Review Article

A Review of the Modeling of Adsorption of Organic and Inorganic Pollutants from Water Using Artificial Neural Networks

Table 4

Summary of the ANN modeling of the multicomponent adsorption of water pollutants at dynamic operating conditions.

AdsorbentAdsorbateANN modelExperimentData used (Training-Testing-Validation)Input variablesOutput variablesPollutants removalReference

BiocharCd2+, Ni2+, Zn2+, Cu2+Feed forward back propagation, feed forward back propagation with distributed time delay, Cascade forward, ElmanFixed-bed994-213-213C0, tCt/C0Cd2+ 5.17 mmol/g, Ni2+ 2.01 mmol/g, Zn2+ 4.38 mmol/g, Cu2+ 5.40 mmol/g[41]
Date palm biocharOrtho-cresol and phenolFeed-forward and nonlinear regression generalized decay-functionFixed-bed column--C0, F, Db, m, tResidual concentration--[49]
Chitosan/zeoliteBisphenol A, carbamazepine, ketoprofen, tonalideThree-layer feed-forward Levenberg Marquardt (2-4-1)Fixed-bed, RSM (CCD)60-20-20C0, pHAdsorption capacityBisphenol A 37.11 mg/g, carbamazepine 41.24 mg/g, ketoprofen 5.92 mg/g, tonalide 3.61 mg/g[77]
Activated carbonPesticidesStatic and recurrent neural network (13-6-1), (8-3-1)Fixed-bed6044-0-3705Solubility, molecular weight, C0, initial total organic carbon, Freundlich parameters, secondary micropore volume, C/Co at time (k-1)TNormalized concentration C/C0 at time kT.160 mg/g[171]
Bone charCu2+, Zn2+Feed-forward network and backpropagation algorithmFixed-bed504-108-108 single, 1008-216-216 binaryMetal properties (molecular weight, electronegativity and ionic radius), C0, F, Db, tCt/C0Cu2+ 51.8 mg/g, Zn2+ 41.1 mg/g[177]
Bone charCd2+, Ni2+, Zn2+Fuzzy ANNFixed-bed70-15-15C0, hydration energy, electronegativity, hydrated ionic radii, molecular weightCt/C0--[178]
Rice branAs3+, As5+Multi-layer feed-forward back-propagation, Levenberg-Marquardt back-propagation (4-7-5-1), (3-5-7-1)Equilibrium and fixed-bed, RSM /CCD)41-0-10Batch: pH, C0, T, m; Dynamic: Db, F, C0Adsorption capacityAs3+ 41.55 μg/g,
A5+ 45.60 μg/g
[306]
Anion base resinResidual color in palm oil mill effluentWavelet ANN (3-6-1)Fixed-bed70 datapH, F, DbResidual color was the experimental response4.63 mg/g[307]

Nomenclature: central composite design (CCD), depth of bed (Db), flow rate (), inlet concentration (), mass of bed (), outlet concentration (Ct), and response surface methodology (RSM).