LSTM-OBE based Interval Prediction of Effluent BOD for Wastewater Treatment
Meng Chu Zhou, Yinyue Zhang, Jing Wang, Tonglai Xue · IFAC-PapersOnLine · 2023
In this paper, a new interval prediction method of effluent BOD for wastewater treatment is proposed. Firstly, a point prediction method of effluent BOD based on a long short term memory (LSTM) neural network model is generated. Then, an improved LSTM-MTOBE algorithm is designed to predict the interval of effluent BOD, in which the minimum trace optimal boundary ellipsoid (MTOBE) algorithm is combined with LSTM to identify the real parameters of its output layer weights. Next, based on the properties of ellipsoids, the determination interval of the model uncertainty is obtained and analyzed. Finally, compared with some existing algorithms, the proposed method is illustrated to have better prediction accuracy and interval performance.