<![CDATA[Indonesia one of the countries that prone to earthquakes which is tectonic and volcanic. Earthquake can damage a building, therefore the planning of earthquake resistant building is a requirement for an earthquake prone areas like in Indonesia and that planning must notice to the response of structure. One of the solution to predict the structural response and the damage level of multilevel steel portals is using the Artificial Neural Network (ANN). Thestructure model that author review is a multilevel steel portals which consisting of 3 models (5 levels, 10 levels and 15 levels). The structure is modeling with the finite element softwareand receives earthquake loads based on the time history of the Cape Mendocino earthquake that will be scaled on 10 capital of the provinces in Sumatera island. Artificial Neural Network(ANN) with backpropagation method is designed by using MATLAB program which is the input of this ANN is displacement, velocity and acceleration of three models that have beendesigned. The output that will be generated is damage level of the steel portal in the category of Safe, Immidiate Occupancy (IO), Life Safety (LS) or in conditions of Collapse Prevention(CP). Artificial neural network is trained by 8 earthquake data that owned by the capital of the provinces, and 2 remaining data from Padang City and Pekanbaru City that use forsimulation process. Based on the research, the model of artificial neural network has the potential accuracy to predict the damage level of structural between 90% -99%. From the result of simulation by ANN with 702 data from capital provinces of Padang and Pekanbaru City, 98,5% data could be predicted correctly by ANN. These results have shown that ANN is able to predict the damage level of multilevel steel portals in all of capital provinces in Sumatera Island.Keywords: multilevel steel portals, 3 structural model, structure response, time history, damage level, Artificial Neural Network.]]>
