在pH 6.5的0.3 mol·L-1氯化钾介质中,在电位值-0.71 V及-0.74 V处可分别见到对硝基氯苯及邻硝基苯酚的灵敏极谱峰,可分别对此两种化合物进行测定,但因两者相互干扰不能在混合物中同时测定.将改进的小波神经网络与BP神经网络相结合,提出一种新的混级联神经网络结构,并用于单扫描示波极谱法同时测定对硝基氯苯和邻硝基苯酚.通过对网络结构的优化和网络参数的调整,加快了训练速度,提高了预测的准确度.该法用于混合样品中同时测定对硝基氯苯和邻硝基苯酚,其相对误差和回收率分别为3.76%,96.2%;4.05%,96.0%.

In a 0.3 mol·L-1 KCl medium of pH 6.5,sensitive polarographic peaks of p-nitrochlorobenzene (p-NCB) and o-nitrophenol (o-NP) were observed at -0.71 V and -0.74 V respectively,but the determination of these 2 compounds by the oscillopolarographic method was not possible in the presence of each other due to very close values of their potentials.A new method of adaptive chemical pattern recognition,i.e.,the cascade artificial neural network (CANN) based on the combination of the improved wavelet ANN and BPANN,was proposed and applied to the simultaneous determination of p-NCB and o-NP in a mixture by the single-seeeping oscillopolarography.By optimization of the structure of CANN and adjustment of the parameters,the training speed and prediction accuracy were remarkably improved.In the analysis of 10 simulated samples of mixtures of p-NCB and o-NP,relative deviation and recovery obtained were 3.76% and 96.2% respectively for p-NCB,and were 4.05% and 96.0% respectively for p-NP.