围岩流变模型参数识别的启发式-支持向量机方法

Heuristic Calculation-Support Vector Machine Method for Parameter Identification in a Rock Rheological Model

针对传统反分析算法在精度与效率方面的缺陷,文章引入支持向量机、启发式智能优化算法及三维数值模型对围岩流变参数进行反演。其具体过程为:采用均匀设计方法,利用三维数值模型构造训练样本集,通过启发式智能算法搜索最佳的支持向量机模型参数,建立岩体力学参数与岩体位移之间的非线性支持向量模型;对任意一组给定的岩石力学参数,利用支持向量机的外推能力代替数值计算软件的正向计算,通过启发式算法对力学反演参数进行调整,即可迭代求出岩体力学参数。将该方法运用于广甘高速公路杜家山隧道工程流变参数的反演中,结果表明,该反演方法合理可靠,可有效地指导隧道工程参数设计和施工稳定控制。

In light of the precision and efficiency problems related to the traditional inversion method, an inversion fl）r the rock rheological parameters was carried out by introducing the support vector machine, heuristic optimization algorithm and 3D numerical model. Specifically, a non-linear support vector model was established for the relationship between the mechanical parameters and the displacement of the rock mass based on training sample sets provided by a uniform design and a 3D numerical model and the optimal parameters of the support vector model obtained by the heuristic algorithm; for any given set of rock parameters, the extrapolating ability of SVM was used in stead of the numerical computation, and an adjustment of the inversion parameters related to the rock＇s mechanical behaviors was conducted by an heuristic algorithm and an iteration calculation. Based on the case of the Dujiashan tunnel on the Guangyuan-Gansu expressway, the above inversion method is proven to be reasonable and reliable for guiding parameter design and construction control of a tunnel.