摘要
特大断面地下洞库爆破开挖工程中涉及到众多的影响因素,为了较准确地预测出爆破振动速度,引入支持向量机理论,建立最小二成支持向量机爆破振动速度预测模型(LS-SVM模型),该模型利用结构风险最小化来提高求解问题的速度和精度。采用该模型对某地下水封LPG洞库工程进行爆破振动速度预测,并与传统的萨道夫斯基回归公式模型(萨氏模型)和模糊神经网络模型(FNN模型)进行对比分析。分析结果表明:LS-SVM模型、FNN模型与萨氏模型的全局均方根相对误差RMSRE分别为4.68%、14.42%与19.33%;LS-SVM模型有14组数据满足预测模型泛化能力误差阀值(6%)的要求,而FNN模型与萨氏模型均不满足要求。因此LS-SVM模型在爆破振动速度预测中的预测性能和泛化能力均优于FNN模型及萨氏模型,可为多因素影响下类似工程爆破振动速度预测提供借鉴经验。
There are many influence factors in blasting excavation engineering of super-large section underground caverns. In order to accurately predict the blasting vibration velocity, the LS-SVM model was established based on support vector machine ,which improved the speed and accuracy of the solving problem with structural risk minimiza- tion. The LS-VSM model was adopted to predict blasting vibration velocity induced by the underground water-sealed LPG caverns in China, and compared with the traditional prediction model as Sadov~ formula model（ SA model）and fuzzy neural network model（ FNN model）. The analysis results indicated that global root mean square relative error （RMSRE）of LS-SVM model was 4.68% compared with 14.42% by FNN model and 19.33% by SA model. Mean- while, there were 14 groups meeting the error threshold value （6%）about the generalization performance of prediction model ,while FNN model and SA model didn＇t meet the requirement. Thus ,in prediction of blasting vibration velocity, regardless of prediction effect or generalization performance, the LS-SVM model was superior to FNN model and SA model.
出处
《爆破》
CSCD
北大核心
2017年第4期12-16,共5页
Blasting
基金
国家自然科学基金资助项目(41672260)
中国地质大学(武汉)教学实验室开放基金资助项目(SKJ2014061
SKJ2016091)
关键词
地下洞室群
爆破振动速度预测
最小二乘支持向量机
模糊神经网络
underground caverns
blasting vibration velocity prediction
least squares support vector machine
fuzzy neural network
