计算量大、精度不高等缺点。通过将该方法应用于营口兴隆大厦桩撑支护体系基坑的岩土参数的反分析中，其反演的等效岩土参数值所计算的位移值与实际位移的平均相对误差仅为 2.83%，验证了该方法的适用性和结果的准确性。最后将反演所得抗剪强度参数与室内试验结果，弹性模量与经验值进行了对比分析，给出了该类型基坑三维数值模拟计算的初始参数建议值，采用该建议参数计算所得位移值与实际观测值平均相对误差为 4.07%，满足实际工程的应用。

Back Analysis Method of Geotechnical Parameters based on IAGA+M-SVR and Its Application

Due to the complexity and uncertainty of geotechnical engineering，it is difficult to obtain the geotechnical parameters accurately. This paper presents a method of back analysis of geotechnical parameters displacement of foundation pit，which combines the advantages of improved adaptive genetic algorithm（IAGA）and multi-output support vector ma chine（M-SVR）. This method makes full use of the advantages of genetic algorithm in global optimization and small sample data modeling of support vector machine，and overcomes the disadvantages of traditional genetic algorithm，such as easily falling into local extremum，low efficiency and instability in the search of complex function optimization，and large calcula? tion amount and low precision in building multi-point model by using single output support vector machine. By applying this method to the back analysis of the geotechnical parameters of the foundation pit in the pile supporting system of Yingkou Xin glong Building，the average relative error between the calculated displacement value with the equivalent geotechnical param eter value and the actual displacement is only 2.83%，which verifies its applicability and accuracy. Finally，through the com? parison of the shear strength parameters obtained from the back analysis with the results of laboratory test，andof the elastic modulus withthe empirical value，the recommended initial parameter value of the three-dimensional numerical simulation for this type of foundation pit is given. The average relative error between the displacement value calculated by the recommended parameter and the actual observation value is 4.07%，which satisfies the application of practical engineering.