An optimal control of wastewater treatment plants (WWTP) has to account for changes in the bio-chemical state of the bioreactors. As many process variables of a WWTP are not measurable online, the development of an efficient control strategy is one of the greatest challenges in the optimization of WWTP operation. This paper presents an approach, which combines the use of Self-Organizing Maps (SOM) and a clustering algorithm to identify operational patterns in WWTP process data. These patterns provide a basis for the optimization of controller set points that are well suited for the previously identified operation regimes of the plant. The optimization is performed using Genetic Algorithms. This approach was developed, tested and validated on a simulation model based on the Activated Sludge Model No.1 (ASM1). The results of this state-based control indicate that the presented methodology is a promising and useful control strategy that is definitely able to address the distinctive energy and effluent limit challenges faced by WWTP operators.