Slope stability estimation with probabilistic neural networks
The purpose of this article is to demonstrate the application of probabilistic neural networks (PNNs) as a classification tool in the slope stability estimation. PNNs are applied to estimate slope stability according to the slope geometric shapes and soil mechanical parameters. Unlike other neural network training paradigms, PNNs are characterized by high training speed and their ability to produce confidence levels for their classification decision. The unit weight of the soil, cohesion of soil, internal fraction angle, slope angle, slope height, and pore pressure ration have been found to be the most significant factors affecting the slope stability. In order to improve the performance of PNN, the stochastic gradient approach is used to train the probabilistic neural networks. Validation is performed to show the efficiency of probabilistic neural networks for estimation of slope stability. The practical application results show that probabilistic neural network models generate higher predicting precision than the conventional linear regression, limit equilibrium method and maximum likelihood estimation.
Author's Name: Li, S., Liu, Y.
Volume: Volume 1
Issues: Issue 4
Keywords: Attack graph, Information fusion, Network security, Sensor, Vulnerability