Han Chern Loh

Accurate Zirconium Detection at Visible Wavelength Using Artificial Neural Network

Abstract A quantitative analysis has been carried out to determine the concentration of Zr4+ ion in aqueous solution by using alizarin red S (ARS) as reagent to form ARS‐Zr complex. ARS‐Zr complex gives a maximum absorption peak at wavelength of 520 nm with pH 2.5. The repeatability study at two different Zr4+ ion concentrations of 10 mg/l and 200 mg/l was found to give RSD values of 1.16% and 0.55%, respectively. The study of interfering ions was carried out with K+, Na+, Ca2+, Pb2+, Al3+ and Fe3+ ions with Zr4+ ion: interfering ions at mole ratio of 1:1, 1:10 and 1:100. The study showed a significant interfering effect at the ratio of 1:100 for all the ions except K+ and Na+ ions. Characterization of ARS‐Zr complex gives a dynamic Zr4+ concentration range of 5–35 mg/l. However, the use of artificial neural network (ANN) was able to extend the dynamic concentration range of Zr4+ to a larger range of 5–200 mg/l with optimized ANN parameters of 14 hidden neurons, 20 epochs, and training rate of 0.001. The sum square error (SSE) was found to have a value of 0.33 with an average error of 0.38.

An Optimization of Optical Fiber Salicylic Acid Sensor Using Artificial Neural Network

Abstract The application of an artificial neural network (ANN) in optimizing the response of an optical fiber salicylic acid (SA) sensor is presented in this paper. This sensor is fabricated based on immobilization of ferric(III) nitrate on Dowex‐50 × 8. The reflectance spectra of the sensor were measured by using an optical fiber spectrophotometer. A backpropagation (BP) ANN was used to analyze the response of the sensor developed. The results showed that the ANN technique was effective and useful in broadening the limited dynamic response range of the SA sensor (0.02 – 0.50 g/L) to an extensive calibration response (0.02–2.00 g/L). It was found that a network with 15 hidden neurons was highly accurate in predicting the response of the optical fiber SA sensor. This network scores a summation of squared error (SSE) skill and low average predicted error of 0.014 g/L and 0.032 g/L, respectively. Scholarship of National Science Fellowship (NSF) toward Han Chern Loh from the Ministry of Science, Technology and Environment (MOSTE), Malaysia is greatly acknowledged.

Application of Artificial Neural Network on Optical Method for Salicylic Acid Determination by Using Copper

Abstract A simple and rapid spectrophotometric method for salicylic acid (SA) determination has been carried out based on complexation. In this study, copper(II) acetate was used as a reagent. SA form a stable yellowish green complex with copper(II) acetate at pH 7. The useful dynamic range is 0.02 g/l–0.30 g/l. It has a maximum absorption at 397 nm and is stable for more than 50 h. The results were used to train up artificial neural networks (ANNs) for data optimisation. For training and validation, the Bayesian Regularization (BR) method and leave one out cross‐validation method were used, respectively. The model scores a correlation and summation of squared error (SSE) skill of 0.9993 and 1.5322 × 10−4 g/l, respectively, for two hidden neurons. The average predicted error is small (3.77 × 10−3 g/L). The Scholarship of National Science Fellowship (NSF) to Han Chern Loh from the Ministry of Science, Technology and Environment (MOSTE), Malaysia is greatly acknowledged.

Usage of Artificial Neural Network (Back Propagation) in Optimising Salicylic Acid Determination with Ferric(III) Nitrate

Abstract The use of artificial neural networks (ANN) in optimizing salicylic acid (SA) determination is presented in this paper. A simple and rapid spectrophotometric method for salicylic acid (SA) determination was carried out based on the complexation of salicylic acid–ferric(III) nitrate, SAFe(III). The SA forms a stable purple complex with ferric(III) nitrate at pH 2.45. The useful dynamic linear range is 0.01–0.35 g/L. It has a maximum absorption at 524 nm and the stability is more than 50 hours. The results were used for artificial neural networks (ANNs) training to optimize data. For training and validation purposes, a back‐propagation (BP) artificial neural network (ANN) was used. The results showed that ANN technique was very effective and useful in broadening the limited dynamic linear response range mentioned to an extensive calibration response (0.01–0.70 g/L). It was found that a network with 22 hidden neurons was highly accurate in predicting the determination of SA. This network scores a summation of squared error (SSE) skill and low average predicted error of 0.0078 and 0.00427 g/L, respectively.

Quantitative Analysis of Formaldehyde Using UV‐VIS Spectrophotometer Pattern Recognition and Artificial Neural Networks

Abstract This study is focused on the quantitative analysis of formaldehyde in aqueous solution using a Fluoral‐P reagent and describes the characterization of the reaction, including the effect of reagent concentrations, pH, response time, dynamic range, reproducibility, photostability, and selectivity by using an ultraviolet‐visible (UV‐VIS) spectrophotometer. The relative standard deviation value was 1.79 to 2.12%. The dynamic range of the complex gives a linear stimulation of 0.00 to 3.60 ppm for the concentration of formaldehyde. The reproducibility of this study is high, with 1.79 and 2.12% for 20 and 40 ppm of formaldehyde, respectively. The interference from acetaldehyde (formaldehyde: acetaldehyde=1:100) was lower than 2.10%. In addition, the application of artificial neural networks to quantitative analysis for formaldehyde has also been done in this study to optimize the dynamic range of formaldehyde involved in the formation of Fluoral‐P–formaldehyde complex. A three‐layer feed‐forward network and the back propagation algorithm‐operated training process were used in this study. For quantitative analysis of formaldehyde, artificial neural networks, networking with 23 hidden neurons and 40,000 cycle numbers with 0.001% learning rate, produce the best training results, with sum‐squared error value 0.5847.

Electrochemical sensor for the lead determination: development and characterisation

Lead electrochemistry sensor had been developed using screen printed electrode (SPE) technique with poly(vinyl) chloride (PVC) as printed matrix. The study has been carried out by using different type of electrodes. However, lead electrochemistry sensor had optimum response when deposited with mercury on the top of working electrode and in the buffer solution of pH 2. The optimum parameters for mercury deposition include deposited time = 15 s, deposited potential = 1000 mV, scanning range = 20 mV/s and amplitude pulse = 25. This sensor has linearity range of 1 - 50 ppm of Pb(II) R/sup 2/ of 0.9997. The LOD of this sensor is lower than 1 ppm of Pb(II).

Fabrication and characterization of novel salicylic acid sensors using different reagents (ferric and copper)

Abstract Fabrication of optical fiber salicylic acid (SA) sensors based on immobilization of ferric(III) nitrate and copper(II) acetate on Dowex‐50x8 is presented in this paper. The SA forms a stable purple complex with immobilized Fe3+ at pH 2.1 with a response time of 10 min while it forms a stable yellowish green complex with immobilized Cu2+ at pH 6.5 with a response time of 8 min. The reflectance spectra of the sensors were measured by using an optical fiber spectrophotometer. The results showed these SA sensors have maximum reflectance at 786 nm and 725 nm for SAFe complex and SACu complex, respectively. The useful dynamic response ranges are 0.02–0.50 g/L (SAFe) and 0.40–1.40 g/L (SACu). These complexes are stable for more than 24 hours. A good reproducibility (0.90%— SAFe; 0.86%— SACu) of measurement was obtained with these sensors.

Fabrication, characterisation and optimisation of novel salicylic acid sensors using different reagents and artificial neural network

Two optical fiber chemical sensors for salicylic acid (SA) have been fabricated using ferric(III) nitrate and copper(II) nitrate as reagent. Ion exchange resin, Dowex 50/spl times/8 had been used as solid support and optimisation was done by using artificial neural network (ANN). SA sensor using immobilised Fe(III) gave maximum reflectance at wavelength of 786 nm and pH 2.1. Dynamic range of SA concentration had been extended using ANN (15 hidden neurons) from 0.02 - 0.50 g/L to 0.02 - 2.00 g/L. SA sensor based on immobilised Cu(II) showed maximum reflectance at pH 6.5 and wavelength of 725 nm. ANN with 5 hidden neurons was able to extend dynamic range of SA concentration from 0.40 - 1.40 g/L to 0.05 - 2.00 g/L. Sum-squared error (SSE) for SA sensor based on Fe(III) and Cu(II) were 0.0140 g/L and 0.0310 g/L, respectively. These two sensors were stable for a period of more than 24 hours.

A preliminary study on. optical sensors for salicylic acid determination

Absaacr A simple and rapid spectrophotometric method for salicylic acid (SA) determination was carried out based on the complexation. In this study, copperpr) acetate was used as a colour developer. SA formed a stahle yellowish green complexation with copperpr) acetate at pH 6.8 7.7. The w o r h g dynamic range was O.O2g/l 0.30g/l. It had a maximum absorption at 4OOnm and stahle for more than 50 hours. The results were used to train an aitificial neural networks (ANN) for forecasting purpose. The model scores a correlation and summation of squared error (SSE) skill of 0.9994 and 1.2006 x g/l respectively for ten hidden neurons. The average