青霉素发酵过程软测量建模仿真研究  

Modeling and Simulating Study of Soft Measurement in Penicillin Fermentation

作　　者：崔浩阳 李凌 Cui Haoyang;Li Ling(College of Information Engineering,Shenyang University of Chemical Technology,Shenyang,110142,China)

机构地区：[1]沈阳化工大学信息工程学院,辽宁沈阳110142

出　　处：《石油化工自动化》2022年第6期1-5,共5页Automation in Petro-chemical Industry

基　　金：国家自然科学基金项目(60874057);国家高技术研究发展计划项目(2008AA042902)。

摘　　要：青霉素发酵过程是一类具有较强非线性、时变性和不确定性的典型间歇过程,其中产物浓度等关键生物参数难以实时在线检测,而离线化验则需要高昂成本。针对青霉素发酵过程软测量建模问题,首先利用k近邻互信息方法选择菌体浓度、基质体积、二氧化碳浓度和补料温度作为辅助变量,然后利用稀疏自编码器提取数据特征,结合最小二乘支持向量机进行软测量建模。仿真实验结果表明:所建软测量模型可达到较好的预测效果,无监督机器学习方法在青霉素发酵软测量建模过程中应用效果良好,为进一步对青霉素发酵过程的菌体浓度控制的研究奠定了基础。Penicillin fermentation is a typical batch process with strong nonlinearity,time variability and uncertainty.Real-time online testing is difficult to be conducted for key biological parameters such as product concentration,while off-line assay is costly.Aiming at the problem of soft measurement modeling for Penicillin fermentation,k-nearest neighbor mutual information method is used to select bacteria concentration,matrix volume,carbon dioxide concentration and supplementary food temperature as auxiliary variables,then,sparse autoencoder is used to extract data features,and least square support vector machine is used to build soft measurement model.The simulation results show that the constructed soft measurement model can achieve better prediction effect.The application effect of unsupervised machine learning method in the construction of soft measurement model of penicillin fermentation is good,which lays a foundation for further research on the control of bacteria concentration in penicillin fermentation.

关 键 词：青霉素发酵 软测量 k近邻互信息 稀疏自编码器 最小二乘支持向量机 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]