Neural networks belong to the type of emperical tools, which compared to physically based models has the advantage of being self-calibrating. In return this type of models can not be generalized, because it only fits to the specific problem, for which it has been trained. The potential possibilities of using neural networks within urban hydrology will be investigated. The first step has been to identify and classify different neural networks and the next step will be concrete applications, which will prove the possibilities.
|Effective start/end date||01/01/1995 → 31/08/1996|