Application of artificial neural networks in multifactor optimization of an FIA system for the determination of aluminium

A methodology based on the coupling of experimental design and artificial neural networks (ANNs) is proposed in the optimization of a new flow injection system for the spectrophotometric determination of Al(III) with Arsenazo DBM, which has for the first time been used as chromogenic reagent in the quantitative analysis of aluminium. An orthogonal design is utilized to design the experimental protocol, in which three variables are varied simultaneously. Feedforward-type neural networks with faster back propagation (BP) algorithm are applied to model the system, and then optimization of the experimental conditions is carried out in the neural network with 3-7-1 structure, which have been confirmed to be able to provide the maximum performance. In contrast to traditional methods, the use of this methodology has advantages in terms of a reduction in analysis time and an improvement in the ability of optimization. The method has been applied to the determination of Al(III) in steel samples and provided satisfactory results. Received: 26 May 1999 / Revised: 29 July 1999 / Accepted: 17 August 1999     

Application of artificial neural networks in multifactor optimization of an on-line microwave FIA system for catalytic kinetic determination of ruthenium (III)

A methodology based on the coupling of experimental design and artificial neural networks (ANNs) is proposed in the optimization of a flow injection system for the spectrophotometric determination of Ru (III) with m-acetylchlorophosphonazo (CPA-mA), which has been for the first time used for the optimization of high-performance capillary zone electrophoresis (J. Chromatogr. A 793 (1998) 317). And since it has been applied in many other regions like micellar electrokinetic chromatography, ion-interaction chromatography, HPLC, etc. (J. Chromatogr. A 850 (1999) 345; J. Chromatogr. A 799 (1998) 35; J. Chromatogr. A 799 (1998) 47). An orthogonal design is utilized to design the experimental protocol, in which five variables are varied simultaneously (Anal. Chim. Acta 360 (1998) 227). Feedforward-type neural networks with extended delta-bar-delta (EDBD) algorithm are applied to model the system, and the optimization of the experimental conditions is carried out in the neural network with 5-5-1 structure, which have been confirmed to be able to provide the maximum performance. In contrast to traditional methods, the use of this methodology has advantages in terms of a reduction in analysis time and an improvement in the ability of optimization. Under the optimum experimental conditions, Ru (III) can be determined in the range 0.040-0.60 mug ml(-1) with detection limit of 0.03 mug ml(-1) and the sampling frequency of 34 h(-1). The method has been applied to the determination of Ru (III) in refined ore as well as in secondary alloy and provided satisfactory results.     

Application of artificial neural networks in multivariable optimization of an on-line microwave FIA system for catalytic kinetic determination of iridium(III)

A new rapid, selective and sensitive on-line microwave flow injection-kinetic method was developed for spectrophotometric determination of micro amounts of Ir(III), based on its catalytic effect on the m-acetylchlorophosphonazo (CPA-mA) and KIO(4) reaction in NaOH media. An on-line microwave oven was employed to accelerate the reaction. The reaction was followed spectrophotometrically by measuring the decrease of the absorbance of CPA-mA at 580 nm. The effect of five variables for the determination of Ir(III) was optimized by means of a multilayer artificial neural network using extended delta-bar-delta (EDBD) algorithms. Under the optimum experimental conditions, Ir(III) could be determined in the range 0.060-0.60 micro gZZZ;mL(-1) with detection limit of 0.02 micro gZZZ;mL(-1) and the sampling frequency of 34 h(-1). The proposed method was applied to the determination of micro amounts of Ir(III) in refined ore and secondary alloy with the recoveries from 91.4% to 109%.     

Application of generalized artificial neural networks coupled with an orthogonal design to optimization of a system for the kinetic spectrophotometric determination of Hg(II)

A simple and sensitive kinetic method for the determination of traces of mercury (70-760 ng ml⁻¹) based on its inhibitory effect on the addition reaction between methyl green and sulfite ion is proposed. The reaction was monitored spectrophotometrically by measuring the decrease in absorbance of methyl green at 596 nm between 2 and 4 min using a fixed time method. Artificial neural networks with back propagation algorithm coupled with an orthogonal array design were applied to the modeling of the proposed kinetic system and optimization of experimental conditions. An orthogonal design was utilized to design the experimental protocol, in which pH, concentration of sulfite, temperature, and concentration of methyl green were varied simultaneously. Optimum experimental conditions in term of sensitivity were generated by using ANNs. The rate of decrease in absorbance is inversely proportional to the concentration of Hg(II) over entire concentration range tested (100-550 ng ml⁻¹) with a detection limit of 45 ng ml⁻¹ and a relative standard deviation at 200-400 ng ml⁻¹ Hg(II) of 3.2% (n=5). A simple preconcentration step improved the limit of detection and linear dynamic range of the method to about 8 and 12-760 ng ml⁻¹, respectively, by about 10 times enrichment of mercury between 12 and 75 ng ml⁻¹. The method was based on enrichment of Hg(II) from dilute samples on an anionic ion exchanger fixed on a plastic strip and was applied to the determination of Hg(II) in environmental samples with satisfactory results.     

Application of a modified simplex method to the multivariable optimization of a new FIA system for the determination of osmium

A methodology based on the coupling of experimental design and a modified simplex method is proposed for the optimization of a new flow injection-kinetic system for the spectrophotometric determination of Os (IV) with m-acetylchlorophosphonazo, which has for the first time been used as chromogenic reagent in the quantitative analysis of this element. An orthogonal array design is utilized to design the experimental protocol, in which six variables are varied simultaneously, and obtain the initial simplex using 25 experiments. A modified simplex method is applied to continuously optimize the data of the orthogonal array design; the search for optimum conditions of ¶6 variables using the modified simplex method required only 25 experiments. The efficiency and simplicity of the coupling of the experimental design and the modified simplex method are attractive for the development of new analytical methods. The method has been applied to the determination of Os (IV) in a refined ore as well as in a secondary alloy and provided satisfactory results.     

Application of a modified simplex method to the multivariable optimization of a new FIA system for the determination of osmium

A methodology based on the coupling of experimental design and a modified simplex method is proposed for the optimization of a new flow injection-kinetic system for the spectrophotometric determination of Os (IV) with m-acetylchlorophosphonazo, which has for the first time been used as chromogenic reagent in the quantitative analysis of this element. An orthogonal array design is utilized to design the experimental protocol, in which six variables are varied simultaneously, and obtain the initial simplex using 25 experiments. A modified simplex method is applied to continuously optimize the data of the orthogonal array design; the search for optimum conditions of 6 variables using the modified simplex method required only 25 experiments. The efficiency and simplicity of the coupling of the experimental design and the modified simplex method are attractive for the development of new analytical methods. The method has been applied to the determination of Os (IV) in a refined ore as well as in a secondary alloy and provided satisfactory results.     

Application of artificial neural networks to the determination of pesticides by linear sweep stripping voltammetry

In this work, artificial neural network (ANN), a powerful chemometrics approach for linear and nonlinear calibration models, was applied to detect three pesticides in mixtures by linear sweep stripping voltammetry (LSSV) despite their overlapped voltammograms. Electrochemical parameters for the voltammetry, such as scan rate, deposit time and deposit potential, were evaluated and optimized from the signal response data using ANN model by minimizing the relative prediction error (RPE). The proposed method was successfully applied to the detection of pesticides in synthetic samples and several commercial fruit samples.     

Application of factorial design to improve a FIA system for penicillin determination

A potentiometric FIA system for penicillin determination, employing penicillinase [E.C. 3.5.2.6] immobilized on silica gel, packed into a reactor, was improved by the use of statistically designed experiments. A two-level and three-factor factorial was used to find the best working conditions evaluating the influence of some parameters on the signal response of the system and the number of determinations per hour. These parameters were analyzed individually obtaining two level of the variables to be used in the factorial design: length of the reactor (1.5 and 2.0 cm), carrier flow rate (1.6 and 2.2 ml min⁻¹) and sample volume (100 and 150 μl). The pure error on the measurements was estimated by authentic repetitions. The ideal working conditions taking into account a compromise between the best response signal and the number of determinations per hour (with the same importance) being chosen the level of factors: length of reactor 1.5 cm, carrier flow rate 2.2 ml min⁻¹ and sample volume of 150 μl. Under these conditions the system allowed to analyze was about 45 samples per hour, during 73 days, with a standard deviation of 2.4% at concentration range between 10⁻¹ and 10⁻³ mol l⁻¹.     

Optimization of a high-performance liquid chromatography system by artificial neural networks for separation and determination of antioxidants

A high-performance liquid chromatography (HPLC) system was used to determine the antioxidants tert-butyl-hydroquinone (TBHQ), tert-butylhydroxyanisole (BHA), and 3,5-di-tert-butylhydroxytoluene (BHT) simultaneously in oils. The paper presents a new methodology for the optimized separation of antioxidants in oils based on the coupling of experimental design and artificial neural networks. The orthogonal design and the artificial neural networks with extended delta-bar-delta (EDBD) learning algorithm were employed to design the experiments and optimize the variables. The response function (Rf) used was a weighted linear combination of two variables related to separation efficiency and retention time, according to which the optimized conditions were obtained. The above-mentioned antioxidants in rapeseed oils were separated and determined simultaneously under optimized conditions by HPLC with UV detection at 280 nm. Linearity was obtained over the range of 10-200 microg/mL with recoveries of 98.3% (TBHQ), 98.1% (BHT), and 96.2% (BHA).     

The study for optimization of chromatographic condition by means of artificial neural networks

In this paper, the optimization of gas chromatographic experimental parameters is investigated using a three layer feed-forward neural network with the back-propagating. The design, development, and testing of the neural network are described in detail. The chosen structure is 4-6-2 system with a learning rate eta of 0.6 and a momentum constant mu of 0.4. The results of several simulations are very satisfactory. Network results are compared with the results obtained by the orthogonal method.     

Citrate and calcium determination in flavored vodkas using artificial neural networks

The development of multianalyte sensing schemes by combining indicator-displacement assays with artificial neural network analysis (ANN) for the evaluation of calcium and citrate concentrations in flavored vodkas is presented. This work follows a previous report where an array-less approach was used for the analysis of unknown solutions containing the structurally similar analytes, tartrate and malate. Herein, a two component sensor suite consisting of a synthetic host and the commercially available complexometric dye, xylenol orange, was created. Differential UV-Visible spectral responses result for solutions containing various concentrations of calcium and citrate. The quantitation of the relative calcium and citrate concentrations in unknown mixtures of flavored vodka samples was determined through ANN analysis. The calcium and citrate concentrations in the flavored vodka samples provided by the sensor suite and the ANN methodology described here are compared to values reported by NMR of the same flavored vodkas. We expect that this multianalyte sensing scheme may have potential applications for the analysis of other complex fluids.     

Multisensor system on the basis of an array of non-specific chemical sensors and artificial neural networks for determination of inorganic pollutants in a model groundwater

The application of a multisensor system to groundwater monitoring is investigated. The sensor system is based on an array of non-specific potentiometric chemical sensors with data processing by artificial neural networks and includes 13 sensors with PVC membranes and 12 solid-state ones. Results of measurements in model solutions containing heavy metals, alkali- and alkali-earth cations and inorganic anions at concentrations typical for groundwater near the city of Braunschweig, Germany, are presented. Both the response of the whole sensor array and the responses of subsystems consisting only of PVC and only of solid-state sensors, respectively, are investigated. It is shown that both subsystems can be used for determination of described ions and that the best results are obtained if the whole array of sensors is used.     

Application of artificial neural networks to the nondestructive determination of ciprofloxacin hydrochloride in powder by short-wavelength NIR spectroscopy

The present study is aimed at providing a new short-wavelength near-infrared (NIR) spectroscopic method for the nondestructive quantitative analysis of ciprofloxacin hydrochloride in powder via artificial neural networks (ANNs). For this purpose, the NIR spectra of 90 experimental powder samples in the range 700–1100 mm were analyzed. Four different pretreatment methods—first-derivative, second-derivative, standard normal variate (SNV), and multiplicative scatter correction (MSC)—were applied to three sets of the NIR spectra of the powder samples. Among all of the ANN models, the first-derivative model is found to be the best. The results presented here demonstrate that the short-wavelength NIR region is promising for the fast and reliable determination of the major components in pharmaceuticals. The degree of approximation as an evaluation criterion prevents the overfitting phenomenon occurring in ANNs. The text was submitted by the authors in English.     

Application of artificial neural networks to predictions in flow-injection spectrophotometry

It is demonstrated that predictions can be obtained in spectrophotometric flow-injection analysis (FIA) based on an experimental parameter, that is, the degree of reaction, which takes into account the hydrodynamic and chemical characteristics of the spectrophotometric reaction used. The search algorithm is based on constructing a model of a chemical-analytical process using a learning artificial neural network that enables the prediction of the degree of reaction for some reagents not studied yet. The trained neural network is used for the a priori evaluation and comparison of a number of reagents for the determination of aluminum by FIA.     

Rapid determination of radon progeny concentration based on artificial neural networks

Journal of Radioanalytical and Nuclear Chemistry - The rapid measurement of radon progeny concentration is of great significance for improving the efficiency of radon exposure dose evaluation in a...     

Simultaneous kinetic determination of sulfite and sulfide using artificial neural networks

Artificial neural networks (ANNs) are proposed for the determination of sulfite and sulfide simultaneously. The method is based on the reaction between Brilliant Green (BG) as a colored reagent and sulfite and/or sulfide in buffered solution (pH 7.0) and monitoring the changes of absorbance at maximum wavelength of 628 nm. Experimental conditions such as pH, reagents concentrations, and temperature were optimized and training the network was performed using principal components (PCs) of the original data. The network architecture (number of input, hidden and output nodes), and some parameters such as learning rate (η) and momentum (α) were also optimized for getting satisfactory results with minimum errors. The measuring range was 0.05-3.6 μg ml⁻¹ for both analytes. The proposed method has been successfully applied to the quantification of the sulfite and sulfide in different water samples.     

Potentiometric sensor arrays for the individual determination of penicillin class antibiotics using artificial neural networks

Arrays of potentiometric sensors with plasticized polymer membranes based on tetraalkylammonium organic ion exchanges with anions of penicillin class antibiotics (benzylpenicillin, ampicillin, oxacillin, and amoxicillin) have been proposed for the individual determination of antibiotics in model mixtures and pharmaceutical preparations. The following cross-sensitivity parameters of sensors have been estimated: the average slope of the electrode function (54 < S _av < 61), the nonselectivity factor (2.8 < F < 74.6), and the reproducibility factor (31.9 < K < 61.2). Artificial neural networks have been applied to the treatment of analytical signals from the multisensor system in the concentration range 2.5 × 10⁻⁴–1 × 10⁻¹ M. The average error of the individual determination of penicillin class antibiotics is 5–7%.     

Enzymatic methods for the determination of alpha-glycerophosphate and alpha-glycerophosphate oxidase with an automated FIA system

A procedure for the enzymatic determination of alpha-glycerophosphate (alpha-GP) has been developed, using an automated in-house FIA system, with immobilized glycerol-3-phosphate oxidase (GPO) on non-porous glass beads, following optimization of the immobilization and analytical parameters. Fabricated single bead string reactors (SBSR) were used in connection with the FIA system, following optimization of its parameters. The half-life of GPO-SBSR regarding reduction of the enzyme activity was found to be 110 days for its use in 20 triplicate measurements daily and storage at 4 degrees C in the appropriate buffer. The regression equation of the calibration graph for the determination of alpha-GP was: A(max)=(10+/-2)x10(-4)+(22 134+/-12)x10(-4) (mmol l(-1)alpha-GP). The lower limit of quantitation was 0.74 mumol l(-1)alpha-GP and the RSD of the method 0.05% (r=0.9999). The same FIA system and procedure can be also used for the determination of the GPO activity, with the alpha-GP as substrate. The regression equation for this calibration graph was: A(max)=(23+/-18)x10(-4)+(190+/-1)x10(-4) (mug ml(-1) GPO), the lower limit of quantitation was 0.782x10(-3) mg ml(-1) (0.782 ppm) GPO and the RSD of the method 0.53% (r=0.9999). Serum samples obtained from hospitalized patients were deproteinized by gel filtration and analyzed under pseudo-first order conditions, at various concentrations of alpha-GP. A kinetic study of the reduction of alpha-GP in serum versus time is given and an observed reaction rate constant k(ob)=106.5x10(-4) min(-1) was determined.     

Development of an FIA system with amperometric detection for determination of bentazone in estuarine waters

On the basis of its electrochemical behaviour a new flow-injection analysis (FIA) method with amperometric detection has been developed for quantification of the herbicide bentazone (BTZ) in estuarine waters. Standard solutions and samples (200 microL) were injected into a water carrier stream and both pH and ionic strength were automatically adjusted inside the manifold. Optimization of critical FIA conditions indicated that the best analytical results were obtained at an oxidation potential of 1.10 V, pH 4.5, and an overall flow-rate of 2.4 mL min(-1). Analysis of real samples was performed by means of calibration curves over the concentration range 2.5x10(-6) to 5.0x10(-5) mol L(-1), and results were compared with those obtained by use of an independent method (HPLC). The accuracy of the amperometric determinations was ascertained; errors relative to the comparison method were below 4% and sampling rates were approximately 100 samples h(-1). The repeatability of the proposed method was calculated by assessing the relative standard deviation (%) of ten consecutive determinations of one sample; the value obtained was 2.1%.     

Multienzymatic-rotating biosensor for total cholesterol determination in a FIA system

Fabrication of an amperometric-rotating biosensor for the enzymatic determination of cholesterol is reported. The assay utilizes a combination of three enzymes: cholesterol esterase (ChE), cholesterol oxidase (ChOx) and peroxidase (HRP); which were co-immobilizing on a rotatory disk. The method is developed by the use of a glassy carbon electrode as detector versus Ag/AgCl/3 M NaCl in conjunction with a soluble-redox mediator 4-tert-butylcatechol (TBC). ChE converts esterified cholesterol to free cholesterol, which is then oxidized by ChOx with hydrogen peroxide as product. TBC is converted to 4-tert-butylbenzoquinone (TBB) by hydrogen peroxide, catalyzed by HRP, and the glassy carbon electrode responds to the TBB concentration. The system has integrated a micro packed-column with immobilized ascorbate oxidase (AAOx) that works as prereactor to eliminate l-ascorbic acid (AA) interference. This method could be used to determine total cholesterol concentration in the range 1.2 μM-1 mM (r = 0.999). A fast response time of 2 min has been observed with this amperometric-rotating biosensor. Lifetime is up to 25 days of use. The calculated detection limits was 11.9 nM. Reproducibility assays were made using repetitive standards solutions (n = 5) and the percentage standard error was less than 4%.