Kim Seng Chia

Predicting the boiling point of diesel fuel using adaptive linear neuron and near infrared spectrum

Monitoring the boiling point of a diesel fuel is an important step to understand the characteristics of the diesel fuel. This study evaluated the feasibility of adaptive linear neuron (Adaline) as a predictive model to predict the boiling point of diesel fuel based on near infrared spectrum. The parameters of learning rate and training cycle that involved in the optimization process were examined and discussed. The best predictive accuracy was achieved by Adaline that used learning rate of 0.001 and 788 adaptation cycles with root mean square error of prediction (RMSEP) of 3.42 OC and correlation coefficient of prediction (rp) of 0.9739. Findings show that Adaline with adaptive learning approach is capable of predicting the boiling point of diesel fuel based on near infrared spectrum without using data reduction approach.

A comparison between single layer and multilayer artificial neural networks in predicting diesel fuel properties using near infrared spectrum

ABSTRACT The implementation of near infrared spectroscopy in monitoring diesel fuel properties is highly dependent on the capability of its predictive model. This study investigates the feasibility of a single layer artificial neural networks among various predictive models in predicting the diesel fuel properties using near infrared spectrum. Results were compared and discussed with that reported in previous studies that used the same data in predicting the diesel fuel properties. Findings show that the proposed single layer outperforms popular models, and is comparable with a recent advanced models in predicting the diesel fuel properties using near Infrared spectrum without data reduction.

Prediction of glucose concentration using near infrared light and adaptive linear neuron

This study evaluates the relationship between near infrared light and glucose concentration by means of adaptive linear neuron. Firstly, the design and the development of the proposed glucose measurement device are presented. After that, the experiment design of acquiring sufficient near infrared data for training and testing is described. Next, adaptive linear neuron was trained and validated to predict the glucose concentration based on near infrared light. Findings indicate that the proposed glucose measurement approach by means of near infrared light with different wavelengths (i.e. 850nm, 860nm, 870nm, and 950nm) and adaptive linear neuron was capable of achieving RMSEC of 82.24 mg/dL, RMSEP of 70.04 mg/dL, rp of 0.9066, and rc of 0.9068 for a glucose concentration range of between 20 mg/dL and 600 mg/dL when the learning rate of 0.001 and 207 adaptation cycles were used.

A five band near-infrared portable sensor in nondestructively predicting the internal quality of pineapples

The determination of the fruit taste and grade depends on the internal quality of the fruit such as total soluble content, pH, and acidity. This paper investigates the feasibility of a non-destructive method to classify the internal quality of the pineapples using near infrared light and artificial neural network. Five near infrared light emitting diodes (LEDs) were used as the light source to emit near infrared light. A photodiode was used to measure the intensity of the reflected near infrared light from pineapples. The data of the acquired near infrared light were used to classify the internal quality of the pineapple using neural network. The random seed and the hidden neurons of the neural network were optimised to maximise the classification accuracy. Findings indicate that the neural network with seven hidden neurons was capable of achieving 30% misclassification.