Evaluating unidimensional convolutional neural networks to forecast the influent pH of wastewater treatment plants
| Autor: | Paulo Novais, Pedro Oliveira, Maria Alcina Pereira, Francisco Aguiar, Bruno Fernandes |
|---|---|
| Přispěvatelé: | Universidade do Minho |
| Jazyk: | angličtina |
| Rok vydání: | 2021 |
| Předmět: |
Time series
Wastewater Treatment Plants 010504 meteorology & atmospheric sciences Convolutional Neural Networks 0207 environmental engineering 02 engineering and technology 01 natural sciences Convolutional neural network Deep Learning Environmental science Sewage treatment Biochemical engineering Influent pH 020701 environmental engineering 0105 earth and related environmental sciences Engenharia e Tecnologia::Engenharia Eletrotécnica Eletrónica e Informática |
| Zdroj: | Intelligent Data Engineering and Automated Learning – IDEAL 2021 ISBN: 9783030916077 |
| Popis: | One of our society’s challenges today is water resources management due to its importance for human life. The monitoring of various substances present in wastewater is a crucial part of the process of Wastewater Treatment Plants (WWTPs). One of these substances is the influent’s pH, which plays a fundamental role in the nitrification and nitration processes. Hence, this paper presents a study to forecast the influent pH in a WWTP for the next two days. For this purpose, several candidate models were conceived, tunned and evaluated, taking into account the one-dimensional Convolutional Neural Networks (CNNs) considering two distinct approaches in the Pooling layer: the channels’ last and the channels’ first. The best candidate model obtained a Mean Absolute Error (MAE) of 0.257, following the channel’s last approach, compared to the channels’ first that obtained a MAE of 0.272. This work is financed by National Funds through the Portuguese funding agency, FCT - Fundação para a Ciência e a Tecnologia within project DSAIPA/AI/0099/2019. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|