APPLICATIONS OF ARTIFICIAL NEURAL NETWORKS TO LAND DRAINAGE ENGINEERING

An artificial neural network (ANN) model was developed and trained by using the simulated midspan water table depths from DRAINMOD, a conven tional water table management model. Compared to DRAINMOD, the model i s very simple to run, and requires only a small amount of data, such a s precipitation, evapotranspiration, and initial midspan water table d epth. The results indicate that the ANN model can make predictions sim ilar to DRAINMOD, with the least root mean square error of 0.1193, and doing this significantly faster and with fewer input data. The result s also indicated that the successful prediction of midspan water table depths depends upon the inclusion of data indicating average as well as extreme conditions, in order to train the ANNs. Given such data, AN Ns perform well under general conditions. Generally, the ANN structure with six processing elements and one hidden layer was sufficient for this study. It was found that the networks should be trained with at l east 145,000 cycles, bur more than 200,000 cycles are unnecessary. A f eedback procedure was implemented which fed the previous water table d epth output back into the current input. In addition, a lag procedure was suggested which improved the performance elf ANNs under irregular situations, such as sudden and large rainstorms. A three-day lag of al l input parameters was the best choice when the weather conditions wer e irregular. The benefits of ANNs are speed, accuracy, ease-of-use and flexibility, thus making ANN models suitable for water table manageme nt systems that require a real-time control.