Comparison of modelling accuracy with and without exploiting automated optical monitoring information in predicting the treated wastewater quality

Tomperi, Jani; Leiviskä, Kauko

doi:10.1080/09593330.2017.1331267

Cited by 2 publications

(2 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Due to the available instrumentation in the Hias Process, this approach cannot be tested either. Data-driven methods do not require a specific set of instrumentation, and several recent papers have reported successful applications to effluent prediction using multivariate linear regression (Tomperi & Leiviskä, 2018), feedforward-backpropagation networks (El-Rawy et al, 2021), and time-series models and machine learning models (Ly et al, 2022). In a recent master thesis (Nermo, 2023) transfer function models were developed to predict the Hias Process effluent phosphorus.…”

Section: Introductionmentioning

confidence: 99%

Estimation of effluent nutrients in municipal MBBR process

Komulainen,

Baqeri,

Nermo

et al. 2023

Linköping Electronic Conference Proceedings

View full text Add to dashboard Cite

The recently updated European Union’s Urban Waste Water Treatment Directive proposal, European Green Deal, Biodiversity Strategy for 2030, and EU’s Energy System Integration highlight a pressing need for innovative biological nutrient removal processes and energy-efficient control methods to reduce pollution and minimize the carbon footprint at water resource recovery facilities. The aim of the PACBAL research project is to develop estimation methods for nutrient profile in a novel industrial Moving Bed Biofilm Reactor (MBBR) process. This study devises and assesses a range of data-driven methods to estimate effluent phosphorus concentration by utilizing a combination of real sensors with software models. The resulting virtual sensor could facilitate the design of energy-efficient control strategies. The case study data are collected from the MBBR process at Hias water resource recovery facility in Norway. Data sets from December 2022 to March 2023 include varying weather conditions, such as rain, dry, and snow. The Hias Process consists of three anaerobic and seven aerobic zones, where biomass carriers removes over 90 percent of the phosphorus from the wastewater in simultaneous biological processes. The industrial online measurements include wastewater flowrate, aeration rates, dissolved oxygen and nutrients COD and NO2/ NO3 at inlet and total suspended solids at outlet. Dynamic data-driven models indluding transfer functions, state-space models and ARX models, were developed and compared to estimate the outlet phosphorus concentration. Model fitness to validation data was around 7% with ARX models, and up to 18% with tranfer function models and state-space models. The first and second order models gave similar results. The state-space models will be developed further and implemented to into virtual sensors that will enable energy-efficient control strategy development.

show abstract

Section: Introductionmentioning

confidence: 99%

Estimation of effluent nutrients in municipal MBBR process

Komulainen,

Baqeri,

Nermo

et al. 2023

Linköping Electronic Conference Proceedings

View full text Add to dashboard Cite

show abstract

“…Tomperi et al [15] modeled the suspended solids using incoming iron concentration, mean and median area of flocs, and amount of filaments. Very recently, real-time optical monitoring of the wastewater treatment plant was proposed by the image analysis-based optimal models [16]. Honggui et al [17] proposed a fuzzy neural network-based modeling of SVI using physicochemical measurements as input to the network.…”

Section: Introductionmentioning

confidence: 99%

Generalized modeling of the sludge volume index and suspended solids to monitor activated sludge bioflocculation using image analysis

Khan¹,

Nisar²,

Ng³

2020

Environ. Protect. Eng.

View full text Add to dashboard Cite

The performance of an activated sludge wastewater treatment plant depends on bioflocculation that is monitored by physical measurements such as the sludge volume index (SVI) and mixed liquor suspended solids (MLSS). The estimation of SVI and MLSS has been proposed using image analysis based modeling which is time-efficient and valid for multiple plants operating in different states. The methodology includes the sequence of image acquisition using bright-field microscopy, a robust segmentation of flocs, partitioning of flocs based on different ranges of their equivalent diameters, extraction of morphological features, and modeling of SVI and MLSS using the features. It is proposed that bright-field microscopy at lower magnification to capture the flocs is sufficient to model SVI and MLSS. A robust approach for image segmentation is adopted by integrating state-of-the-art image segmentation algorithms. It is hypothesized that flocs in different ranges of equivalent diameter respond differently to the variation in the operating state. Hence, flocs and their respective image analysis features are categorized based on the range of equivalent diameter. Finally, stepwise regression is used for feature selection and model identification to explore the feasibility of generalization of models to multiple plants in different states regarding SVI and MLSS.

show abstract

Comparison of modelling accuracy with and without exploiting automated optical monitoring information in predicting the treated wastewater quality

Cited by 2 publications

References 23 publications

Estimation of effluent nutrients in municipal MBBR process

Estimation of effluent nutrients in municipal MBBR process

Generalized modeling of the sludge volume index and suspended solids to monitor activated sludge bioflocculation using image analysis

Contact Info

Product

Resources

About