论文部分内容阅读
目的 对神经网络反传算法 (BP)加以改进并用于复方药物的测定。方法 采用改进算法 (MBPN) ,并用XOR问题考察改进前后的收敛速度。结果 将MBPN算法用于复方扑热息痛中三组分的同时测定 ,相对误差为 - 6 .0～ +9.0 %。结论 MBPN法经优化选择传播参数γ加快了神经网络NN的收敛速度 ,增强了反传算法BP的可用性和实用性。用于复方药物的同时测定结果良好、操作简便。
Objective To improve the BP neural network algorithm and use it in the determination of compound drugs. Methods The improved algorithm (MBPN) was used and the convergence rate before and after the improvement was investigated by XOR. Results The MBPN algorithm was applied to the simultaneous determination of three components in compound paracetamol with the relative error of -6.0 ± 9.0%. Conclusion The MBPN method optimizes the choice of propagation parameters γ to speed up the neural network NN convergence speed and enhances the usability and practicability of back propagation algorithm BP. Simultaneous determination of compound drugs results are good, easy to operate.