Modeling and predictive analyses related to piezometric level in an earth dam using a back propagation neural network in comparison on non-linear regression

The objective of the dam safety guidelines and rules is to provide design engineers with information on dam planning, design, construction, operation and maintenance activities, procedures and requirements. However, the hydraulic and mechanical properties may react differently than expected based on the design or in rare case dam fail. The purpose of the dam monitoring is to illustrate the dam response face different solicitations. Furthermore, the results of monitoring dam are used to determine the kind and timing of maintenance or repair procedures while also keeping tabs on how the structure's behavior evolves over time. This can make it possible to compare the performance of the structure to project forecasts. In this paper, a typical zoned dam with central core El-Izdihar dam, located in the northwest of Algeria was considered as an example. The non-linear regression models combined with the theory of Back Propagation Neural Network (BPNN) were adopted to ascertain the optimum model for the prediction of the piezometric levels in the dam. The performance of the model has been judged using the correlation coefficient R, and Root Mean Square Error (RMSE). The achieved result shows that the BPNN model has the ability to predict piezometric level, and make that more accurate and reasonable even in case of nonlinear analyses.