基于BP神经网络的微差爆破震动预测研究  

Millisecond Blasting Vibration Prediction based on BP Neural Network

作　　者：王涛[1] 张建华[2] 

机构地区：[1]广东宏大爆破股份有限公司,广州510623 [2]武汉理工大学,武汉430070

出　　处：《爆破》2015年第2期140-143,共4页Blasting

摘　　要：爆破震动危害是矿山安全评估的一个重要指标,爆前对其震动预测,做好安全防护,是保证矿山安全生产的重要措施。爆破对象及其地质情况的不确定性、爆区周边环境的复杂性,使得震动监测难以用一套统一的公式进行准确预测。人工神经网络可实现复杂环境、多因素影响下的仿真模拟,利用BP神经网络建立爆破震动预测模型,将爆破的原始数据及监测数据输入模型,利用Matlab软件自带的神经网络软件包自编程序对其进行训练,使得模型的传递函数达到最优。实践证明,将模型用于爆前震动预测,能够有效的预测震动,指导施工。The blasting vibration hazard was an important indicator in mine safety assessment. Blasting vibration forecast was important to ensure the safe production in mines. The uncertainty of the blasting object,including the ge- ological condition and complexity surrounding environment of the blasting zone, made it difficult to accurately predict vibration by a unified formulas. The artificial neural network can obtain simulation of muhiple factors under the com- plex environment, and establish the blasting vibration prediction model. By training the original blasting and monito- ring data by Matlab software with neural network packagel, the best transfer function model was achieved. The model was effectively used to predict blasting vibration.

关 键 词：爆破震动 BP神经网络 网络模型 震动预测 

分 类 号：TD235.46[矿业工程—矿井建设]