Artificial Neural Networks in Water Distribution Systems: A Literature Synopsis

High computational requirements are commonly associated with the hydraulic simulation of large-scale water distribution. The convergence of the cumbersome iterative procedures involved has been a well-debated issue for the past decades. The large-scale and non-linear properties pose a great hindrance towards the development of online applications for water distribution network (WDN) analysis and pressure control thereof. Consequently, there has been a great interest in the deployment of model-free techniques to mimic the rather computationally expensive non-linear hydraulic simulations. As the hydraulic simulation based research is still being conducted, the advantages of model-free techniques make them more suitable alternatives. Artificial neural networks (ANN) is one of the most successful model-free methods for WDN analysis and management. In this paper, a literature synopsis of existing applications of model-free approaches in water distribution is presented. The technical advantages of applying such technique in a large-scale non-linear network are brought up in this paper.