Application of a Moving-Window-Adaptive Neural Network to the Modeling of a Full-Scale Anaerobic Filter Process

Abstract

To explore the complex dynamics of a full-scale anaerobic filter process treating the wastewater from a purified terephthalic acid manufacturing industry, a new modeling approach based on a moving-window-adaptive neural network is proposed. The essential feature of this modeling approach is that the neural network model is automatically updated whenever a new data block is available so that it can effectively capture the slowly changing process dynamics. To elucidate the advances of the proposed method, four different modeling approaches combined with the concepts of autoregressive with exogenous (ARX) input and a finite impulse response model were evaluated and compared. During each model identification process, a modified crossvalidation technique was used to avoid the overfitting problem of a neural network. Among the tested models, a moving-window-adaptive ARX neural network model showed the best prediction ability with the smallest validation error. To investigate the feasibility of this model, various dynamic simulations were performed. Although some limitations such as ambiguity in sensitivity analysis and instability in long-term simulation were identified, it is considered that the movingwindow-adaptive ARX neural network model could provide a useful guideline to explore the complicated dynamics of the anaerobic filter process.

Hybrid water treatment cost prediction model for raw water intake optimization

“…For instance, Baruch et al (2005) built three adaptive neural network control structures to regulate a biological wastewater treatment process. Another neural networkbased control system was developed by Lee et al (2005) to efficiently operate small plants with significant variance in the influent loadings. They used an internet-based remote monitoring system that input oxidation-reduction potential (ORP) as the main sensor for the control.…”

Nonlinear semisupervised principal component regression for soft sensor modeling and its mixture form

“…In contrast, through building relationships between the quality variable and other easy-to-measure secondary variables based on the historical dataset of the process, the data-based method can often successfully construct empirical models for soft sensor utilizations [1,2]. Conventionally used data-based soft sensors include principal component regression (PCR) [3][4][5][6], partial least squares (PLS) [7][8][9][10], artificial neural networks [11][12][13][14][15], kernel learning methods [16][17][18][19][20][21], and Bayesian inference-based approach [22][23][24].…”

Treatment of Wastewater from Purified Terephtalic Acid (PTA) Production in a Two-stage Anaerobic Expanded Granular Sludge Bed System

“…The PTA scrubbing section generates the maximum quantity of wastewater corresponding to 5-20 kg-COD/m 3 with 3-10 m 3 wastewater/t PTA production [3,4]. Accordingly, terephthalate, acetate, benzoate, and p-toluate are the major organic contaminants, but trimellitic acid, o-phthalic acid and 4-carboxybenzaldehyde are also contaminants found to be removed in the wastewater [5,6]. Anaerobic processwas developed to treat more complicated and higher strength organic wastewater such as aromatic compounds in the PTA wastewaters [7,8].…”