Neural network–genetic programming for sediment transport

Abstract

The planning, operation, design and maintenance of almost all harbour and coastal engineering facilities call for an estimation of the longshore sediment transport rate. This is currently and popularly done with the help of empirical equations. In this paper an alternative approach based on a combination of two soft computing tools, namely neural networks and genetic programming, is suggested. Such a combination was found to produce better results than the individual use of neural networks or genetic programming. The ability of the neural network to approximate a non-linear function coupled with the efficiency of the genetic programming to make an optimum search over the solution domain seems to result in a better prediction.