The wastewater treatment sector has a major impact on the aquatic environment by preventing contaminants from entering. However, wastewater treatment plants are complex systems. The operator simultaneously needs to safeguard the stability of the plant, make sure to meet the effluent standards, and keep the operational costs as low as possible. Unfortunately, often only the first objective (stability) is achieved in practice. There is thus ample room for improvement of the operational control strategy. We identified the following areas of improvement. The plant should be considered as a whole by implementing 1) a plant-wide control strategy that uses 2) all available data. 3) Temporal trends should be taken into account and 4) the control strategy should adapt continuously to changes in the process. By taken these considerations into account, the operational performance of wastewater treatment plants can be improved. We propose the use of Deep Reinforcement Learning (RL) as an alternative control strategy. Deep RL is a machine learning technique that automatically learns to take optimal actions through repeated interaction with the wastewater treatment plant. Moreover, Deep RL meets all four of the abovementioned areas of improvement. In this proposed research project, we will investigate the performance of Deep RL as a control strategy in theoretical and real wastewater treatment plants and ultimately perform a full-scale implementation.