【摘要】 以福建泉州南惠高速公路NH5标段路基爆破开挖工程为实例,运用人工神经网络原理,以孔径、孔深、孔距、排距、最大单孔药量、单段最大药量、总药量和爆源距离作为影响爆破振动的主要因素,建立BP小波神经网络模型.对质点的水平径向、水平切向、垂直方向等3个方向分别预测其爆破振动速度峰值及频率,并将预测结果与BP神经网络、支持向量机的预测结果进行对比.实验结果表明:BP小波神经网络的爆破振动速度峰值-频率模型预测收敛快、精度高,优于标准BP网络和支持向量机模型,其结果更加符合国家标准GB 6722-2003《爆破安全规程》的评价要求.更多还原

【Abstract】 Taking the Nanhui expressway NH5 section′s subgrade blasting excavation project in Quanzhou,Fujian as example,adopting artificial neural network theory,the BP wavelet neural network model was established,which considers various main factors,such as the charge hole diameter,distance,and depth,column distance between charge holes,line maximum charge of single hole,maximum charge weight per delay interval,total charge and explosive distance.By the BP wavelet neural network model,the blasting vibration peak value and main frequency were predicted in three directions separately,namely horizontal radial,horizontal tangential and vertical.The prediction results were compared with BP neural network and support vector machine model.The results show that: BP wavelet neural network model of blasting vibration peak value and main frequency owns fast convergence and high precision,so BP wavelet neural network mode is better than BP neural network model and support vector machine model,it meets well the requirements of "Demolition Safety Regulation"(GB 6722-2003).更多还原