Paper Information

Journal:   JOURNAL OF SCIENCE (UNIVERSITY OF TEHRAN) (JSUT)   SPRING 2007 , Volume 33 , Number 1 (SECTION: GEOLOGY); Page(s) 57 To 64.
 
Paper: 

APPLICATION OF ARTIFICIAL NEURAL NETWORK FOR LANDSLIDE HAZARD ZONATION

 
 
Author(s):  RAKEEI BABAK, KHAMEHCHIAN MASHA ELAH*, ABD ALMALEKI P., GIAH CHI P.
 
* DEPARTMENT OF ENGINEERING GEOLOGY, TARBIAT MODARRES UNIVERSITY, TEHRAN, IRAN
 
Abstract: 

There are several methods for zonation of landslide hazard. They can be generally divided into two groups; direct and indirect methods. In this study, we used Artificial Neural Network (ANN) as an indirect method for landslide hazard zonation in Semnan province. A total of 49 landslides or slide zones, that overlaid on the topographical map with the scale of 1:50000, were studied. Maps of factors such as lithology, slope, aspect, land use, buffer of faults, DEM, precipitation were then prepared. The sedata were then normalized, according to the maximum value of each factor. The normalized data was then fed into a multi-layer preceptor with back error propagation algorithm. The network had 3 layers, first layer as input layer, had 7 input elements each of which related to one factor. The second layer as hidden layer had 20 process element. The last layer as output layer consisted of one process elements which have been trained to offer5 level of hazard risk. This structure was found as the best optimized structure through extensive simulations. Our data base consisted of 2016 records. This database was randomly divided into the two separate groups. One as train database consisted of 1626 records which used to train the established ANN. The second group named the testing database consisted of 400 records, which used to test the performance of the ANN. The accuracy of network for predicting landslide hazard was measured about 91.25%. Results reveal that artificial neural network model the landslide hazard zonation better that other approaches.

 
Keyword(s): ARTIFICIAL NEURAL NETWORK, LANDSLIDE, ZONATION, PERCEPTRON
 
References: 
  • ندارد
 
  Persian Abstract Yearly Visit 77
 
Latest on Blog
Enter SID Blog