Probabilistic forecasting of rainfall response in a Danish stormwater tunnel

Sustainable urban drainage is an economically expensive necessity, partially due to the operation of water pumps. Reliable forecasting stormwater response following a rainfall event has potential reduce those expenses, because it can be used in model predictive control schemes that optimize energy consumption pumps significantly better than commonly applied real-time systems. Urban systems are traditionally designed around highly complex, deterministic models where assessment uncertainty forecast either absent or relies on computation–heavy simulations. With offset Danish tunnel, we propose much faster, but reliable, non-linear continuous-discrete-time state-space based stochastic differential equations which generate probabilistic forecasts contain complete information about distribution uncertainty. We explain step-by-step how structure built from simple physical assumptions, then parameters estimated maximum likelihood principles and finally demonstrate capabilities model. believe this would well-suited for subsequent scheme.