| dc.description.abstract | Demand capacity ratio (DCR) is the ratio between the number of demand
to the capacity available in a system in a given period of time. If the DCR
exceeds 1, it means that the bridge has exceeded its capacity and has the
potential to collapse. In order to avoid the risk of accidents on bridges, it
is very important to know the DCR value of a bridge. However, the DCR
calculation process is too detailed because it includes many variables
such as vehicle weight, speed, wind, and other environmental factors.
This is quite time-consuming because it requires a process of trial and
error for each step. This study aims to approach using artificial neural
networks (ANN) to predict DCR so that the process is shorter. ANN is
used to predict the DCR value on the upper structure of the class A steel
truss bridge of Highways with a span of 45 meters. As input data, the
ultimate stress, span length, and steel profile area are used, while the
output is the DCR value. Input and output data were obtained from SAP
2000 modelling results by varying bridge dimensions, material
properties, and profile types after a combination of loading on the bridge.
The result obtained is that the ANN model made is able to predict the
DCR value of the upper structure of the truss bridge and does not
experience overfitting. But the value of the referenced accuracy
parameter is still large so that the resulting predictions are not good and
require further research by increasing the amount of data and trying
other ANN architectures. | en_US |
