摘要
根据观测的爆破振动响应数据和有限元正演振动分析数据,建立了基于人工神经网络的爆炸冲击荷载参数识别方法。采用Levenberg-Marquardt优化方法修正网络的权值和阈值,大大提高了神经网络的收敛速度。研究了观测噪音对爆炸荷载参数识别结果的影响。数值计算结果表明,所建立的基于人工神经网络的爆炸冲击荷载参数识别方法具有良好的鲁棒性和抗观测噪声能力。
The artificial neural network model is used to identify the loads parameters from underground explosion according to the field measurement data of ground vibration and computed data with finite element method. The convergence rate of the artificial neural network is improved by using Levenberg-Marquardt optimization method. The influence of measurement noise on identified results is studied. The results of numerical computation show that the proposed identification method is of robustness and ability against measurement noise.
出处
《岩石力学与工程学报》
EI
CAS
CSCD
北大核心
2003年第11期1870-1873,共4页
Chinese Journal of Rock Mechanics and Engineering
基金
国家自然科学基金(10072014)
高等学校博士点基金(2000014107)资助项目。
关键词
人工神经网络
爆炸
冲击荷载
反问题
振动响应
观测噪音
Acoustic noise
Finite element method
Identification (control systems)
Inverse problems
Neural networks
Optimization
Vibrations (mechanical)
