Neural Network Predictive Tool of Ground Settlement Due to Dewatering Activities

May 1999

Proceedings of the Twelfth International Florida Artificial Intelligence Research Society Conference (FLAIRS 1999)

Volume

Proceedings of the Twelfth International Florida Artificial Intelligence Research Society Conference (FLAIRS 1999)

All Papers

The present paper reports the implementation of a neural network technique to predict ground settlement as occasioned by dewatering activities of a long-term large-scale drainage project. The objective of the project is to lower and maintain the level of the ground water table in urban areas through a system of a nonstop pumping wells whose layout and operation rate are determined and implemented in a pilot project at a district level. To make use of the wealth of data collected and to aid in the decision making regarding extending the project to other areas, a neural network predictive tool is developed to estimate ground settlement induced by the drainage activites near and further away from the pumping wells. The backpropagations architecture is employed to design two networks for drawdown settlement relationship and attenuation of settlements further away from the pumping wells. Incorporating relevant soil properties in the input nodes considerably improved the performance of both networks.