基于最小二乘支持向量机的发酵过程混合建模

Hybrid Modeling of Fermentation Process Based on Least Square Support Vector Machines

提出了一种综合先验知识与最小二乘支持向量机的发酵过程建模方法,并且采用遗传算法进行最小二乘支持向量机的参数优化选取。该模型应用到一个具体发酵过程状态变量的预估中,仿真结果表明基于最小二乘支持向量机的混合模型具有很高的精度与范化能力,同时也表明了最小二乘支持向量机是软测量建模的一种有效方法。

A method for synthesizing fed-batch fermentation that combines prior knowledge and least square support vector machine （LS-SVM） was presented. Prior knowledge enters the hybrid as a simple process model and first principle equations. The genetic algorithms was investigated to select the parameters of LS-SVM models as a means of improving the LS-SVM predictions. The hybrid model based on LS-SVM was applied to predication of state variable in a fed-batch fermentation. The simulation results show that the hybrid model gives better estimates of state variable. And it also has good generalization capabilities. At the same time, it also indicates that LS-SVM is of potential application in soft sensor.