基于多种群遗传算法考虑浅层基岩及粘弹性的路基模量反演方法

提出一种考虑浅层基岩和粘弹性的路基模量反算新思路。对于路基模型，分别采用位移边界条件和开尔文模型来描述浅层基岩深度和粘弹性，并利用有限元软件对便携式落锤弯沉仪现场试验进行模拟，采用多种群遗传算法对路基模量最优值进行迭代。有限元模拟计算结果表明，新方法考虑浅层基岩的正算模型的反演模量平均误差为7.0%，而考虑半空间的正算模型却高达16.2%，说明忽略浅层基岩会使反演模量产生较大误差；随着浅层基岩深度的增加，误差减小，当深度为3 m时，影响几乎不存在。对于考虑粘弹性的有限元模型，在正算模型中忽略该特性的最大误差可达27.9%，而考虑粘弹性的最大误差仅为7.4%。由于浅层基岩深度勘探难度大，仅从理论方面进行研究。

Back-calculation of subgrade modulus considering shallow bedrock and viscoelasticity based on multi population genetic algorithm

This article proposes a new idea for back-calculation of the subgrade modulus considering shallow bedrock and the viscoelastic characteristics. For the subgrade model, displacement boundary conditions and the Kelvin model are adopted to describe the depth of the shallow bedrock and the viscoelasticity, respectively. The portable falling weight deflectometer (PFWD) field test is simulated by ABAQUS general finite element (FE) software, and the optimal value of the modulus is iterated by a multi population genetic algorithm (MPGA). Based on the new method, back-calculation results from FE simulation tests show that the modulus average error of the forward model considering shallow bedrock is 7.0%, while that of the forward model considering half space is as high as 16.2%, indicating that negletct of the shallow bedrock in the forward model of the back-calculation program may cause a significant error in the inversion modulus, but its influence decreases with the increase of the depth of the shallow bedrock, and the depth limit is 3 m. Similarly, for the FE model considering viscoelasticity, the maximum error for neglect of this attribute in the forward model reaches 27.9%, compared with the error of only 7.4% when considering viscoelasticity in the forward model. Due to the difficulty exploring the depth of shallow bedrock, examinations are only conducted from the theoretical aspect.