Artificial intelligence techniques for flood risk management in urban environments

Abstract

Urban flooding is estimated to cause £270 million pounds worth of damage each year in England and Wales alone. There has, therefore, been a clear need to develop improved methods of identifying intervention strategies to reduce flood risk in urban environments. This paper describes ground-work performed towards evaluating the relative suitability of several algorithms applied to multi-objective optimisation of flood risk intervention strategies in an urban drainage network. An effective methodology is described for reducing an array of return period/duration rainfall files to a minimum, and it is described how this methodology makes possible comparisons of optimisation algorithms. This work has been undertaken as part of a STREAM-IDC EngD project which is a collaborative effort between the University of Exeter, and HR Wallingford.