圆弧滑动边坡稳定性分析的神经网络方法

Neural Network Approach for the Stability Analysis of Circular Failure Slope

搜集了大量圆弧滑动边坡工程实例,并根据实例边坡的实际稳定状况及其几何、物理力学参数构建了训练数据集和测试数据集。应用所构建的训练数据集,对所建立的用于圆弧滑动边坡稳定性分析的神经网络进行训练,并采用训练数据集和测试数据集对其拟合能力和推广预测能力进行测试,最终获得满意结果。应用所建立的神经网络对某高速公路边坡的最危险圆弧滑动面的安全系数、稳定状态和圆心位置作了预测,结果与实际情况吻合。

The numerous cases of circular failure slope were collected, based on the practical stability status and geometry, physics, mechanics parameters of the slope cases the data set for training and the data set for testing were formulated. The artificial neural network (ANN) for the stability analysis of circular failure slope was trained with the data set for training. The data set for training and the data set for testing were used to test the fitting and general-prediction capabilities of the trained ANN. The trained ANN was used to predict the safety factor, the circle center of the critical circular face and the stability status of the excavated slope on a certain express highway, and the predicted results are very close to the actual situations.