The application of this method to dynamic modeling of continuous stirred tank reactor (CSTR) shows that the GRPE trained DRNN has higher modeling precision and faster convergence than BP trained MLP.

• 利用所提出方法对CSTR动态建模结果表明，基于GRPE训练的DRNN比基于BP训练的MLP模型精度高，收敛速度快。