考虑多因素交互作用的岩石强度预测

影响岩石强度的因素相互作用与相互耦合，常规方法难以快速准确描述影响因素间相互耦合作用对岩石强度的影响。结合前人研究成果与实践经验，引入岩石工程系统（RES）理论的交互作用矩阵，采用增大系数的双曲正切函数为激活函数的BP网络编码方式，提出了基于RES交互作用矩阵的BP网络岩石强度预测模型，并进行网络训练。结果表明：运用RES理论系统分析思想，在构造相互作用矩阵基础上，采用改进的BP网络，不仅能够快速进行岩石强度预测，而且也提高了预测精度。该方法不仅为岩石强度预测提供一种新思路，同时也为工程岩体评价提供了一种新途径。

PREDICTION OF ROCK STRENGTH CONSIDERING MULTI-FACTORS INTERACTION

The rock strength involves the interaction and inter-coupling of many influencing factors. The conventional methodis unable to evaluate the interaction relationship among the influencing factorsaccurately. Therefore according to previous scholars' achievements and practical experience, the essay introducesthe interactive matrix of Rock Engineering System(RES) theory, and applies the increscent coefficient ofhyperbolic tangent function asthe BP network coding way of activation function. The thesis also presents the predicted model ofinteractive matrix of BP network of rock strength based on RES and involves network training. The final result shows that applying theanalytical thought of REstheoretical system on the basis of the construction of interactive matrix and with the improved BP networknot only predict the rock strengthquickly, but also improve the prediction accuracy. This research not only provides a new idea for rock strength prediction, but also provides a new way forrock mass engineering evaluation.