电子世界
電子世界
전자세계
ELECTRONICS WORLD
2014年
11期
74-75
,共2页
聚氯乙烯%T-S模糊神经网络%汽提过程%最邻近聚类算法
聚氯乙烯%T-S模糊神經網絡%汽提過程%最鄰近聚類算法
취록을희%T-S모호신경망락%기제과정%최린근취류산법
PVC%T-S fuzzy neural network%Stripping process%New nearest neighbor clustering algorithm
由于汽提过程是一个具有高度非线性和时变等特点的复杂工艺过程，其准确的数学模型难以建立，过程参数也较难控制，运用传统的控制方法难以达到高精度的控制效果。基于T-S模糊神经网络具有较强的逼近和学习能力，又需要较少的先验知识，能够在线学习等特点，本文采用最邻近聚类法对传统的T-S模糊神经网络方法进行改进，同时，采用共轭梯度法和递归最小二乘法来确定模型得参数，结合现有的该公司的实际运行数据，建立了汽提塔系统的模型。仿真结果验证了采用T-S模糊神经网络方法建模的有效性。
由于汽提過程是一箇具有高度非線性和時變等特點的複雜工藝過程，其準確的數學模型難以建立，過程參數也較難控製，運用傳統的控製方法難以達到高精度的控製效果。基于T-S模糊神經網絡具有較彊的逼近和學習能力，又需要較少的先驗知識，能夠在線學習等特點，本文採用最鄰近聚類法對傳統的T-S模糊神經網絡方法進行改進，同時，採用共軛梯度法和遞歸最小二乘法來確定模型得參數，結閤現有的該公司的實際運行數據，建立瞭汽提塔繫統的模型。倣真結果驗證瞭採用T-S模糊神經網絡方法建模的有效性。
유우기제과정시일개구유고도비선성화시변등특점적복잡공예과정，기준학적수학모형난이건립，과정삼수야교난공제，운용전통적공제방법난이체도고정도적공제효과。기우T-S모호신경망락구유교강적핍근화학습능력，우수요교소적선험지식，능구재선학습등특점，본문채용최린근취류법대전통적T-S모호신경망락방법진행개진，동시，채용공액제도법화체귀최소이승법래학정모형득삼수，결합현유적해공사적실제운행수거，건립료기제탑계통적모형。방진결과험증료채용T-S모호신경망락방법건모적유효성。
The stripping process is a highly nonlinear and time-varying complex process,it is difficult to establish accurate mathematical model,and process parameters are difficult to control,to achieve the use of traditional control method control effect with high accuracy.Based on T-S fuzzy neural network has a good approximation and learning ability,and less the prior knowledge repured,to online learning characteristics,The nearest neighbor clustering method is applied to improve the traditional T-S fuzzy neural network method in this paper. At the same time,using the conjugate gradient method and the recursive least square method is used to determine the model parameters,combined with the actual operation data of the company's existing,stripper system model is established.The simulation results verify the validity of the model.
