神经网络在研究岩石断裂韧度尺寸效应的应用

Application of Neural Network to the Study of Size Efect in Rock Fracture Toughness

介绍了人工神经网络在预测人字形切槽巴西圆盘试样(CCNBD)断裂韧度时的应用,通过分析表明,神经网络得到的断裂韧度预测值比CCNBD尺度律公式得到的值更加接近实验值,神经网络直接基于实验数据进行训练,不需要作任何假设,对于构形比较复杂,试样制备时难以满足完全几何相似的CCNBD试样,研究其尺寸效应是比较有利的。

This paper introduces the method to predict the rock fracture toughness of the cracked chevron notched Brazilian disc(CCNBD) by using neural network.As a result,the fracture toughness from ANN forecast is much more close to the experimental value than that from scaling law calculated.Moreover,since Neural networks directly use experimental results in training,there is no need for any assumption.This is of advantage to the complex samples of CCNBD considering the size effect of incomplete geometrically similar samples.