Rock slope displacement prediction based on multi-source information fusion and SSA-DELM model

Abstract

In order to solve the inefficient use of multi-source heterogeneous data information cross fusion and the low accuracy of prediction of landslide displacement, the current research proposed a new prediction model combining variable selection, sparrow search algorithm, and deep extreme learning machine. A cement mine in Fengxiang, Shaanxi Province, was studied as a case. The study first identified the variables related to landslide displacement of rock slope, and removed redundant variables by using Pearson correlation and gray correlation analysis. To avoid the impacts of random input weights and random thresholds in the DELM model, the SSA algorithm is used to optimize the model’s parameters, which can generate the optimal parameter combinations. The results showed an enhanced generalization ability of the model by removal of redundant variables by Pearson correlation and gray correlation analysis, and higher accuracy in the prediction of landside displacement of rock slope by SSA-DELM compared to other traditional machine learning algorithms. The current study is significant in the literature on rock slope disaster analysis.