【摘要】 为了解决工程实践中复杂岩体的岩体力学参数的取值问题,利用偏最小二乘回归方法,在自变量间存在强相关性的情况下建立计算模型,有效解决了变量间的多重共线性问题,尤其当样本点少、变量个数多时,该方法更有效.以岩体变形模量的取值为例进行了分析研究,结果表明,PLS拟合值与实测值相差很小,拟合值与实测值的趋势基本一致,说明该方法实用有效.计算量小,预测精度高.更多还原

【Abstract】 In engineering practice, how to value the rock mechanical parameters has been one of the hard problems that the engineering field has been paid attention to because of the complexity of the rock mass. Therefore, a new and effectively method,partial least-squares regression is used. Its advantage is able to build up the computing model between the values with strong relativity, and resolve the multi-collinearity among variables, especially when there are few sample points and many variables. The case of the rock deformation module shows there is little difference between the fitting values and the measured values, and the basically consistent changed trend. The errors reach ranging into 15 percent, with the most error 15.5 percent and the least error 0.8 percent. So the method has little calculation_work and prediction is of high precision by analyzing. The result from above method is valid for the value of the rock deformation module, and provides a new method for selecting the rock mechanical parameters.更多还原