小脑模型在精馏塔浓度预测中的应用

摘要:小脑模型（cerebella model articulation controller 简称CMAC）是一种局部学习网络，CMAC 算法收敛速度很快，结构简单，并具有一定的泛化能力。精馏塔塔顶、塔底产品浓度含量的预测控制是精馏过程中非常重要的部分。研究表明，在某些情况下双线性系统可以很好地描述精馏塔的动态特性，因此本文通过双线性精馏塔机理模型给出网络学习样本数据，结合小脑模型对精馏塔塔顶、塔底产品浓度含量进行预测。仿真实验表明，该网络可以精确地描述精馏塔实际工程对象，取得良好的预测结果。
关键词CMAC 神经网络双线性精馏塔模型浓度预测
Abstract Cerdbella Model Articulation Controller（CMAC）is one kind of local-learning networks. CMAC features quick convergence，simple architecture and certain capability of generalization. The prediction for the concentration of product at the top and the bottom of distillation column is the most important part in the distilling process. Studies show that in some cases the dynamics property of distillation column can be well described by bilinear system.Therefore，according to the theory model of bilinear system，the learning data can be achieved to train the network. In combination with above training data，CMAC neural network is applied for distillation column to predict the concentration of distillated product at the top and the bottom. The simulation proves that practical engineering object of distillation column can be precisely described by CMAC neural network，and exccllcnt predictive result
is obtained.
Keywords CMAC neural network Model of bilinear distillation column Prediction of concentration