摘要
针对海洋生物酶发酵过程关键生物参数(如基质浓度、菌体浓度等)难以实现直接在线测量的问题,提出了一种基于贝叶斯准则的最小二乘支持向量机(LS-SVM)软测量建模方法。以典型的海洋生物酶——海洋蛋白酶为研究对象,在分析海洋蛋白酶发酵过程机理的基础上,确定了发酵过程软测量模型的辅助变量和主导变量。考虑到在LS-SVM建模中,正规化参数和核参数的优化是建模的难点,采用贝叶斯准则对LS-SVM参数进行优化,进而建立了基于贝叶斯准则的LS-SVM软测量模型。仿真验证结果表明,该模型比传统的LS-SVM具有更高的预测精度和泛化能力。
In order to solve the problems of real-time measurement of crucial biological variables （ such as biomass concentra- tion, substrate concentration and so on） in the marine biologic enzyme fermentation process, a soft sensor modeling method based on least squares support vector machine （ LS - SVM） and Bayesian criteria was proposed, taking a typical Marine biological enzyme- marine protease as the research object. Firstly, the auxiliary variables and the dominant variable of soft sensor model were deter- mined based on the analysis of the mechanism of Marine protease fermentation process. Secondly, considering the normalisation pa- rameter and kernel parameter were the difficulties in LS -SVM modeling, Bayesian criteria was used to optimize parameters of LS - SVM, and then soft sensor model was established by LS - SVM based on Bayesian criteria. The simulation results show that the modeling has higher prediction precision and better generalization ability than LS - SVM.
出处
《仪表技术与传感器》
CSCD
北大核心
2014年第8期92-94,110,共4页
Instrument Technique and Sensor
基金
"十二五"国家863重点科技项目(2011AA09070301)
江苏省高校优势学科建设工程资助项目(苏政办发(2011)6号)
江苏省科技计划项目(BE2010354)
关键词
贝叶斯准则
最小二乘支持向量机
参数优化
海洋蛋白酶
软测量
Bayesian criteria
least squares support vector machine （ LS - SVM ）
parameters optimization
marine protease
soft sensor
