Modeling of an activated carbon sludge process for effluent prediction – a comparative study using ANFIS and GLM regression
No Thumbnail Available
Date
2018
Journal Title
Journal ISSN
Volume Title
Publisher
Springer
Abstract
In this paper, nonlinear system identification
of the activated sludge process in an industrial wastewater
treatment plant was completed using adaptive
neuro-fuzzy inference system (ANFIS) and generalized
linear model (GLM) regression. Predictive models of
the effluent chemical and 5-day biochemical oxygen
demands were developed from measured past inputs
and outputs. From a set of candidates, least absolute
shrinkage and selection operator (LASSO), and a fuzzy
brute-force search were utilized in selecting the best
combination of regressors for the GLMs and ANFIS
models respectively. Root mean square error (RMSE)
and Pearson’s correlation coefficient (R-value) served as
metrics in assessing the predicting performance of the
models. Contrasted with the GLM predictions, the obtained
modeling results show that the ANFIS models
provide better predictions of the studied effluent variables.
The results of the empirical search for the dominant
regressors indicate the models have an enormous
potential in the estimation of the time lag before a
desired effluent quality can be realized, and preempting
process disturbances. Hence, the models can be used in
developing a software tool that will facilitate the effective
management of the treatment operation.
Description
Keywords
Wastewater treatment process modeling, Predictive models, ANFIS, Fuzzy exhaustive search, GLM regression, LASSO regularization