Research on Dosing Time Series Prediction in Wastewater Turbidity Removal

• Published in: 2022 7th International Conference on Computational Intelligence and Applications (ICCIA) pp. 103 - 107
• Main Authors: Chen, Yuan, Zhai, Zhengang, Zhu, Yunya, Fang, Xusheng, Wang, Jiangang, Wang, Biao
• Format: Conference Proceeding
• Language: English
• Published: IEEE 24.06.2022
• Subjects: Adaptation models; Analytical models; DA-RNN; Effluents; MARS; Predictive models; Time series analysis; time series prediction; Time-frequency analysis; turbidity removal treatment; Water quality
• DOI: 10.1109/ICCIA55271.2022.9828409

In wastewater treatment plants, turbidity is an essential and universal indicator to measure water quality. How to quickly determine the dosing frequency in turbidity removal treatment online is important for efficient and intelligent control. After analyzing potential relationships contained in the multi-dimensional water quality data, this paper introduce multivariate adaptive regression splines (MARS) and dual-stage attention-based RNN (DA-RNN) to construct time series regression models respectively to achieve high-precision of dosing frequency based on inlet and outlet water quality data. The experiment of 95-day water quality data from a wastewater plant has comprehensively verified that the models can mine the features of high-dimensional data and realize time series prediction with goodness of fit and prediction accuracy.