The wastewater treatment process (WWTP) is highly nonlinear and complex, and its dynamics are difficult to model. Process data obtained under multiple operating conditions that have a number of operating models and changes in operating conditions, such as biological WWTP, should be modeled carefully. In this paper, we propose a new hybrid technique for modeling wastewater quality that is based on the Takagi-Sugeno-Kang fuzzy model with artificial intelligence learning and employs real-coded genetic algorithms. The hybrid learning framework integrates the multiple fuzzy model, which can find the rule base and the membership functions of the fuzzy model simultaneously. We applied the proposed method to predict the effluent chemical oxygen demand in a full-scale coke wastewater treatment plant. The modeling results show that the hybrid model can accurately model the real nonlinear processes under diverse operating conditions, and that it is able to identify various operating regions in a full-scale WWTP.
All Science Journal Classification (ASJC) codes
- Environmental Chemistry
- Waste Management and Disposal