Integration of grey analysis with artificial neural network for classification of slope failure

With the advent of technology and introduction computational intelligent methods, prediction slope failure using machine learning (ML) approach is rapidly growing for past few decades. This study employs an “artificial neural network” (ANN) to predict failures based on historical circular cases. Using feed-forward back-propagation algorithm with a multilayer perceptron network, ANN powerful ML method capable predicting complex model However, result can be improved by integrating statistical analysis method, namely grey relational (GRA), model. GRA identifying influencing factors input data correlation level reference sequence comparability dataset. analyze eliminate unnecessary samples improve performance. Grey analysis-artificial network (GRANN) was developed six factors: unit weight, friction angle, cohesion, pore pressure ratio, height, factor safety (FOS) as output factor. The results were analyzed accuracy percentage receiver operating characteristic (ROC) values. It shows that GRANN has outperformed giving 99% 0.999 ROC value, compared 91% 0.929.