针对岩石力学实验声发射三维定位常规算法精度较低的问题,分析各种改进的定位算法。通过使用波速均匀的样品进行人工触发声发射实验发现,随机误差是岩石实验中声发射定位不准的最主要因素。为消除随机误差,采用最小绝对偏差法进行定位。绝对偏差最小法是一种适合于存在离群点时的稳健估计算法,可以克服最小二乘法仅在误差为正态分布时才有效的缺点。实验结果表明,与最小二乘法相比,当数据中存在个别误差较大的离群点时,采用最小绝对偏差法可以使得定位精度得到明显改善。
The acoustic emission(AE) 3D location is widely used in rock mechanics experiments, and one of its main disadvantages of the current technique is the low accuracy of general algorithms. It is realized that the random error is the dominating factor which influences the accuracy of AE location the most based on an analysis of many algorithms and a check with an artificial AE shock excitation. Therefore, least absolute deviation, which is more robust than least squares especially for Gaussian noise, is selected to reduce the random error. Experimental results show that compared with least squares, least absolute deviation can improve location accuracy effectively considering the existence of several outliers.