Calibration of a Potentiometric Multi-Sensor Array for the Determination of Na+, K+ and NH4+ Ions by Using Artificial Neural Networks

Abstract

A potentiometric multi-sensor array, comprising of all-solid-state selective and non-selective electrodes, was constituted and calibrated via artificial neural networks (ANNs) for the determinations of NH₄⁺, Na⁺ and K⁺ ions in aqueous model solutions. Various artificial neural network configurations were constituted and their root mean square errors of calibration (RMSEC) and root mean square errors of prediction (RMSEP) values were calculated. The model which has the lowest RMSECxRMSEP value was preferred as the best model. Artificial neural network models calculated for Na⁺ and K⁺ have nearly similar prediction ability to univariate calibrations which were performed using the main ion concentrations of the calibration solutions and respective ion-selective electrode responses. Artificial neural network model calculated for NH₄⁺ had the superior prediction ability when compared with its univariate calibration model.