小波变换-插值-Elman神经网络5-Br-PADAP分光光度法同时测定铝合金中铜铁镍

Wavelet transform-interpolation-Elman neural network spectrophotometry for the simultaneous determination of copper,iron and nickel in aluminium alloy with 5-Br-PADAP

通过对铝合金中铜、铁、镍测定条件的研究,提出以5-Br-PADAP为显色剂,以OP为增溶增敏剂,在pH3.5的乙酸盐缓冲液条件下,用小波变换-插值-Elman神经网络分光光度法同时测定铝合金中的铜、铁、镍的方法。该法在小波变换对吸收光谱数据进行降噪处理的基础上,采用一维线性插值处理使训练集样本对待辩识空间形成较好的覆盖,从而使Elman网络能够更有效地提取信息,改善网络性能。研究结果表明,该法可以显著降低样品测定的相对误差。对铝合金样品中铜、铁、镍的测定,其结果与原子吸收光谱法一致。

On the basis of the study of conditions for determining copper,iron and nickel in aluminium alloy ,a new method for simultaneous determination of copper, iron and nickel was proposed by spectrophotometry combining with wavelet transform-interpolation-Elman neural network under the conditions of using 5-Br PADAP as chromogenic agent in the presence of the sensitizer/solubilizer OP and in acetate buffer solution of pH 3.5. In this method , noise in visible spectra was eliminated by wavelet transform and linear interpolation was used to make specimen of training set cover recognizable room better. As a result, Elman neural network could collect characteristic information more efficiently and improved its performance. The results showed that relative error of sample determination was reduced remarkably. The method was developed to analyze copper, iron and nickel in aluminium alloy samples. The results were consistant with that obtained by AAS.