Application of artificial neural networks coupled with an orthogonal design and optimization algorithms to multifactor optimization of a new FIA system for the determination of uranium(vi) in ore samples

Abstract

A sensitive and selective spectrophotometric flow injection method has been developed for the determination of uranium(VI) in ore samples, based on the reaction of uranium(VI) with p-acetylchlorophosphonazo (CPA-pA) in a HNO₃ medium. Most of the interfering ions were effectively eliminated by the masking reagent, diethyleneaminepentaacetic acid (DTPA). Artificial neural networks coupled with an orthogonal design and penalty algorithm were applied to the modeling of the proposed flow injection system and optimization of the experimental conditions. An orthogonal design was utilized to design the experimental protocol, in which three variables were varied simultaneously. ANNs with a faster back propagation (BP) algorithm were used to model the system. Optimum experimental conditions were generated automatically by using jointly ANNs and optimization algorithms in terms of sensitivity and sampling rate. In the U(VI)–CPA-pA system, Beer’s law was obeyed in the range 1.0–23.0 μg mL⁻¹, the detection limit for uranium(VI) was 0.3 μg mL⁻¹ and the sampling rate was 100 h⁻¹. The method was applied to the determination of uranium(VI) in ore samples with satisfactory results. It was shown that this method had advantages over traditional methods in respect of improvement in the ability of optimization and reduction in analysis time.