基于人工神经网络的水下爆破振动预测技术  

Prediction Technology of Underwater Blasting Vibration Based on Artificial Neural Network

作　　者：石晨晨 陈宏涛 杨波 周泰安 赵军 SHI Chen-chen;CHEN Hong-tao;YANG Bo;ZHOU Tai-an;ZHAO Jun(Chengdu New Technology Blasting Engineering Co.,Ltd.of China Railway No.2 Group,Chengdu 610000,China;Guangxi University,Nanning 530004,China;Blasting Safety Technology R&D Center of China Railway Group Limited,Chengdu 610000,China)

机构地区：[1]中铁二局集团成都新技术爆破工程有限公司,成都610000 [2]广西大学,南宁530004 [3]中国中铁爆破安全技术研发中心,成都610000

出　　处：《价值工程》2022年第34期133-135,共3页Value Engineering

基　　金：中国中铁爆破安全技术研发中心资助项目(20210810-02);广西大学工程爆破研究所开发基金资助项目(20210601-12)。

摘　　要：为了研究梧州市某水道航段疏浚、炸礁爆破振动对临近建构物产生影响的规律。根据现场监测的振动数据,利用MATLAB软件进行RBF神经网络建模与分析,选取比例药量、场地系数、孔网参数、孔数作为输入层元素,以实测爆破合振速峰值为输出层进行网络训练,进而进行爆破振动预测,并与工程爆破振动预测中使用最为普遍的BP神经网络进行对比。得出人工神经网络中RBF模型的RMSE与R2都要比BP模型预测精度高,RBF模型更适用于水下爆破振动预测。In order to study the influence of underwater blasting vibration on adjacent structures in a channel in Wuzhou City,according to the vibration data monitored on site,the RBF neural network modeling and analysis are carried out by using MATLAB software.The proportional charge,site coefficient,hole network parameters and hole number are selected as the input layer elements,and the measured blasting peak speed is used as the output layer for network training,so as to predict the blasting vibration,which is compared with the BP neural network method which is the most commonly used in the blasting vibration prediction.It is concluded that the prediction accuracy of RMSE and R2 of RBF model in artificial neural network is higher than that of BP model,and RBF model is more suitable for underwater blasting vibration prediction.

关 键 词：水下爆破 爆破振动 振动预测 MATLAB 人工神经网络 

分 类 号：TV542[水利工程—水利水电工程]