小腦模型在精餾塔濃度預(yù)測中的應(yīng)用

摘要:小腦模型（cerebella model articulation controller 簡稱CMAC）是一種局部學(xué)習(xí)網(wǎng)絡(luò)，CMAC 算法收斂速度很快，結(jié)構(gòu)簡單，并具有一定的泛化能力。精餾塔塔頂、塔底產(chǎn)品濃度含量的預(yù)測控制是精餾過程中非常重要的部分。研究表明，在某些情況下雙線性系統(tǒng)可以很好地描述精餾塔的動(dòng)態(tài)特性，因此本文通過雙線性精餾塔機(jī)理模型給出網(wǎng)絡(luò)學(xué)習(xí)樣本數(shù)據(jù)，結(jié)合小腦模型對精餾塔塔頂、塔底產(chǎn)品濃度含量進(jìn)行預(yù)測。仿真實(shí)驗(yàn)表明，該網(wǎng)絡(luò)可以精確地描述精餾塔實(shí)際工程對象，取得良好的預(yù)測結(jié)果。
關(guān)鍵詞CMAC 神經(jīng)網(wǎng)絡(luò)雙線性精餾塔模型濃度預(yù)測
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