粗集理论在萃取精馏塔软测量中的应用

丁二烯萃取精馏过程中,副产品抽余液(BBR)的质量(丁二烯含量)和很多工艺参数有关,工艺参数之间又是相互关联、耦合的,并具有噪声。应用粗集方法将这些工艺数据进行压缩和抽提,解决了工艺参数间的相关问题,同时去掉了一些信息量不大,并带来噪声的成分。用模糊C均值聚类算法将训练集分成具有不同聚类中心的子集,每一子集用径向基函数(RBF)网络进行训练来获得子模型,然后用模糊聚类产生的隶属度将各子模型的输出加权求和得到BBR中顺丁烯的含量,由顺丁烯的含量来估计丁二烯含量。结果表明,这种软测量算法具有较好的建模效果,由于采取了数据分组训练,大大节省了建模的训练时间,比单纯的基于神经网络的方法要快得多。

Application of Rough Sets for Soft-Sensing in Tower of Extractive Distillation

In the process of extractive distillation of butadiene, the quality of By-product Butane-butene Raffinate (BBR) is related to many process parameters, which are coupling and with noises. In the paper, the process parameters are compressed and abstracted by the method of Rough Sets. The problem of coupling of process parameters is solved, and the components with noises and less information are omitted. Fuzzy C-means clustering(FCM) algorithm is used for separating a whole training data set into several clusters with different centers, each subset is trained by Radial Base Function Networks(RBFN).The degree of membership is used for combining several models to obtain the finial content of Z- butene in the BBR, which can be used to estimate the content of butadiene. The obtained results demonstrate that this soft-sensing algorithm is of a high modeling accuracy, and the time of training by this means is much shorter than by pure neural net work means because of dividing training data set into several subsets.