An improved DWKNN-Based rock fall distance prediction model

Abstract:

Predicting the rock fall distance of a slope are of primary concern in identifying potential dangerous area and the prediction method plays crucial roles in accurate prediction of rock fall distance. Recently, machine learning has been widely used for prediction of rock fall distance, and the Dual Weighted K-nearest Neighbor (DWKNN) algorithm, one of machine learning techniques, showed good performance in pattern recognition. In this study, the DWKNN algorithm is first attempted to apply to the prediction of rock fall distance, and the prediction model of the rock fall distance based on the DWKNN algorithm is proposed. Then, based on the typical rock fall cases, rock fall distance prediction of the slope is carried out using the prediction model, which is the necessary step before the major projects construction near the slope, and our proposed prediction model achieves a higher accuracy than the other three popular methods including K-nearest neighbor (KNN) algorithm, distance-weighted k-nearest neighbor (WKNN) algorithm and support vectors machine (SVM) algorithm. At last, we applied our proposed prediction model to a series of laboratory tests. Extensive experimental results for rock fall distance prediction demonstrate that the effectiveness of our proposed prediction model of the rock fall distance based on the DWKNN algorithm.