Application of genetic algorithms and fuzzy control to a combined sewer pumping station

In the present study, fuzzy logic control and genetic algorithms are applied to achieve improved pump operations in a combined sewer pumping station. Pumping rates are determined by fuzzy inference and fuzzy control rules corresponding to input variables. Genetic algorithms are used to automatically improve the fuzzy control rules through genetic operations such as selection, crossover and mutation. The effects of different fitness functions and learning conditions are investigated using a stormwater runoff model. It is found that current pump operations can be improved by adding the sewer water quality to the input variables and to the fitness function; the improved operations can reduce not only floods in the drainage area but also pollutant loads discharged to the receiving waters.