路堑开挖爆破中邻近民房风险评价的网格支持向量机模型

应用支持向量机理论并结合路堑开挖爆破特点,提出路堑开挖爆破中临近民房安全性评价的支持向量机回归模型。考虑爆破参数、地质条件和民房结构状况因素,选取最小抵抗线、孔距、排距、炸药单耗和民房的自振周期等16个影响较大的因素作为该模型的输入参数,房屋安全等级系数作为模型输出,利用网格搜索寻优方法对支持向量机模型的参数进行了优化。以19组路堑开挖爆破实测数据作为学习样本进行训练,对另外3组待判样本进行判别,并与多元回归、BP神经网络回归和实测结果进行对比。研究结果表明:建立的支持向量机回归模型对路堑开挖爆破中临近民房安全性评价效果良好,具有较高的预测精度。

THE MODEL OF SUPPORT VECTOR MACHINE WITH GRID SEARCH METHOD TO EVALUATE THE RISKS OF RESIDENTIAL HOUSES BY BLASTING FOR CUTTING EXCAVATION

On the basis of support vector machine(SVM) theory and the characteristics of cutting excavation blasting,a regression model to evaluate the safety for residential houses by blasting for cutting excavation was proposed.Considering the factors of blasting parameters,geological condition and structure of houses,sixteen influence factors such as minimum resistant line,hole spacing,row spacing,unit explosive consumption,natural vibration period of houses were used as the input parameters for the SVM regression model,and the safety grade coefficient of house was forecasted by the proposed model which was optimized by grid search method(GSM).The nineteen sets of measured data were trained as learning samples,and other three groups were used as testing samples to distinguish.The prediction results of the SVM regression model were compared with those of BP neural network,multivariate linear regression(MLR) model and the measured values.The results showed that the SVM regression model worked well in predicting the safety for residential houses by blasting for cutting excavation,and the prediction accuracy was acceptable.