基于BP神经网络的岩石损伤声发射事件源定位研究

Study on acoustic emission event source location of rock damage based on BP neural network

工程岩体一般包含大量的节理、裂隙等各类缺陷,利用AE技术可以对缺陷进行定位,但仍然有很大误差。为进一步减小AE定位误差,确定缺陷位置,做好安全预警工作,研究提高岩石声发射定位精度的优化方法。利用断铅实验模拟产生声发射信号,将声发射信号特征参数中的幅值、能量、振铃计数以及到达时差作为输入,断铅点真实坐标作为输出,构建BP神经网络模型,并通过实验确定最优隐含层数等参数,最后通过交叉学习训练交叉预测得出断铅点坐标,并与基于到达时差的传统定位法以及考虑材料各向异性条件下的波速优化算法进行对比。结果表明:BP神经网络算法与传统方法相比,板状岩石声发射定位的误差显著减小,定位精度提高;岩石试件定位误差的误差波动更为平稳,有效改善了声发射源位置对定位效果的影响。本方法为岩石缺陷的精确定位提供了1种较好的技术手段。

The engineering rock mass generally contains a large number of joints,cracks and other defects,the acoustic emission(AE)technology can be used to locate the defects,but there are still large errors.In order to further reduce the error of AE localization,determine the defect location,and do a good job of safety warning,an optimization method to improve the acoustic emission localization accuracy of rock was studied.The acoustic emission signals were simulated and generated by the lead breaking experiment,then the amplitude,energy,ringing count and time difference of arrival of the AE signal characteristic parameters were taken as the input,and the real coordinates of lead breaking point were taken as the output.So a BP neural network model was constructed,and the optimal number of hidden layers and other parameters were determined by experiment.Finally,the coordinates of lead breaking point were obtained by the cross learning training and cross prediction,and compared with the traditional localization method based on the time difference of arrival and the wave velocity optimization algorithm considering the material anisotropy.The results showed that compared with the traditional method,the BP neural network algorithm could significantly reduce the error of acoustic emission localization for plate-like rock and improve the localization accuracy.The positioning error of rock specimen fluctuates more smoothly,and the error fluctuation is more stable,which effectively improves the influence of AE source location on the positioning effect.This method provides a better technical means for the accurate localization of rock defects.