岩体变形模量确定方法有室内外试验法、数值分析法、反分析法、岩体分类法等。上述方法均存在很大缺陷，而神经网络法的日益完善使通过建模预测岩体参数成为可能。以溪洛渡水电站的88组数据为基础，考虑岩石质量指标RQD、RMD、Vp等因素，建立了基于模式搜索法的改进RBF神经网络模型，并用该模型预测岩体变形模量。为了验证模型的准确性，将西藏如美水电站岩体的17组数据代入，将其预测结果与BP神经网络模型结果及原位数据作对比。结果表明，改进RBF模型更适于硬岩岩体变形模量的预测。%The deformation modulus of rock mass can be determined by in-situ and indoor test, numerical analysis, back a-nalysis and rock mass classification methods etc. These methods have limitations, but the improvement of the artificial neural net-work ( ANN) method makes it possible to predict the rock mass deformation modulus accurately. On the basis of 88 sets of test data of Xiluoda Hydropower Station and by consideration of several rock quality index such as RQD, RMD and Vp, an improved RBF ( Radial Basis Function) neural network model is established to predict the deformation modulus of rock mass, in which, the pattern search method is used to solve the optimal SPREAD value. For verifying the correctness of the model, 17 sets of data of Rumei Hydropower Station, Tibet, were put into the model and the results were compared with the measured data and the results obtained by BP neural network respectively. It is concluded that the improved RBF neural network model is more suitable for the prediction of deformation modulus of hard rock mass.
展开▼
