Abstract

Nowadays, the management of pressurised irrigation networks requires plenty of information to provide an efficient and reliable service to farmers. Perturbation is the propagation of pressure waves through the networks pipes which could expose the network to a serious risk that could cause components damaging. Several computational codes were developed to simulate such phenomenon. The most recent ones are efficient enough to provide a good image of the perturbation occurrence through different indicators, but they are time and computationally expensive. For real time decision making and more flexible management, there is a need for faster models to be developed. In this study the directly programmed models were used as big data generators to train a developed deep learning model. This approach was applied on a pressurised on-demand irrigation system located in south of Italy that consists of 19 hydrants (service outlets) and covers 57 ha. Two thousand configurations (operational scenarios) were simulated using a predeveloped directly programmed model and fed to train a deep learning model with the objective of forecasting the maximum pressure occurred due the perturbation at each section. The occurred pressure is represented as classes according to the case sensitivity and the required precision. In the present work, scenarios for 1, 2 and 3 bars steps were simulated. The model proved to be significantly time saving compared to previous approaches as the results are produced instantaneously with a forecasting accuracy of 85 %. Furthermore, from the called confusion matrix, the error committed by the model is of one class lower or higher that may be considered tolerable according to the system sensitivity. Thus, modelling the perturbation in the on-demand pressurised irrigation networks would add a significant contribution to provide practical recommendations for real-time decision-making processes.