Design of smart urban drainage systems using evolutionary decision tree model

Abstract

Recently, as an alternative method for monitoring of drainage systems, Internet of Things (IoT) technology is initiated in smart cities. IoT is used for detection of the location of the sediment deposition within the drainage pipe system to alert for repairing before complete blocking. However, from the hydraulic point of view, it is reasonable to design the drainage and sewer pipes to prevent the deposition of the sediment based on the physical parameters. To this end, instead of detection of blockage location, monitoring the flow characteristics is of more importance to keep pipe bottom clean from sediment deposition. Accordingly, smart sensors mounted in the drainage and sewer pipes should read the flow velocity and alert once the flow reaches a velocity in which sediment deposition is occurred. In order to determine the sediment deposition velocity, this study models sediment transport in drainage systems by means of evolutionary decision tree (EDT) technique. EDT results are compared with conventional decision tree (DT) and evolutionary genetic programming (GP) techniques. A large number of experimental data covering wide ranges of sediment and pipe size were used for the modeling. Evaluation of the developed models in terms of verity of statistical indices showed the outperformance of the proposed EDT model. The EDT, DT and GP models were found superior to their traditional corresponding regression models existing in the literature. Results are helpful for determination of the flow characteristics at sediment deposition condition in drainage systems maintained using IoT technology in smart cities.