基于动态BP神经网络的漂白过程辨识研究

Research on Identification of Bleaching Process based on Dynamic BP Neural Network

针对纸浆漂白过程的大惯性、大时滞、非线性,时变和多干扰的难控特点,本文提出利用动态BP神经网络来建立纸浆漂白过程的模型,仿真结果表明,该网络具有较好的准确性,能够较为真实的反应漂白过程。然后利用训练好的网络模型来模拟实际生产过程,在此基础上通过阶跃响应法来辨识对象模型的参数,得出了过程的具体数学模型,为纸浆漂白的优化控制提供了可能。

Paper bleaching process is a large inertial, large time-delay, non-linear, time-varying process with much random disturbance, so it is difficult to control. Modeling the process based on dynamic BPNN is brought forward in this paper, the simulation result proves that the network can imitate the process accurately. Adopting the step response based on the model to identify the parameter, the math model of the process has been given, it is helpful to optimization control of paper bleaching.