一种盾构施工引起的地面沉降预测方法

针对盾构施工引起的地面沉降的影响因素众多、关系错综复杂,常规数学模型难以准确预测沉降量的问题,该文提出了采用主成分分析法和粒子群优化的支持向量机方法来建立预测模型。并结合工程实例将预测结果与常用的多元线性回归模型和基于Levenberg-Marquardt算法改进的BP神经网络模型(LM-BP)的预测结果做了比较。结果表明:PCA-PSO-SVM的预测结果精度较多元线性回归模型和LM-BP神经网络有很大的提高,证明了该研究方法具有一定的理论意义和参考价值。

A prediction method of the land subsidence caused by shield construction of PCA-PSO-SVM

In view of the fact that the influence factors of the land subsidence caused by shield construction are numerous and complicated,and it is difficult to predict the subsidence accurately by the conventional mathematical method,the principal component analysis method and support vector machine model optimized by particle swarm optimization were used to establish the prediction model.Combined with the engineering example,the predicted subsidence values were compared with those of linear regression model and BP neural network model updated by Levenberg-Marquardt algorithm.The results show that the prediction accuracy of PCA-PSO-SVM model is improved more greatly than that of LMR model and LM-BP neural network model,proving that the proposed method has certain theoretical significance and reference value.