CAJ | 학술논문

由于生化反应过程的复杂性和高度非线性,多数简单的数学模型不能准确描述.该文基于Matlab软件,利用改进的支持向量机(v-SVR)对植物乳酸杆菌发酵这一典型生化过程进行研究,应用遗传算法估计模型最优参数,建立植物乳杆菌的菌体密度预测模型.同时建立传统的logistic动力学模型以进行比较.结果表明,采用结合遗传算法的v-SVR预测模型拟合误差小,皮尔森相关系数(R)更高,可以较好地预测乳酸杆菌的发酵过程,为其优化控制及放大提供依据.
유우생화반응과정적복잡성화고도비선성,다수간단적수학모형불능준학묘술.해문기우Matlab연건,이용개진적지지향량궤(v-SVR)대식물유산간균발효저일전형생화과정진행연구,응용유전산법고계모형최우삼수,건립식물유간균적균체밀도예측모형.동시건립전통적logistic동역학모형이진행비교.결과표명,채용결합유전산법적v-SVR예측모형의합오차소,피이삼상관계수(R)경고,가이교호지예측유산간균적발효과정,위기우화공제급방대제공의거.
Due to the complexity and high non - linearity of bioprocess,most simple mathematical models cannot describe the exact behavior of biochemistry systems. An approach via v support vector regression (v - SVR) based on Matlab software platform is proposed for modeling the fermentation of Lactobacillus plantarum. With the methods of genetic Algorithm, We evaluated parameters involved in the SVR model, and developed a predictive model of L. plantarum density. According to the contrast of the predicting capability, the logistic model was also developed. The results indicated that v - SVR model can pre - estimate fermentation process of L. plantarum, which had smaller fitting error and bigger Pearson correlation coefficient. This v - SVR model can be useful for optimizing and up scaling lactic acid bacteria fermentation process.