Dynamic multi-objective optimization control for wastewater treatment process based on modal decomposition and hybrid neural network

Qing Liu,Xiangyuan Jiang

DOI: https://doi.org/10.1016/j.jwpe.2024.105274

IF: 7

2024-04-18

Journal of Water Process Engineering

Abstract:To achieve the desired quality of wastewater treatment and minimize energy consumption, the paper proposes a multi-objective optimization control algorithm in the framework of improved modal decomposition, meta-heuristic algorithm, and hybrid neural network, by establishing an accurate objective function model and obtain effectively optimized setpoints for control variables in the wastewater treatment process. Firstly, in order to improve the accuracy of the objective optimization model, the parameters of the improved complete ensemble empirical modal decomposition with adaptive noise (ICEEMDAN) are optimized using the improved dung beetle optimizer (IDBO) algorithm, and the raw influent data undergo decomposition. The decomposed data is then fed into the proposed hybrid neural network model for the prediction of the performance metrics of wastewater treatment. Secondly, the paper employs the algorithm of multi-objective ant lion optimizer (MOALO) to optimize water quality and energy consumption and obtain dynamic optimal setpoints. Finally, A Linear active disturbance-rejection control (LADRC) is utilized to track and control the setpoints. The proposed method is validated using the BSM1 simulation model, and the results demonstrate a significant reduction in both energy consumption (EC) and improvements in effluent quality index (EQ) compared to other methods.

engineering, chemical, environmental,water resources