小波变换与神经网络结合用于示波计时电位测定

Application of the Combination of Wavelet Transform and Neural Networks in Oscillographic Chronopotentiometric Determination

小波变换与神经网络的结合是化学计量学中一个十分活跃的研究领域。作者介绍了小波变换与神经网络的两种结合方式———辅助式结合和嵌套式结合 ,并将二者用于单组分示波计时电位信号定量计算的比较。结果表明 :辅助式结合训练次数少 ,预测误差小 ;小波神经网络计算的自动化程度高 。

The researches on the combination of wavelet analysis and neural networks have attracted much attention in recent years. Two combination types, the ancillary type and the embedded type, are described and applied to predict quantitatively the concentration of mono-component in oscillographic chronopotentiometry. The experimental results of the prediction show that the ancillary type has higher convergence rate and prediction accuracy, but the embedded type is more automatic. Compared to standard back propagation neural network, two types are better than artificial neural network.