基于部分最小二乘法的流化床反应器的神经网络模型

利用部分最小二乘法能将高维输入数据投影到低维特征向量空间，以及神经网络可以近似任何非线性的特性，针对某丙烯氨氧化合成丙烯腈流化床反应器的试验数据，讨论了神经网络与部分最小二乘的组合方法，提出了基于部分最小二乘法的流化床反应器神经网络估计模型。仿真结果表明了该模型的有效性。

Neural Network Model Based on Partial Least Square for Fluidized Bed Reactor

Partial least square(PLS) is capable of projecting the information in high dimensional space down tolow dimensional space and neural networks have universal approximation nonlinear property. The paper discussesthe combination algorithm of neural network and PLS. A neural network model of the fluidized bed reactorbased on partial least Square proposed using the experimental data of the fluidized bed reactor for synthesizingacrylonitrile from amrnxidation of propylene. The simulation results show the proposed model is effective.