Controlling biological wastewater treatment plants using fuzzy control and neural networks

Improving the performance of wastewater treatment plants by optimising the control systems is a very cost efficient method but it involves risks caused by the time variant and non linear nature of the complex biochemical processes. Key problem is the removal of nitrogen combined with an optimal processing of the sludge water. Different control strategies are discussed. A combination of neural network for predicting outflow values one hour in advance and a fuzzy controller for dosing the sludge water are presented. This design allows the construction of a highly non-linear predictive controller adapted to the behaviour of the controlled system with a relatively simple and easy to optimise fuzzy controller. The system has been successfully tested on a municipal wastewater treatment plant of 60.000 inhabitant equivalents.