Fuzzy Model Predictive Control for Wastewater Treatment Process under Multi-Operating Conditions

The operating conditions of wastewater treatment process is time-varying and nonlinear. In the study of biochemical reaction process, there are coupling relations and size differences among variables of non-single controlled objects. In order to solve the multi-variable control problem, this paper use the Model Predictive Control (MPC) based on neural network. A number of control factors are considered in the design process and by constructing the cost function, the controller is developed in the expected direction. In addition, at the actual working condition change, the single working condition change is very difficult to maintain for a long time. In this paper, a multi-condition identification model predictive control is established by using T-S model. The scheme can make the controller more fit with the change of state. In order to verify the effectiveness of the designed controller and guarantee the data authenticity, the multi-state condition of this paper is constructed by using the real data provided by BSM1.The experimental results show that the designed controller is more stable at single condition and multi-condition, and the control accuracy has been improved obviously. The advantages of the controller are also illustrated by the evaluation indexes of IAE (Integral Absolute Error) and ISE (Integral Square Error).