基于Fisher方法的滑坡稳定性预测

Prediction study of landslide stability based on Fisher method

采用Fisher判别分析法,将影响滑坡失稳的外形特征、滑面特征等4项内在因素和河流作用、暴雨强度等5项外在因素作为判别因子,建立基于Fisher理论的滑坡稳定性预测的判别分析模型。以国内典型滑坡工程发生失稳的实测资料作为学习样本进行训练,建立相应线性判别函数对待判样本进行预测。研究结果表明,Fisher判别分析模型预测性能良好,预测精度高,回判估计的误判率为零,是滑坡稳定性预测的一种有效方法,可以在实际工程中应用。

A discriminant analysis model of landslide stability based on Fisher method is established,using 4 internal discriminant factors,such as appearance characters,slip surface characters,etc,and 5 external discriminant factors,such as fluviation,rainstorm intensity,etc.Discriminant functions are obtained by training measured data from typical landslide projects,and then the functions can be used to predict the sub-sample data.The results show that the Fisher prediction model is good in performance and precision,and its false positive rate is zero.This is a new method to predict landslide stability and can be used in practical engineering.