摘要
本文利用改进的“反向传播”神经网络模型,在滴定突跃附近,建立了E-V曲线的神经网络插值模型,由其二阶微商求得滴定终点。计算实例中,拟合最大相对误差不超过0.1％,计算机CPU时间不超过20s,实验结果表明,该方法性能良好,在电容量分析方面有广阔的应用前景。
The approximation model for E-Vcurve near titration steep is constructed by an improved back-propagation neural network model. By calculating the second order derivative of the constructed model, titration end-point is determined. In our examples, the maximum fitting relative error does not exceed 0.1% and the total calculating time does not exceed 20 s. The experimental results show that the neural network is good,and it might be widely used in electro-volumetric analysis.
出处
《分析化学》
SCIE
EI
CAS
CSCD
北大核心
1993年第4期439-442,共4页
Chinese Journal of Analytical Chemistry
关键词
电位滴定终点
人工
神经网络
Potential titration end-point, Artificial neural network, Improved back-propagation model,Second order derivative.
