Rock fall risk zoning on Khalkhal to Shahroud road using Multilayer Perceptron algorithm

Document Type : Research Article


1 Professor of Geomorphology, Faculty of Social Sciences, University of Mohaghegh Ardabili, Ardabil, Iran

2 Assistant professor of Geomorphology, Faculty of Literature and Human sciences, Ferdowsi University of Mashhad, Mashhad, Iran

3 Masters graduate, Geomorphology and Environmental Management, Faculty of Social Sciences University of Mohaghegh Ardabili. Ardabil, Iran



Rockfall is one of the phenomena that causes a lot of damage, especially in mountainous areas. Therefore, it is necessary to evaluate the areas prone to rock fall in mountainous areas. The purpose of this research is to zonate the risk of rock falls on Khalkhal-Shahroud road using multi-layer perceptron algorithm. The multi-layer perceptron algorithm is one of the new machine learning models that has the ability to solve complex problems. In order to identify the important factors in the occurrence of rock falls, according to field studies, 8 factors have been identified, which include: height, vegetation, slope direction, distance from the fault, distance from the road, geology, land use, slope. After pre-processing, all layers are entered into SPSS MODELER software and the modeling is designed with 8 input neurons, 8 intermediate neurons and 1 output. The results of this research showed that in the multi-layer perceptron algorithm, the highest weight value was assigned to the geological layer with a value of 0.20 and for the land use layer and distance from the road with values of 0.14 and 0.12, respectively. Also, in the validation section of the model, the results showed that the AUC value shows 0.9810 in the training section and 0.9876 in the testing section of the network, which indicates that the model has high validity in both the training section and the testing section. And it is highly ranked. Finally, it is suggested to use other machine learning models such as support vector machine in future studies to study and evaluate rock fall and range movements.


Main Subjects


Articles in Press, Accepted Manuscript
Available Online from 07 June 2024
  • Receive Date: 30 April 2024
  • Revise Date: 29 May 2024
  • Accept Date: 05 June 2024