Академический Документы
Профессиональный Документы
Культура Документы
2, 239-247
http://omicron.ch.tuiasi.ro/EEMJ/
______________________________________________________________________________________________
Abstract
Some samples consisting in silica functionalized in different degrees with dihydroxy-azomethine groups have been tested as
sorbents for copper (II) ions from aqueous solutions in different conditions. An indirect monitoring procedure for the complexing
process was approached: UV-VIS absorption intensity – time variation for the metal-containing solution in which complexing
silica was soaked. In order to find the conditions in which the maximum efficiency in the metal absorption can be obtained
(maximum decreasing of the UV absorbance), a method based on neural networks and genetic algorithms (NN-GA) was applied.
The model included into the optimization procedure was a feedforward neural network with one hidden layer. The optimization
problem was solved with a standard genetic algorithm, which computed the optimal values for reaction conditions (CuCl 2 solution
concentration, pH, complexing groups density in silica) leading to minimum relative absorbance of the metal in shortest time.
Key words: complexed metals, functionalized silica, genetic algorithm optimization, neural network modeling
Author to whom all correspondence should be addressed: e-mail: silvia_curteanu@yahoo.com
Piuleac et al. /Environmental Engineering and Management Journal 9(2010), 2, 239-247
systems immobilized on silica gel are available. In a previous paper (Piuleac et al., 2008)
Between them, Schiff bases, having multidentate details of preparing silica xerogels with different
coordination sites, are known to form complexes with contents in chelating groups and its use to extract
transition metal ions readily (Che and Cheng, 1986; metal ions (Cu2+) from aqueous solution were
Ueno and Martell, 1955). presented. In order to find the conditions in which a
Silica xerogels functionalized in different maximum efficiency in the metal absorption can be
degrees with hydroxyazomethyne groups linked reached and to study the evolution in time of the
through a propylene moiety to silica network were absorbance as a function of others process parameters
tested in this paper for the uptake of copper (II) from (saturation time, concentration of the solution,
aqueous solutions. An indirect monitoring procedure chelating groups content and pH), a series of
for the complexing process was approached: UV-VIS experiments have been performed in different
absorption intensity – time variation for the metal- conditions. The acquired data were used for process
containing solution in which complexing silica was modeling based on artificial intelligence instruments.
soaked. It was emphasized the option of using simple neural
For complex chemical processes, neural networks or stacked neural networks with different
networks (NN), which posses the ability to recognize architectures to model the wastewater treatment
and reproduce cause–effect relationships, has gained process based on silica xerogels functionalized with
popularity in various areas of environmental hydroxyl-azomethine groups.
approaching (Karatzas and Kaltsatos, 2007; The main goal of the present paper is to
Manolakos et al., 2007; Mjalli et al., 2007). Neural develop a general procedure based on neural networks
networks are efficient modelling tool for complex and genetic algorithms, which could be applied to the
processes where the chemical and physical laws are complex optimization problems. Neural networks are
not completely known because they are black boxes used as efficient modeling tools and genetic
based on input – output data. algorithms as solving methods of optimization; a
Genetic algorthms (GA) are part of the so- neural model of the process is included in the
called evolutionary algorithms and represent an optimization procedure. In this herein case study, the
optimization instrument with increasing application in estimation of the metal absorbance is analyzed as a
scientific problems. These kinds of approaching do function of the process parameters: time, CuCl2
not require any piece of information about the search solution concentration, values of complexing groups
space area, just being necessary an objective/fitness density in silica and pH values. There is followed a
function that assigns a value to any solution (Deb, maximum value of the absorbance decreasing in the
1999). context of using minimum resources. It was
The basic algorithm, simple GA (SGA), has demonstrated that GA optimization procedure is easy
the advantages that it does not require good initial to apply and provides accurate results.
guesses for the values of the decision variables, like
traditional methods. It uses a population of several 2. Experimental
points simultaneously along with probabilistic
operators (reproduction, crossover and mutation), 2.1. Materials
inspired by natural genetics. Additionally, SGA has
the advantage of using only the values of the Some solutions differing by copper dichloride
objective functions and not any derivatives, as concentration (5.5 and 11 g /L) and pH (4, 5 and 6)
required by gradient search techniques (Guria et al., were prepared by using adequate buffer solutions.
2005). Silica functionalized with
The applicability of these algorithms has dihydroxyazomethine groups (Fig. 1) in different
covered a wide variety of problems because of their degrees (Table 1) was prepared, as already described
flexibility, ease of operation, minimal requirements (Cazacu et al., 2008), by one-pot procedure, both
and global perspective that they offer (Deb, 2001). silica formation and its functionalization occurring in
GAs have found various applications in chemical the same sol-gel system.
engineering including process control, gas pipeline
design, pattern recognition of multivariate chemical
data, optimization of reaction rate parameters,
multipurpose chemical batch plant design, and
O
scheduling (Yan et al., 2003). Details about different
types of genetic algorithms and a series of their
O - Si N = HC - OH
applications in optimization of the chemical processes
have been given in the reviews of Bhaskar et al.
(2000) and Nandasana et al. (2003). O
HO
Our group applied GA optimization method
for different chemical systems (Caliman et al., 2008;
Curteanu et al., 2006; Curteanu and Leon, 2007;
Curteanu and Cazacu, 2007). Fig.1. The silica functionalized with dihydroxyazomethine
groups
240
Optimization by NN – GA technique of the metal complexing process - potential application in wastewater treatment
Table 1. The functional groups content in the prepared to a multilayer perceptron having one hidden layer
silica with 6 neurons, MLP(4:6:1). Before using, the neural
model was tested in training and validation phases.
Sample Functional groups content in silica (mol/g) Good agreement between experimental data and
FS1 4.35 x 10-4 neural network predictions (average relative errors
FS2 8.70 x 10-4 less than 0.3298 % and correlation over 0.9939)
FS3 13.04 x 10-4 registered against “unseen” data (validation phase).
Maximum absorbance at every time is noted
At and reported to the absorbance of the initial
2.2. Metal uptake experiments solution (A0). For a better comparability of the
results, eliminating the influence of the ratio between
UV-VIS spectrum was registered for each sorbent and solution amounts, the relative absorbance,
CuCl2 solution. A certain amount of functional silica Ar=At/A0, can be used.
gel bringing a certain amount of chelating groups was In this study, the optimization problem
introduced in such a solution and UV-VIS spectrum includes the neural model, which is represented as:
was registered (by using SPECORD M42
spectrophotometer with quartz cells of 1 cm NN [Inputs: Time, CuCl2 solution concentration,
thickness) at certain time intervals until the maximum complexing groups density in silica, pH; Output:
absorbance remains constant. The corresponding Absorbance] (1)
time, saturation time, was denoted as ts, when either
sorption equilibrium was reached or all functional The vector of decision variables, u, has the
groups were occupied by metal complexing (Cazacu components:
et al., 2008).
u = [Time, CuCl2 solution concentration, complexing
2.3. Formulation of the optimization problem groups density in silica, pH] (2)
241
Piuleac et al. /Environmental Engineering and Management Journal 9(2010), 2, 239-247
not required and such points do not influence the approaches for genetic algorithm based optimization
final solution (Leboreiro and Acevedo, 2004). which use binary solution representation, as it is the
The optimization procedure includes the simplest type of encoding, in which chromosomes are
neural model MLP(4:6:1) and it is solved with a composed only of 1’s and 0’s. Even the number of
simple genetic algorithm. Genetic algorithm alleles is thus rather small (two), this encoding is very
provides, after an iterative calculus, the optimal common, because it is very easy to use. However,
values for decision variables (time, CuCl2 solution value encoding is more general, because genes are
concentration, complexing groups density in silica, real numbers. Some experiments (Michalewicz, 1996)
pH), which are the inputs for the neural network have shown that real value encoding is more time
model. With these inputs, the neural network efficient, with better precision of the solutions.
computes the final value of the relative absorbance. If The initial population is generated randomly.
the two values are identical or there is a very tight Offspring is created by genetic operators and it is
difference between them, it can be conclude that the stored in a population pool that is a collection of
task of the optimization, represented by minimum of offspring and their parents.
the objective function, J, is achieved. Fig. 2 Selection compares the chromosomes in the
illustrates this optimization procedure. population aiming to choose those which will take
part in the reproduction process. The selection occurs
with a given probability on the base of fitness
functions. The fitness function plays a role of the
environment to distinguish between good and bad
solutions. The individuals with higher fitness must
have more chances to reproduce. For each individual
to be created in the next generation, two parents are
thus selected.
There are different methods for selection
phase; our paper uses rank selection which first ranks
the population and then every chromosome receives
fitness from this ranking. The worst will have fitness
1, second worst 2 etc. and the best will have fitness n
(number of chromosomes in the population).
The recombination (crossover) has as main
purpose the recombination the features of two
Fig. 2. The optimization method based on NN and GA randomly selected parents from the mating pool with
the aim of producing better offspring. The variant of
2.4. GA optimization technique crossover used in this study supposes different points
for all genes, which means that the new individual
Genetic algorithms are intelligent stochastic will no longer be on the line segment that links its
optimization techniques based on the mechanism of parents. The offspring will look more like one parent
natural selection and genetics. GAs start with an regarding a feature and less regarding another.
initial set of solutions, called population. Each After recombination, offspring undergoes to
solution in the population is called chromosome (or mutation. Generally, the mutation refers to the
individual), which represents a point in the search creation of a new chromosome from one and only
space. The chromosomes are evolved through one individual with predefined probability. Mutation
successive iterations, called generations, by genetic is used to produce small perturbations on
operators (selection, crossover and mutation) that chromosomes to promote diversity of the population.
mimic the principle of natural evolution. A set of Our GA includes a variant of mutation named
solutions are analyzed and modified by genetic resetting.
operations simultaneously, where selection operator A gene value is reset to a random value in its
can select some “good” solutions as seeds, crossover search interval. The purpose is to refresh the search
operator can generate new solutions hopefully process, in case when the genetic diversity of the
retaining good features from parents, and mutation population decreases (so no longer converges to the
operator can enhance diversity and provide a chance solution) or the algorithm has converged into a local
to escape from local optima (Wang et al., 2006). optimum. Each gene is independently considered,
In a GA, a fitness value is assigned to each and mutation gives it a new random value in the
individual according to a problem-specific objective initialization interval. Only some genes change
function. Generation by generation, the new (possibly all, but unlikely).
individuals, called offspring, are created and survive After the three operators are carried, the
with chromosome in the current population, called offspring is inserted into the population, replacing
parents, to form a new population. the parent chromosomes in which they were derived
In the GA model, They were used real values from, producing a new generation. The best
encoding for the chromosomes. There are other individual is copied directly into the new population
(the elitism technique) and the rest of the individuals
242
Optimization by NN – GA technique of the metal complexing process - potential application in wastewater treatment
Relative absorbance
0.98
that the overall solution of the GA will not get worse. G1_net
The stopping criterion determines when GA 0.96
G2_exp
G2_net
will break. In other words, the genetic operations are
G3_exp
repeated until a termination condition is finding. In 0.94
G3_net
this implementation GA was stopped if the maximum 0.92
G4_exp
number of generations has been executed or the pre- G4_net
0.96 G5_net
G6_exp
0.94 G6_net
0.92 G7_exp
O
G7_net
0.90 G8_exp
O - Si N = HC - OH
0.88 G8_net
O
O Cu O 0.86
O
0.84
HO - CH = N Si - O 0 200 400 600 800 1000 1200 1400 1600
Time, min
O
243
Piuleac et al. /Environmental Engineering and Management Journal 9(2010), 2, 239-247
algorithms. The values of these parameters must be The optimization procedure is implemented in
specified before the execution of GA and they depend Matlab 7.5 with original software, as specific
on the nature of the objective function. One important functions were programmed for each phase of the
thing in using GA as solving method is to adjust its genetic algorithm.
parameters to the particular problem approached, so Since GA is a stochastic algorithm, it was run
as to obtain good solutions and to preserve the for many times, for each situation (each row in the
algorithm from a preliminary convergence. tables of results) to get statistically meaningful values
The fitness function in GA is the objective for the time taken. Longer runs are not necessary as
function of the optimization procedure. The results of GA is already a population based procedure that
the optimization are represented by the values of the works with multiple solutions. One of the solutions
decision variables (time, CuCl2 solution (the best solution) is chosen and inserted into the
concentration, complexing groups density in silica, results Tables.
pH) that lead to a minimum value of the objective Table 2 shows the influence of initial
function, which means the achievement of minimum population. There is known that the grower parameter
value for the relative absorbance (maximum decrease is the better results of optimization are obtained, but
of the absorbance). this kind of rule is applied until one certain limit.
Time = 2199.3954
Conc = 5.5111
2 100 100 Gr_dens = 0.0010 0.9427
pH = 5.5008
Time = 2173.9741
Conc = 5.5024 Gr_dens =
3 150 100 0.0010 0.9428
pH = 5.5003
Time = 2194.5754
Conc = 5.5082 Gr_dens =
4 200 100 0.0010 0.9429
pH = 5.5018
Time = 2190.0300
Conc = 5.5062
5 300 100 Gr_dens = 0.0010 0.9428
pH = 5.5005
Time = 2187.1616
Conc = 5.5092 Gr_dens =
6 400 100 0.0010 0.9429
pH = 5.5015
Time = 2190.0520
Conc = 5.5039 Gr_dens =
7 500 100 0.0010 0.9429
pH = 5.5032
Time = 2198.7309
Conc = 5.5191
8 600 100 Gr_dens = 0.0010 0.9429
pH = 5.5024
Time = 2186.6470
Conc = 5.5005 Gr_dens =
9 800 100 0.0010 0.9429
pH = 5.5023
Time = 2193.6061
Conc = 5.5072
10 1000 100 Gr_dens = 0.0010 0.9427
pH = 5.5006
244
Optimization by NN – GA technique of the metal complexing process - potential application in wastewater treatment
Starting with a dimension of initial population population = 50, number of generations = 50,
of 50, the results obtained are good, although at high crossover rate = 0.9 and mutation rate = 0.03.
values there are obtaining better results. The Running GA optimization with these
improving is becoming insignificant, so the growth of parameters, the following results were obtained: time
the execution time doesn’t justify. For this case, the = 2100 min, CuCl2 concentration = 5.85 g·L-1, group
relative absorbance is around 0.94, the values of time density = 0.0010 mol·g-1 and pH = 5.5, with a
are at superior limit of the considered interval maximum value of relative absorbance of 0.945.
(experimental time interval), the CuCl2 concentration Generally, the optimization results are not
about 5.5, pH = 5.5 and the group density around significantly influenced by the values of GA
1·10-3 (relatively a high value). algorithm. Taking into account the elitism included in
In Table 2, Conc refers to CuCl2 solution the optimization procedure, this technique does not
concentration (g∙L-1) and Gr_dens represents allow the solution becomes worse.
complexing groups density in silica (mol∙g-1). On the other hand, there is compensation
A similar problem is about the influence of the between the parameters values; for instance, if the
number of generations (Table 3). Taken into account crossover or mutation rates are unsuitable, high
a range between 50 and 1000 generations with a values for initial population dimension and number of
dimension of population established at 50, considered generations are necessary to reach optimal solution.
as an optimal value, the results seem to be similar in It could be better to obtain a maximum
all the cases. The influence of crossover and mutation decrease of the absorbance in the shorter time. That
rates was also studied, by varying the crossover rate why several runs were performed narrowing the time
between 0.2 and 2 and the mutation rate between interval (Table 4). One can see that the value of
0.001 and 1. The best values for these two relative absorbance increases as the time interval
parameters were 0.9 and 0.03, respectively. The becomes narrower. The dependence between the
decreasing of the cross rate could lead at worse values of the relative absorbance and the optimal
results. Consequently, the best GA parameters for the value of time is given in Fig. 7.
approached case study were: dimension of initial
1 50 50 Time = 2184.2895
Cross Rate=0.8 Conc = 5.5595
Mute Rate=0.04 Gr_dens = 0.0010
pH = 5.5214
2 50 100 Time = 2147.1225
Conc = 5.5066
Gr_dens = 0.0010
pH = 5.5096
3 50 150 Time = 2193.0200
Conc = 5.5207
Gr_dens = 0.0010
pH = 5.5044
4 50 200 Time = 2198.6052
Conc = 5.5064
Gr_dens = 0.0010
pH = 5.5076
5 50 300 Time = 2190.5278
Conc = 5.5027
Gr_dens = 0.0010
pH = 5.5010
6 50 400 Time = 2194.4349
Conc = 5.5051
Gr_dens = 0.0010
pH = 5.5002
7 50 600 Time = 2198.6412
Conc = 5.5003
Gr_dens = 0.0010
pH = 5.5003
8 50 800 Time = 2199.9137
Conc = 5.5005
Gr_dens = 0.0010
pH = 5.5015
9 50 1000 Time = 2196.8627
Conc = 5.5018
Gr_dens = 0.0010
pH = 5.5002
245
Piuleac et al. /Environmental Engineering and Management Journal 9(2010), 2, 239-247
One can see that a minimum relative algorithms and neural networks, applied to a
absorbance is associated with maximum time. wastewater treatment by implying silica xerogel
Consequently, a compromise between the limit compounds.
situations presented in Fig. 5 supposes the values of The genetic algorithm solves the optimization
relative absorbance of 0.950 0.955 with time around problem and the neural network constitutes the model
250 minutes. included in the optimization procedure. Simple
architecture of the neural network is proposed for
0.99 process modeling: a simple feed forward neural
0.985 network with one hidden layer. The simple genetic
0.98 algorithm proves to be a good tool for solving the
Relative absorbance
4. Conclusions References
This paper provides a general and simple Bhaskar V., Gupta S.K., Ray A.K., (2000), Applications of
optimization methodology, based on genetic multiobjective optimization in chemical engineering,
Reviews in Chemical Engineering, 16, 1-54.
246
Optimization by NN – GA technique of the metal complexing process - potential application in wastewater treatment
Blain S., Treguer P., (1995), Iron (II) and iron(III) Thessaloniki, Greece Simulation Modelling Practice
determination in sea water at the nanomolar level with and Theory, 15, 1310-1319.
selective on-line preconcentration and Kocjan R., Prezeszlakowski S., (1992), Calcon-modified
spectrophotometric determination, Analytica Chimica silica gel sorbent. Application to preconcentration or
Acta, 308, 425-432. elimination of trace metals, Talanta, 39, 63-68.
Bowe C.A., Martin D.F., (2004), Extraction of heavy metals Leboreiro J., Acevedo J., (2004), Processes synthesis and
by 2-mercaptoethoxy groups attached to silica gel, design of distillation sequence using modular
Journal of environmental science and health. Part A, simulators: a genetic algorithm framework, Computers
Toxic/hazardous Substances and Environmental and Chemical Engineering, 28, 1223-1236.
Engineering, 39, 1479-148. Liu C.-Y., Cheng L., (1993), Chelating Polymers. In:
Cazacu M., Vlad A., Airinei A., Alexandru M., (2008), Facets of Coordination Chemistry, Agarwal B.V.,
Silica Encapsulating Lanthanum Complexes by Sol- Munshi K.N., Dey A.K. (Eds.), World Scientific.
Gel Technique, Polymers International, in press. Mahmoud M.E., Osman M.M., Amer M.E., (2000),
Che C.M., Cheng W.K., (1986), Manganese(III) amide Selective pre-concentration and solid phase extraction
complexes as a new class of catalyst for efficient of mercury(II) from natural water by silica gel-loaded
alkene epoxidation, Journal of Chemical Society, dithizone phases, Analytica Chimica Acta, 415, 33-40.
Chemical Communications, 1443-1444. Mahmoud M.E., Gohar G.A., (2000), Silica gel-
Curteanu S., Leon F., Galea D., (2006), Alternatives for immobilized-dithioacetal derivatives as potential solid
Multiobjectives Optimization of a Polymerization phase extractors for mercury(II), Talanta, 51, 77-87.
Process, Journal of Applied Polymer Science, 100, Manolakos E., Virani H., Novotny V., (2007), Extracting
3680-3695. knowledge on the links between the water body
Curteanu S., Cazacu M., (2007), Neural networks and stressors and biotic integrity, Water Research, 41,
genetic algorithms used for modeling and optimization 4041-4050.
of the siloxane-siloxane copolymers synthesis, Journal Michalewicz Z., (1996), Genetic Algorithms + Data
of Macromolecular Science, Part A Pure and Applied Structures = Evolution Programs, Springer Verlag,
Chemistry, A45, 23-36. Berlin.
Curteanu S., Leon F., (2007), Optimization strategy based Mjalli F.S., Al-Asheh S., Alfadala H.E., (2007), Use of
on genetic algorithms and neural networks applied to a artificial neural network black-box modeling for the
polymerization process, International Journal of prediction of wastewater treatment plants performance
Quantum Chemistry, 108, 617-630. Journal of Environmental Management, 83, 329-338.
Caliman F., Curteanu S., Betianu C., Gavrilescu M., Nandasana A.D., Ray A.K., Gupta S.K., (2003),
Poulios I., (2008), Neural networks and genetic Applications of the Non-Dominated Sorting Genetic
algorithms optimization of the photocatalytic Algorithm (NSGA) in chemical reaction engineering,
degradation of Alcian Blue 8GX, Journal of Advanced International Journal of Chemical Reactor
Oxidation Technologies (ACS), 11, 316-326. Engineering, 1, 1-15.
Deb K., (1999), An introduction to Genetic Algorithms, Piuleac C.G., Curteanu S., Cazacu M., (2008), Stacked
Sadhana-Academy Proceedings in Engineering neural network modeling applied to the process of
Sciences, 24, 293-315. metal ion sorption by the functional silica xerogel,
Deb K., (2001) Multiobjective Optimization Using Polymer-Plastics Technology and Engineering, in
Evolutionary Algorithms, Chichester, Wiley. press.
Dey R. K., Jha U., Singh A.C., Samal S., Ray A.R., (2006), Prado A.G.S., Airoldi C., (2001), Adsorption,
Extraction of metal ions using chemically modified preconcentration and separation of cations on silica gel
silica gel covalently bonded with 4,4′- chemically modified with the herbicide 2,4-
diaminodiphenylether and 4,4`- dichlorophenoxyacetic acid, Analytica Chimica Acta,
diaminodiphenylsulfone-salicylaldehyde Schiff Bases, 432, 201-211.
Analytical Sciences, 22, 1105-1110. Seshadri T., Haupt H.J., (1988), Silica-immobilized 2-[(2-
Dudler V., Lindoy L.F., Sallin D., Schlaepfer C.W., (1987), (triethoxysily)ethyl)thio]aniline as a selective sorbent
An oxygen-nitrogen donor macrocycle immobilized on for the separation and preconcentration of palladium,
silica gel. a reagent showing high selectivity for Analytical Chemistry, 60, 47-52.
Copper(II) in the presence of Cobalt(II), Nickel(II), Singh D.K. ,. Mishra N.K, (1991), Zinc silicate-bonded
Zinc(II) or Cadmium(II), Australian Journal of diethyldithiocarbamate - a selective adsorbent for
Chemistry, 40, 1557 – 1563. separating transition metal ions and preconcentration
Fernandes, F.A.N., Lona, L.M.F., (2005), Neural network of palladium(II), Chromatographia, 31, 300-302.
applications in polymerization processes, Brazilian Ueno K., Martell A.E., (1955), Infrared study of metal
Journal of Chemical Engineering, 22, 323-330. chelates of bisacetylacetoneethylenediimine and
Guria C., Bhattacharza P.K., Gupta S.K., (2005), Multi- related compounds, Journal of Physical Chemistry, 59,
objective optimization of reverse osmosis desalination 998-1004.
units using different adaptations of the non-dominated Wang L., Zhang L., Zheng D.Z., (2006), An effective
sorting genetic algorithm (NSGA), Computers and hybrid genetic algorithm for flow shop scheduling with
Chemical Engineering, 29, 1977-1995. limited buffers, Computers and Operations Research,
Jal P.K., Patel S., Mishra B.K., (2004), Chemical 33, 2969-2971.
modification of silica surface by immobilization of Yan X.F., Chen D.Z., Hu, S.X., (2003), Chaos-genetic
functional groups for extractive concentration of metal algorithms for optimizing the operating conditions
ions, Talanta, 62, 1005-1028. based on RBF-PLS model, Computers and Chemical
Karatzas K.D, Kaltsatos S., (2007), Air pollution modelling Engineering, 27, 1393-1404.
with the aid of computational intelligence methods in
247