摘要
通过RBF、BP神经网络及SVM算法3种预测方法,对爆破震动特征参量进行预测,并与传统萨道夫斯基公式进行对比分析研究。结果表明,3种方法预测精度均优于传统萨道夫斯基公式。当样本数有限时,BP、RBF神经网络在爆破振动峰值振动速度及主频率的预测中效果欠佳,SVM算法的预测精度优于RBF、BP神经网络,在实际工程应用中SVM算法对爆破振动特征参量的预测具有极强的适应性。
In this paper, three prediction methods of RBF and BP neural network and SVM algorithm are used to predict blasting vibration characteristic parameters, and comparative analysis and research are made with traditional Sadaovsk formula. The test results show that forecasting accuracy of these three methods is superior to traditional Sadaovsk formula. In the limited samples, the methods of BP and RBF neural network in predicting the Blasting vibration peak velocity and the main frequency are not good, and SVM algorithm is far superior to RBF and BP neural network. The SVM algorithm in the practical engineering application of prediction of blasting vibration characteristic parameters has a strong adaptability.
出处
《公路》
北大核心
2017年第4期12-17,共6页
Highway
基金
贵州省交通运输厅科技项目
项目编号2015122046
关键词
爆破震动特征参量
预测精度
SVM
BP神经网络
RBF神经网络
blasting vibration characteristic parameters
prediction accuracy
SVM
BP neural network
RBF neural network
