高地应力下大型地下洞室群开挖顺序与支护参数组合优化的智能方法

INTELLIGENT METHOD OF COMBINATORIAL OPTIMIZATION OF EXCAVATION SEQUENCE AND SUPPORT PARAMETERS FOR LARGE UNDERGROUND CAVERNS UNDER CONDITION OF HIGH GEOSTRESS

针对高地应力下围岩变形破坏的特殊性以及大型地下洞室群开挖支护优化计算量大的特点,在三维弹塑性数值计算的基础上,采用反映高地应力下脆性岩石变形破坏特点的新本构模型,提出基于弹性释放能、塑性区体积、洞室周边位移与支护费用的地下洞室群开挖顺序与支护参数组合方案的综合优化新指标,综合集成粒子群与支持向量机的智能技术,提出高地应力下地下洞室群开挖顺序与支护参数的智能优化新方法。该方法通过典型施工方案的数值计算构建学习样本,采用支持向量机方法对样本进行学习与预测,建立起施工方案与综合优化指标之间的非线性映射关系,在具有一定约束条件的全局空间下,通过粒子群优化算法搜索出开挖顺序与支护参数的全局最优组合方案。将该方法应用于高地应力区黄河拉西瓦水电站地下厂房洞室群的开挖顺序和支护参数优化分析,结果表明该方法的可行性。

Aiming at distinctness of deformation and failure of rockmass under high geostress and that optimization of excavation schemes and support schemes for large caverns is a complicated problem having a large search space and large scale of numerical calculation,a new intelligent optimization integrated method is proposed for optimization of excavation sequence and support parameters for large underground caverns under condition of high geostress.The method,which takes the integration optimization indexes including elastic release energy,plastic zone volume,displacement around caverns,support cost as fitness,integrates the 3D numerical method based on a new constitutive model which performs excellently under condition of high geostress,and the intelligent technique including particle swarm optimization（PSO） and support vector machine（SVM）.In detail,learning samples are established by numerical analysis for some typical construction schemes firstly.Then,using SVM trained by learning samples,the nonlinear mapped relationship of excavation sequence and support parameters with integrated optimization indexes are established.Finally,the globally optimum excavation sequence and support parameters are achieved by PSO search technique.Using the method mentioned above,the excavation sequence and support parameters of the large caverns of Laxiwa Hydropower Station located on the Yellow River in China are optimized.The result proves that the proposed method is feasible.