Early Warning System of Landslide Disaster using Generalized Neural Network Algorithm

Abstract

Landslides are frequently happened Indonesia, as many as 274 districts / cities are prone to landslides. There are many parameters that affect the landslide occurrence such as rainfall, land slope, soil moisture, and vibration. It is needed to provide a system that not only able to process data parameters to provide early warning of landslide disaster, but also increase the readiness of the population to minimize losses caused by this disaster. Generalized Regression Neural Network method is used to identify the effect of each parameter on the occurrence of landslide disaster. Tests conducted on field conditions and simulations on safe, alert, and danger condition to know the calculation result of artificial neural network. The simulation results are compared with the artificial neural network feed forward back propagation and manual calculations to demonstrate the effectiveness of the proposed method. The validation test on field condition using simulation shows average error of Generalized Regression method and Feed Forward Backpropagation method are 0.00115 and 0.08702, respectively. Furthermore, the Mean Square Error performance of the former method is better than that of the latter with values of 2.9157e-06 and 0.0112, severally.