基于支持向量回归积和改进粒子群算法的特定区间盾构机作业参数选取

为实现特定区间盾构机作业参数更准确的选取，提出了基于支持向量回归积(e-SVR)和改进惯性权重降低速度粒子群优化(IIWDSPSO)算法的盾构机作业参数选取模型。基于e-SVR构建掘进参数、地层参数、几何参数与地表沉降值之间的非线性关系模型，并基于实际盾构施工数据与人工神经网络模型、随机森林模型进行性能对比分析；通过10组仿真实验分析惯性权重降低速度对算法性能的影响，基于分析改进的粒子群优化算法优化特定地层参数和几何参数区间的掘进参数。结果表明，e-SVR模型对盾构施工数据样本具有更好的拟合和泛化能力，所提出的IIWDSPSO算法具有更好的准确性、稳定性和收敛概率。实际工程应用结果也验证了所提模型求解出的特定区间掘进参数能使地表沉降值相对更小，得到的掘进参数能够为实际工程提供更准确和可靠的参考。

Selection of Shield Construction Parameters in Specific Sections Based on e-SVR and Improved Particle Swarm Optimization Algorithm

To achieve more accurate selection of shield construction parameters in specific sections, a shield construction parameter selection model was proposed based on e-SVR and IIWDSPSO algorithm. Firstly, the nonlinear relationship model among tunneling parameters, formation parameters, geometric parameters and surface settlement was established based on e-SVR, and the performance was compared with ANN(artificial neural network) and RF(random forests) models based on the actual shield construction data. Secondly, the influences of inertia weight decreasing speed on algorithm performance were studied by 10 groups of simulation experiments. The tunneling parameters in specific formation parameters and geometric parameter intervals were optimized based on the improved PSO algorithm. The results show that e-SVR model has better fitting and generalization ability for shield construction data samples, and the proposed IIWDSPSO algorithm has better accuracy, stability and convergence probability. The practical engineering application results also verify that the proposed model may obtain the tunneling parameters in a specific section, which may make the ground settlement value relatively smaller. The obtained tunneling parameters may provide a more accurate and reliable reference for practical engineering.