摘要
针对赖氨酸发酵过程中关键生物参数(菌体浓度、基质浓度、产物浓度)难以在线测量的问题,提出一种基于自适应提升法(Adaboost)与核主元回归(KPCR)的软测量建模方法。利用Adaboost算法具有将弱学习算法提升为强学习算法的性能,将其引人KPCR中,提升KPCR模型。以氨基酸典型菌种赖氨酸发酵过程为研究对象,采用基于Adaboost算法与KPCR的软测量模型进行预测仿真,仿真结果表明该模型能够对赖氨酸发酵过程的三个关键生物参数进行较准确的预测,与单一的KPCR模型相比,泛化能力强,预测精度高。
To solve the problem that the crucial biochemical parameters are measured difficultly on hne in lysine fermentation process, a method based on Adaboost algorithm and KPCR is proposed. Adaboost algorithm has the capability to elevate weak learning algorithm to strong learning algorithm. In this paper, the performance of KPCR is improved by Adaboost algorithm. The simulation results indicate that the proposed method can be used to predict crucial biochemical parameters of lysine fermentation actually. Comparing with the traditional KPCR soft sensor model, it has good stability and meets the accuracy requirements.
出处
《信息技术》
2015年第11期105-108,共4页
Information Technology
