Adaptive critic control for wastewater treatment systems based on multiobjective particle swarm optimization

Given the need to ensure that effluent quality meets the standards and reduces energy consumption in urban wastewater treatment systems, the operation process is considered a multiobjective optimization control problem. An adaptive critic control scheme is developed based on multiobjective particle swarm optimization. This scheme is divided into two parts: upper optimization and bottom tracking control. First, considering the characteristics of nonlinear, multivariable, and large time variation in a wastewater treatment system, the mechanism model is difficult to establish accurately. To preserve quality and reduce consumption, an accurate operation index model of the wastewater treatment process must be designed. The data of the inlet and outlet components are analyzed using a data-driven framework. A multiobjective optimization model reflecting effluent quality and energy consumption is constructed. A radial basis function neural network is used for modeling and compared with a back-propagation neural network. Then, combined with powerful optimization capabilities, the multiobjective particle swarm optimization algorithm is used to solve the multiobjective optimization problem. Combining the practical importance of the two indicators of energy consumption and water quality, a decision method is designed to select the preferred solutions from the optimal solution set. The preferred solutions can be defined as the optimal set concentrations of dissolved oxygen and nitrate nitrogen. Next, the bottom tracking control part adopts an auxiliary controller based on adaptive dynamic planning to supplement the control strategy of a proportional–integral–derivative algorithm, compensating for the shortcomings of the poor adaptive ability of the traditional control algorithm. In addition, this proportional–integral–differential algorithm provides an initial stable control strategy for the adaptive dynamic programming algorithm, overcoming the poor control effect of the learning algorithm in the early stage and ensuring the safety and reliability of the wastewater treatment process. Ultimately, the controller successfully achieves the tracking control of the optimal setting value. To verify the optimization effect and control performance of the proposed scheme, we use benchmark simulation model no. 1 to complete the simulation. Using the indicators of water quality and energy consumption, we also compare the proposed scheme with other multiobjective optimization schemes. The results show that the proposed algorithm effectively improves the operational performance of the wastewater treatment process. It not only ensures that the effluent water quality meets the standards but also effectively reduces the energy consumption of wastewater treatment.