Research Article

Prediction of Effective Width of Varying Depth Box-Girder Bridges Using Convolutional Neural Networks

Table 3

Statistical parameters of 2 models.

ModelPositionMean percentage error (%)Max percentage error (%)Variance of percentage error (%)

ANNPier0.6190.3934.1901.9770.3290.096
Midspan0.7100.3276.2631.8410.4030.081

CNNPier0.5300.2333.1691.8040.3050.041
Midspan0.4720.1941.9191.0360.1380.028