马克斯克鲁维酵母菌发酵过程的软测量

Soft Sensor Modeling for Kluyveromyces Marxianus Fermentation Process

针对马克斯克鲁维酵母菌生长发酵环境的复杂性和高度非线性,提出采用最小二乘支持向量机建立马克斯克鲁维生长发酵条件的预测模型,以四个主要影响因素温度、时间、PH值、培养基中葡萄糖含量为输入,细胞生物量为输出。通过预测在一定影响因素下的细胞生物量,从而找到该酵母菌最佳的发酵条件。仿真实验结果表明该方法具有建模速度快、预测精度高、操作简便等优点,不仅克服了常规的BP预测模型的不足,而且性能优于标准支持向量机预测模型。

Aiming at the complexity and high non-linearity of Kluyveromyees Marxianus fermentation process, a novel model based on least square support vector machine was proposed to pre-estimate the optimal fermentation condition. We used the main factors： temperature, time, PH, dextrose as input of prediction model, and cell life-form quantity as output. The prediction roodel will pre-estimate the cell life-form quantity in the given condition, by doing so we can easily find out the optimal fermentation condition. Experiment results showed that our proposed model method has the advantage of quick speed, high precision, simpleness and so on; it also performed more outstandingly than traditional SVM.