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

Table 3

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

Adsorbent	Adsorbate	ANN model	Experiment	Data used (Training-Testing-Validation)	Input variables	Output variables	Pollutant removal	Reference
Bone char	Naproxen	Feed forward back propagation algorithm (2-10-1)	Fixed-bed	98-21-21	C0, m, t	Ct/C0	3.2 mg/g	[36]
Bone char	Acid blue 25, acid blue 74, reactive blue 4	Feed forward network back propagation (5-10-3)	Kinetic, equilibrium and fixed-bed	70-15-15	C0, pH, t, ratio M/W, T	Adsorption capacity and Ct/C0	AB25 34.91 mg/g, AB74 32.17 mg/g, RB4 27.98 mg/g	[38]
Montmorillonite–iron oxide composite	Cs^1+ and Sr^2+	Hybrid Freundlich isotherm-ANN Levenberg-Marquardt back propagation (4-25-1)	Fixed-bed	80-0-20	C0, t, Db, F	Ct	~ 20 mg/L	[42]
Date palm biochar	Ortho-cresol and phenol	Feed forward and nonlinear regression generalized decay-function	Fixed-bed	--	C0, F, Db, m, t	Residual concentration	--	[49]
Activated carbon, Amberlite XAD-2	Rhamnolipid	Back propagation (4-6-1)	Fixed-bed	66.7-0-33.3	F, Db, C0, t	Ct/C0	--	[50]
Macroporous resins	Solanesol	Feed forward back propagation with gradient descent	Fixed-bed	331-148-0	C0, F, t, Ratio of Height/Diameter, bed void volume fraction, particle void volume fraction, Ps, BET area	Ct	110 mg/mL	[55]
Bone char	Fluoride	Hybrid model based on Thomas equation and feed forward ANN (3-18-18-1)	Fixed-bed	192-96-96	C0, m, t	Ct/C0	2.52 mg/g	[57]
Pseudomonas aeruginosa immobilized in polyacrylamide gel	La, Eu and Yb	Multi-Layer perceptron (6-3-1)	Fixed-bed	392 data	C0, Db, superficial liquid velocity based on empty column, Reynolds number	Ct/C0	190 μmol/g	[169]
Brown seaweed Ecklonia biomass	Cr^6+	Feed-forward back-propagation (6-5-3-1)	Fixed-bed	127 data	C0, m, pH, Db, F, T	Ct	50.2 mg/g	[170]
Eucalyptus camaldulensis barks	Basic blue 41, reactive black 5	Multi-Layer Perceptron (3-5-1)	Fixed-bed	--	Volume of water, Db, C0	Ct/C0	BB 41 162.2 mg/g, RB5 4.8 mg/g	[172]
Posidonia oceanica (L.) Biomass	Methylene blue	Multi-Layer feed forward (3-15-1)	Fixed-bed	608-304-304	F, Db, t	Ct/C0	482.6 mg/g	[173]
Hydrated ferric oxide-based nanocomposite	Phosphate	Three-layer feed-forward back-propagation (4-(6-20)-1)	Equilibrium and fixed-bed	75-25-0	Batch: pH, C0, T, m,; Dynamic: pH, C0, T, F, Db	Removal efficiency, Ct	99.50%	[174]
Activated carbon	Phenol, 2-chlorophenol, 4-nitrophenol	Multi-layered feed forward back propagation	Fixed-bed	106-20-20	Db, F, C0, T	Ct/C0	--	[175]
Aluminum-doped bone char	Fluoride	Feed forward back propagation (3-2-1)	Fixed-bed	474-237-237	F, C0, t	Ct/C0	18.5 mg/g	[176]
Nanofibers, rice polish, living bacillus, walnut husk, black cumin, oleifera seed, hyacinth root, flax meal and rice straw	As^3+, Pb^2+, Cu^2+, Red G	Feed-forward back propagation perceptron (5-(6-16)-1)	Fixed-bed	215-0-47	Ratio of the contact time to the maximum time, BET area, rotation of packing, ratio of inertia to viscosity, mass ratio of pollutants to packing per unit volume	Normalized adsorption capacity	--	[179]
Pongamia	Zn^2+	Multilayer perceptron Levenberg Marquardt back propagation (3-7-1)	RSM (CCD)	60-20-20	pH, C0, T, m, F, Db	Removal percentage	66.29 mg/g	[180]
Jackfruit, mango and rubber leaves	Cd^2+	Hybrid artificial neural network–genetic algorithm	Fixed-bed	70-20-10	C0, F, Db, t	Removal percentage	98.26%	[181]
Activated carbon from leaves of Calotropis Gigantea	Methylene blue	ANFIS	Fixed-bed	40-20-0	C0, Db, pH, F, T	Removal percentage	--	[182]
Sodium dodecyl sulfate modified graphene	Cu^2+, Mn^2+	Artificial neural network (3-5-1)	Fixed-bed	--	T, pH, m, C0	Removal efficiency	Cu^2+ 48.83 mg/g,; Mn^2+ 45.62 mg/g	[183]
Activated carbon	Phenol, toluene, benzene, caffeine, ciprofloxacin, flumequine, diclofenac	Feed forward back propagation (8-45-1)	Fixed-bed	4761-595-595	Molar weight, C0, F, Db, Ps, surface area, pore diameter, t	Ct/C0	--	[184]
NaOH-modified rice husk	Methylene blue	Feed forward back propagation (3-13-1)	Fixed-bed	50-25-25	F, Db, t	Ct	101.32 mg/g	[209]
Silver nanoparticles	Methylene blue	Hybrid artificial neural network–particle swarm optimization (6-14-1)	Fixed-bed	75-13-12	pH, m, sample flow, sample volume, eluent volume and eluent flow	Removal percentage	99.40%	[211]
Light expanded clay aggregate (LECA)	Aniline	Feed forward back propagation (3-(2-7)-(2-5)-2)	Fixed-bed	298-0-122	t, F, influent chemical oxygen demand	Reactor efficiency, Ct	90%	[292]
Shells of sunflower	Cu^2+	Multi-layered neural network back propagation (7:7-5-1:1)	Fixed-bed	162-81-81	t, C0, m, pH, F, Db, Ps	Ct/C0	25.95 mg/g	[293]
Citric-acid-modified rice straw	Crystal violet	Feed forward back propagation (3-17-1)	Fixed-bed	50-25-25	F, Db, t	Ct	77.18 mg/g	[294]
Zinc oxide nanoparticles–chitosan	Methyl orange	Cuckoo Search–ANN Levenberg–Marquardt (5-11-1)	Fixed-bed	29-0-12	pH, volume of elution solvent, m, F, elution solvent	Removal percentage	481 mg/g	[295]
Alginate-based composite	Ni^2+	Three-layered feed-forward (4-10-2)	Dynamic	22-5-5	C0, m, t, pH	Adsorption capacity and Removal percentage	248.7 mg/g	[296]
Hyacinth root	Pb^2+	Multi-layer perceptron with Levenberg–Marquardt back-propagation (4-(2-20)-1)	Fixed-bed	534 data	F, Db, t	Ct	10.94 mg/g	[297]
Hydrous ferric oxide	Fluoride	Hybrid model based on Thomas equation - ANN	Fixed-bed	--	C0, m, t	Ct/C0	6.71 mg/g	[298]
Shells of sunflower	Co^2+	Multi-layered feed forward back propagation (7-5-1)	Fixed-bed	147-74-74	C0, t, m, pH, F, Db, Ps	Ct/C0	11.68 mg/g	[299]
Silica-molecular imprinting	Cholesterol	Feed forward back-propagation Levenberg-Marquardt	Fixed-bed	50-25-25	Db, t, F	Removal efficiency	67.80%	[300]
Bacillus subtilis beads	Cd^2+	Two layer feed forward back propagation combined with the Thomas and Yan models (6-(2-22)-1)	Fixed-bed	273-59-59	C0, t, Db, m, column internal diameter of column	Ct/C0	104.2 mg/g	[301]
Activated carbon	Dyes	Feed forward back propagation (4-(9-12-8)-1)	Fixed-bed	156-0-52	High gravity factor, liquid Reynolds number, adsorption time to the maximum adsorption time and packing density to liquid concentration	Normalized adsorption capacity	--	[302]
Ignimbrite	Fe^3+	Multi-layered back propagation (6-4-1)	Fixed-bed	293-147-147	C0, t, m, pH, F, Db, Ps	Ct/C0	3.65 mg/g	[303]
Spherical nanoscale zerovalent iron on cellulose	Phosphate	Multilayer perceptron back propagation	Fixed-bed	--	C0, pH, t, T	Removal percentage	561.mg/g	[304]
Olive stone, pinion shell	Cu^2+	ANN -Fuzzy Inference System	Fixed-bed	--	F, C0, Db, t	Ct/C0	Olive stone 5.06 mg/g, pinion shell 7.14 mg/g	[305]

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