The paper reports on the latest developments of a neural network-based method of accurately estimating truck attributes (such as axle loads) from strain response readings taken from the bridge over which the truck is traveling. The approach is designed to remove the need for intrusive devices (such as tape switches) on the deck of the bridge to obtain such data so as to provide a convenient and viable means of collecting bridge loading statistics. Specifically, this paper compares the performance of three radically different types of neural network used for identifying the class of truck crossing the bridge. Of the methods considered, a binary networking system is found to be the most efficient. The paper concludes with some recommendations for further study.
[1]
Ian Flood.
Modeling dynamic engineering processes using radial-Gaussian neural networks
,
1999,
J. Intell. Fuzzy Syst..
[2]
Pedro Albrecht,et al.
Computing truck attributes with artificial neural networks
,
1994
.
[3]
John Moody,et al.
Fast Learning in Networks of Locally-Tuned Processing Units
,
1989,
Neural Computation.
[4]
F Moses.
Instrumentation for weighing trucks in motion for highway bridge loads
,
1983
.
[5]
Nabil A. Kartam,et al.
A Binary Classifier with applications to poorly defined engineering problems
,
1998,
Artif. Intell. Eng. Des. Anal. Manuf..