基于贝叶斯网络的软测量建模方法

A soft sensor based on Bayesian network

软测量技术对化工生产过程中提高产品质量和保证安全生产具有重要的作用,因此对化工软测量建模方法的研究具有重要意义。本文将贝叶斯网络应用于化工软测量建模,采用高斯混合模型来近似表达贝叶斯网络模型中的联合概率分布,通过Expectation Maximization算法求解出高斯混合模型参数并给出了贝叶斯网络估计公式。应用此法分别对某炼油厂脱丁烷塔塔底丁烷含量和某双酚A生产过程中脱水塔出口组分苯酚含量建立了软测量模型,取得了良好的离线估计结果。仿真结果表明,与支持向量机相比,在估计精度相当的情况下,省去了许多过程参数的估计,是1种有效的软测量建模方法。

Soft sensor has played an important role in improving product quality and can ensure the safety of chemical production processes,therefore the investigation of soft-modeling method is very important.A new approach based on Bayesian network applied to chemical soft sensor is proposed. The joint probability distribution is approximated by gaussian mixture model in Bayesian network,and parameters of the Gaussian mixture model are obtained through the EM algorithm,and then the estimated formula for Bayesian network is been given.Two soft sensor models are respectively established with this method to estimate the C4 in the bottom of debutane tower and phenol at the outlet of Dehydration Tower.And the models based on Bayesian network show good results.Compared with support vector machine,the Bayesian network eliminates a number of the process parameters estimation and has considerable accuracy.It is an effective method for soft sensor modeling.