文中提出双模型结构RBF(RadialBasisFunction)神经网络,结合工艺机理和相关分析法,筛选出影响较大的变量。对现场数据,用小波分析法,剔除噪声和故障数据,考虑各输入信号对软仪表影响时间的区别,分别采用不同的滞后时间,建立了常减压蒸馏装置质量软仪表模型,取得较好的结果。
Abstract
A RBF neural network with a double model structure was proposed. Combining the mechanism of a process with a correlation analysis, the important variables were selected. The defect and the high level noise in signals of the field site were eliminated by using the wavelet analysis method. The time difference of the various multiple input variables acting on the software instrument were considered and the difference delay time was used. Through applying the RBF neural networks to the atmospheric and vacuum distillation units,a good estimation of the production quality was showed in this paper.
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1 ] 方崇智,萧得云. 过程辨识[M] . 北京:清华大学出版社,1988 ,160 - 180
[2 ] 王永骥,涂 健. 神经元网络控制[M] . 北京:机械工业出版社,1997 ,68 - 86
[3 ] Chen S , Billing S A , Cowan C F N. Practical identification of NARMAX models using radial basis functions[J ] . INT J Control ,1990 ,52(6) :1327 - 1350
[4 ] Chen S , Cowan C F N , Grant P M. Orthogonal least squares learning algorithm for radial basis function networks [ J ] . IEEE Transactions on Neural Networks , 1991 ,2 (2) :302 - 309
[5 ] 俞金寿,刘爱伦,张克进. 软测量技术及其在石油化工中的应用[M] . 北京:化学工业出版社,2000
[6 ] 李海青. 软测量技术原理及应用[M] . 北京:化学工业出版社,2000
{{custom_fnGroup.title_cn}}
脚注
{{custom_fn.content}}