PREDIKSI KERUSAKAN ABUTMEN JEMBATAN AEK MALAU DENGAN METODE ARTIFICIAL NEURAL NETWORK

Abstract

Indonesia is one of the most earthquake-prone regions, due to its position on a trajectory of earthquake hotspots dubbed the Pacific Ring of Fire. Along this path are rows of active volcanoes and tectonic plates that move and collide with each other. Indonesia has experienced many earthquakes, including in Aceh and Padang. Earthquakes are very threatening to the structures that stand on them. Structural collapse caused by earthquakes is generally very sudden and prone to evacuation procedures. Structural performance is needed to determine the condition of structural collapse during an earthquake. The calculation of structural performance uses various methods including static and dynamic methods. These methods require earthquake acceleration records and Response Spectrum Analysis.

This research was conducted on the Aek Malau Bridge Abutment located in Samosir Regency. In this study, the analysis was carried out non-linearly with the pushover analysis method. Bridge Abutments are given lateral loads in the form of earthquake loads and other loads at predetermined reference points. After entering the specified load, it can be known the level of structural performance that refers to the ATC-40 standard according to the capacity curve that occurs. After that, the level of structural performance is analyzed using Artificial Neural Network.

The results of this study were derived from some data with the Samosir Regency earthquake spectrum response. Abutments are declared in a safe state. The accuracy of of experiment  is 99.99996201% for Train data and 99.99997015% for test data.