International Journal of Interdisciplinary Research
Intelligent techniques and optimisation algorithms | Research content | Prediction accuracy | References |
---|---|---|---|
ANN | This research identified the optimal conditions for the decolourisation process of C.I. Direct Red 23 dye solution by investigating how operating parameters affect decolourisation performance | The achieved prediction performance was satisfactory with an R2 value of 0.958 | Khataee et al. (2013) |
This research investigated the prediction of optimal conditions for biodegradation of Remazol Brilliant Blue R dye using a MLP neural network | The optimal biodegradation conditions predicted by the MLP neural network resulted in a biodegradation index of 96% | Rosaa et al. (2013) | |
This research focused on using ANNs, specifically by UV/H2O2, to remove chemical oxygen demand (COD) and colour from extracted wastewater from textiles | The neural model achieved a correlation coefficient of 99% for COD and 97% for colour | Yonar and Kilic (2014) | |
This research evaluated the nZVI-Fenton process for reducing colour and COD from azo-dye textile wastewater using a regression model and an ANN | The ANN provided very accurate predictions with R2 values ranging from 0.96 to 0.99, while the regression model provided predictions with R2 values between 0.92 and 0.95 | Yu et al. (2014) | |
This research focused on modelling the coagulation process of chitosan extracted from spent shrimp shells for the treatment of reactive dye in an aqueous solution | R2 between the estimated and observed outputs was 0.986 and the root mean square error (RMSE) was 2.951 | Bui et al. (2016) | |
This research focused on modelling and evaluating the electro-oxidation of wastewater containing CBSOL LE red wool dye | The regression coefficient between the training and target values was 0.995, indicating that the ANN model was trained accurately | Kaur et al. (2015) | |
This research focused on predicting the removal of Disperse Blue 79 dye colour from water treatment residues using different ANNs | The R2 values of the ANNs models were greater than 0.90, indicating a high level of accuracy in predicting the removal of Disperse Blue 79 dye from water treatment residues | Gadekar and Ahammed (2016) | |
This research investigated the prediction of dye removal efficiency (colour and COD values) of Sunfix Red S3B in an aqueous solution during electrocoagulation | The low RMSE value (9.844%), MAE (13.776%) and high R2 (0.836) validated the validity of the ANN model predictions | Manh (2016) | |
This research extended an ANN to estimate the degradation efficiency of acid blue by martensitic nanoparticles of different sizes based on experimental results | The results showed suitable performance with R2 = 0.955 | Rahmani et al. (2016) | |
This research investigated the use of heterogeneous sono-Fenton-like treatment to remove acid blue 92 and estimated its removal efficiency using an ANN | The correlation coefficient (R2) was found to be 0.9836, based on experimental data, indicating a high level of accuracy in estimating the removal efficiency of acid blue 92 in the treatment process | Dindarsafa et al. (2017) | |
This research focused on predicting the removal efficiency of suspended matter and colour in real textile wastewater after chemical coagulation and sedimentation | The ANN models provided accurate predictions with R2 ranging from 0.93 to 0.96 for both suspended matter and colour removal efficiency | Yu et al. (2017) | |
This research involved the analysis and modelling of reactive grey BF-2R dyes using ANNs | The ANN model accurately predicted the experimental data, with a correlation exceeding 0.999 | do Nascimento et al. (2018) | |
This research aimed to develop a prediction model for treating dyes in wastewater by adding alum through electrocoagulation of a rotating anode, using regression analysis and an ANN model | The results showed that the ANN model performed better than the regression analysis method, and the R2 value was 0.928, indicating a high level of accuracy | Taha et al. (2020) | |
This research used ANN to predict parameters related to the plasma decolourisation of organic dyes [reactive orange 16 (RO 16), reactive blue 19 (RB 19), and direct red 28 (DR 28)] in wastewater. The oxidation was the key parameter for decolourisation performance of RO 16 and DR 28, while the argon flow rate was the main parameter for decolourisation of RB 19 | ANN can be used to optimise the processing parameters to enhance removal efficiency | Mitrović et al. (2020) | |
This research aimed to predict the removal efficiency of pollutants using FeCl3 coagulant under different variables | The output results all had R2 above 0.90, indicating high accuracy and the ability to simulate the coagulation process | ||
This research focused on estimating the removal efficiency of coagulation and adsorption of iron/copper nanoparticles during textile wastewater treatment using ANNs | The R2 value between the predicted and actual values exceeded 0.98, indicating high accuracy, and the ANN could be used with confidence to predict the removal efficiency of the target parameters | Mahmoud et al. (2021) | |
This research aimed to optimize and predict the adsorbent on the reactive Pr Red Hegxl dye using factorial design methods and ANNs | The proposed model was able to reasonably predict the removal efficiency (%) with a model correlation coefficient R2 greater than 0.97 and RMSE less than 0.0317 | Yargic et al. (2021) | |
ANN and GA | This research focused on the use of an ANN model to optimize and predict the behaviour of an adsorption system for the removal of phenol red by Au-NP-AC and TiO2-NP-AC loaded on activated carbon | The ANN model had a mean squared error (MSE) of 3.19 × 10–4 and 0.0022 for Au-NP-AC and TiO2-NP-AC loaded on activated carbon, respectively, and fitted R2 of 0.9962 and 0.9729, respectively, indicating that it could fit well with the removal efficiency of phenol red | Ghaedi et al. (2014) |
RSM and GA | This research aimed to optimize the photocatalytic degradation of Basic Blue 41 (BB41) and Basic Red 46 (BR46) dyes using RSM and GA methods | The optimal conditions for the photocatalytic degradation of BB41 and BR46 were found to be 72.56% and 67.89% using RSM and 72.36% and 68.34% using GA, respectively. The values predicted by both methods closely matched the experimental values | Mahmoodi et al. (2017) |
ANFIS and GA | This research focused on estimating the dye and silver nanoparticle concentrations when treating silk fabrics with nanosilver using spectrophotometric colour matching | The ANFIS method outperformed the GA method in predicting the dye and AgNP concentrations, with MAE and RMSE values for the ANFIS method being 0.087 and 0.103 for the dye concentration, and 0.002 and 0.003 for the AgNP concentration | Nateri et al. (2019) |
ANN and RSM | This research aimed to optimize the electrochemical degradation of synthetic wastewater containing reactive black 5 textile dye by using RSM and ANNs | The final neural network model showed an exponential R2 of 0.982 and an MSE of 0.0146, indicating effective predictive performance. The optimized conditions resulted in complete colour removal and up to a 73.77% reduction in COD in a 180-min treatment process | Viana et al. (2018) |
This research conducted a comparative study of textile wastewater decolourisation using nano zerovalent iron, activated carbon, and green synthetic nano zerovalent iron, and utilized ANNs and RSM for modelling and prediction of the colour adsorption process | The ANN data matched the regression analysis results from the RSM optimization, indicating that both ANN and RSM models were able to predict and model the colour adsorption process | ||
This research investigated the effectiveness of nanomaterial adsorbents for dye adsorption and the effect of operational parameters on it | The experimental colour removal rates matched the predicted values obtained by using RSM-ANN modelling | Thomas et al. (2021) | |
This research investigated the optimization of the electrochemical oxidation process of textile wastewater at graphite electrodes using RSM and ANN techniques, with a focus on the influence of pH, time, and current on the process | The R2 correlation coefficients of the RSM and ANN models for colour removal were 0.91 and 0.98, respectively, indicating a high level of accuracy in predicting colour removal during the electrochemical oxidation process | Saleh et al. (2021) | |
This research aimed to predict and optimize the electrocoagulation process of cationic dyes using an ANN combined with an RSM | The R2 of the model was close to 1, indicating that the process was successfully trained by the ANN and optimized by the RSM | Kothari et al. (2022) | |
RSM and GA-ANN methods | This research focused on modelling and optimizing the removal of azo dyes in the sono-Fenton process using RSM and GA-ANN methods | When conditions were optimal, the prediction errors of the hybrid GA-ANN and RSM were 0.002% and 3.225%, respectively | Baştürk and Alver (2019) |
ANFIS and hSADE-NN | This research aimed to model and optimize the removal of acidic blue 193 by UV and persulfate processes | hSADE-NN had an MAE (Mean Absolute Error) of 3.61%, and ANFIS had an MAE of 5.18%, indicating good prediction results | Vasseghian and Dragoi (2018) |
A grey-based Taguchi neural network approach | This research aimed to improve the optimum process conditions for acrylic fibre dyeing using a grey-based Taguchi neural network approach | The grey relationship rating was improved from 0.6630 to 0.7749 using the grey-based Taguchi ANN approach, resulting in an improvement of approximately 17% | Zeydan (2014) |
Non-linear multiple regression and a hybrid ANN model based on PSO | This research focused on modelling and evaluating the pollutant sorption process in industrial wastewater using non-linear multiple regression and a hybrid ANN model based on PSO | The results showed that the non-linear regression and hybrid PSO-ANN models could effectively model the pollutant sorption process in industrial wastewater with an R2 of 0.91 | Aryafar et al. (2019) |
ANFIS, ANN, and polynomial regression | This research aimed to predict the dye concentration on AgNPs and AgNP-treated silk fabrics using ANFIS, ANN, and polynomial regression methods | The ANFIS system outperformed other methods, with a prediction error of 0.07% for the system | Nateri et al. (2021) |
SVM, Random Forest | This research investigated the treatment of real textile wastewater using spinning disc technology and aimed to model the experiments | The best model in terms of regression correlation was the SVM model | Zaharia et al. (2021) |
RSM, ANN, and ANFIS | This research investigated the removal efficiency of coagulant treatment of COD and colour from textile wastewater based on various conditions using RSM, ANN, and ANFIS models | The ANFIS model was found to be more suitable than ANN and RSM. The R2 and MSE values for ANFIS were 0.9997 and 0.0003, respectively. The R2 and MSE values for ANN were 0.9955 and 0.0845, respectively, and for RSM were 0.9474 and 1.0494, respectively | Nnaji et al. (2022) |