Simulation of Water Distribution System Using Deep Learning Approaches

Deep learning, often known as DL, is an alternative to traditional ANNs in that it may build an unguided model out of a given dataset by learning the features of that dataset in layers. DL has been effective in a wide variety of other applications as well, including big data analytics, the identification of objects in pictures, the categorization of data, and voice recognition. This study presents a proposal for an integrated data-driven modelling method that integrates DL with a well-known evolutionary optimisation tool. This paradigm for high performance computing is mixed and scalable, and it is the basis for the research. Modellers are now able to construct more realistic simulations by making optimal use of the data that is accessible, thanks to the connected infrastructure. Simulation of water distribution networks, optimisation of those networks, and decision making that is driven by data are only some of the uses that have been proven for the technology. Both the development of a metamodel to replace physics-based models (hydraulic and water quality) for regulating water distribution and the identification of irregularities in time series data (including pressures, flows, and consumptions) are the main objectives of this study.