人工神经网络-微分脉冲伏安法同时测定尿液中的多巴胺、尿酸及抗坏血酸

Simultaneous Determination of Dopamine,Uric Acid and Ascorbic Acid in Urine Sample by Differential Pulse Voltammetry and Artificial Neural Networks

研究了多巴胺、尿酸和抗坏血酸在玻碳电极上的伏安行为。在pH 5.7的Britton-Robinson缓冲溶液中,采用微分脉冲伏安法进行电化学扫描,3种化合物均有良好的氧化峰,但其波谱重叠严重,常规伏安法难以同时测定。采用化学计量学方法中的偏最小二乘法(PLS)、主成分回归法(PCR)、径向基人工神经网络法(RBF-ANN)和主成分分析-径向基人工神经网络(PCA-RBF-ANN)方法并在实验数据处理后以化学计量学方法对其进行解析,结果发现采用PCA-RBF-ANN方法时可获得最小预报误差,避免了进一步的分离和提纯步骤,简化了测定过程,并实现了人工合成样品中3组分的同时测定。将该方法用于实际样品的测定,结果满意。

The voltammetric behaviours of ascorbic acid（AA）,uric acid（UA） and dopamine（DA） were studied at glass carbon electrode.Well-defined volammetric oxidation peaks for the three compounds were obtained by differential pulse voltammetry（DPV） in pH 5.7 Britton-Robinson buffer.However,serious overlappings of individual voltammetric peaks were observed for the mixtures of these compounds,which were difficult to determine by normal methods.To facilitate the resolution of these complex voltammograms,different calibration models were investigated based on the different chemometrics methods,including principal component regression（PCR）,partial least squares（PLS）,radial basis function-artificial neural networks（RBF-ANN） and principal component analysis-radial basis function-artificial neural networks（PCA-RBF-ANN）.The results indicated that the lowest relative prediction error（RPE） was obtained with the PCA-RBF-ANN method.The method was successfully applied in the analysis of synthetic mixtures and urine sample without separation and purification.The developed method was then applied in the analysis of the compounds with satisfactory results.