An Efficient Procedure for Identifying the Prediction Model Between Residual Stress and Barkhausen Noise

Journal of Water Process Engineering

Koivuranta

Leiviskä

2017

Self Cite

Wastewater samples taken from the aeration tank of a full-scale activated sludge plant were analyzed using an automatic optical monitoring device. Five variable selection methods were utilized to find the optimal subsets of input variables to develop predictive models for the important parameters of the wastewater treatment process efficiency and the quality of the effluent, including suspended solids, biochemical oxygen demand, chemical oxygen demand, total nitrogen and total phosphorus. The dependencies between the selected variables were also inspected. The study showed that the models based solely on the optical monitoring variables can be used to predict the level of the effluent quality parameters hours before the traditional sampling and analyses. Thus, predictive modelling based on the optical monitoring variables is a potential tool to be used assistance in a process control, keeping the process in a stable operating condition and avoiding environmental risks and economic losses.

Section: Variable Selectionmentioning

confidence: 99%

Predicting the effluent quality of an industrial wastewater treatment plant by way of optical monitoring

Journal of Water Process Engineering

Koivuranta

Leiviskä

2017

Self Cite

“…In this study, no interpolation or deletion of [15,16] For very large datasets two variable selection methods can be used one after another: one variable selection method is used for the variable elimination before the final variable selection by another method. In Sorsa et al [17] a successive projections algorithm (SPA) together with a modified genetic algorithm method was used and it was found that SPA applied before genetic algorithm search greatly improves the reliability of the genetic search. The genetic algorithm is able to find the global optimum well and the computational load is greatly reduced by SPA.…”

Section: Data Pretreatmentmentioning

confidence: 99%

Modelling effluent quality based on a real-time optical monitoring of the wastewater treatment process

Environmental Technology

Koivuranta

Kuokkanen³

et al. 2016

Self Cite

A novel optical monitoring device was used for imaging an activated sludge process in situ during a period of over one year. In this study, the dependencies between the results of image analysis and the process measurements were studied, and the optical monitoring results were utilized to predict the important quality parameters for the wastewater treatment process efficiency: suspended solids, biological oxygen demand, chemical oxygen demand, total nitrogen and total phosphorous in biologically treated wastewater. The optimal subsets of variables for each model were searched using five variable selection methods. It was shown that online optical analysis results have clear dependencies on some process variables and the purification result. The model based on optical monitoring and process variables from the early stage of the treatment process can be used to predict the levels of important quality parameters, and to show the quality of the biologically treated wastewater hours in advance. This study confirms that the optical monitoring method is a valuable tool for monitoring a wastewater treatment process and receiving new information in real time. Combined with predictive modelling, it has the potential to be used in process control, keeping the process in a stable operating condition and avoiding environmental risks.

“…In the earlier study [4], in addition to the temperature and anoxic proportion of volume, influent total nitrogen (12), influent sulphate (23), and mechanically treated wastewater nitrate nitrogen (17) and iron (26) were found important input variables in the suspended solids model, iron (26) was important in modelling the BOD, sludge concentration (30) and PO4-P (11) in COD model and total nitrogen (13), pH (21) and total phosphorus (9) of mechanical treated wastewater were important in nitrogen model, as mainly also in this study.…”

Section: Resultsmentioning

confidence: 99%

“…The amount of input variables should be kept decent because using too many input variables increases the risk to tuning parameter values, which are optimized manually one by one. [24,25] For a very large dataset one variable selection method, for example SPA, can be used for the variable elimination before the final variable selection by another method, for instance GA, to improve the reliability of selection [26].…”

Section: Variable Selectionmentioning

confidence: 99%

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

Environmental Technology

Leiviskä

2017

Self Cite

Traditionally the modelling in an activated sludge process has been based on solely the process measurements, but as the interest to optically monitor wastewater samples to characterize the floc morphology has increased, in the recent years the results of image analyses have been more frequently utilized to predict the characteristics of wastewater. This study shows that the traditional process measurements or the automated optical monitoring variables by themselves are not capable of developing the best predictive models for the treated wastewater quality in a full-scale wastewater treatment plant, but utilizing these variables together the optimal models, which show the level and changes in the treated wastewater quality, are achieved. By this early warning, process operation can be optimized to avoid environmental damages and economic losses. The study also shows that specific optical monitoring variables are important in modelling a certain quality parameter, regardless of the other input variables available.