级联神经网络用于单扫描示波极谱信号的测定  

Determination of single-sweep oscillopolarographic signals via cascade neural network

作　　者：仲红波[1] 周长利[2] 张霞[2] 陈明俊[2] 赵笛思[2] 李关宾[1] 

机构地区：[1]山东大学化学化工学院,济南250100 [2]济南大学化学化工学院,济南250022

出　　处：《分析试验室》2005年第12期18-23,共6页Chinese Journal of Analysis Laboratory

摘　　要：将改进小波神经网络与BP神经网络相结合,提出一种新的混级联神经网络结构,用于单扫描示波极谱信号的同时测定。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。用该法对邻、间硝基氯苯混合样进行了预测,结果满意。对级联神经网络法与单一BP神经网络法的预测结果进行了比较,表明级联神经网络优于单一BP神经网络。By combining the improved wavelet neural network with BP neural network, a new structure based on mixed cascade neur~ new neural network have network was established. This been used in the determination of single-sweep oscillopolarographic signals. The training speed and the predication accuracy can be enhanced by optimizing the network structure and parameters. Concentrations predicated for m-Nitrochlorobenzene （m-NCB in short） and p-Nitrochlorobenzene （p-NCB in short） mixture are satisfactory. All the results suggested that the cascade neural network method was better than the single BP neural network method.

关 键 词：级联神经网络 单扫描示波极谱 硝基氯苯 

分 类 号：O657.1[理学—分析化学]