Abstract: Wastewater treatment is a highly nonlinear process, the water quality changes violently,and there exists strong coupling among the state variables.For these problems,this paper proposes an adaptive fuzzy neural network controller.Simulation results show that this controller can adjust the membership function of input variables and optimize the control rules in an on-line way.When applying to the wastewater treatment system based on activated sludge,this method can remove contamination quickly,makes the sludge maintain good performance of decontamination and deposition,and is of great robustness.