【摘要】 针对支持向量机(SVM)计算复杂度高、参数不容易确定等局限性,提出一种基于相关向量机(RVM)的赖氨酸反应过程关键参量的软测量方法.根据过程经验,确定发酵液的溶解氧浓度、pH值、二氧化碳释放率、氧吸收率和葡萄糖流加速率为辅助变量,利用相关支持向量机的拟合与泛化能力,建立了赖氨酸反应过程基质浓度、菌体浓度、产物浓度等不可直接测量参量的软测量模型.基于L-赖氨酸反应过程开展的试验研究表明:所建立的相关向量机软测量模型拟合精度高、泛化能力强,较好地满足了赖氨酸反应过程的控制要求.更多还原

【Abstract】 To overcome the high computational complexity and difficulty in design kernel parameters of support vector machine,the soft sensor model of lysien fermentation biological parameters is proposed based on relevance vector machine.According to procedure experience,dissolved oxygen,the parameters of pH value,discharge rate of CO2,absorption rate of O2,and flow acceleration rate of glucose are chosen as auxiliary variables,and the concentration of matrix,thallus and product are chosen as primary variables.The soft sensor of lysien fermentation procedure is built using the fitting and generalization capacity.The experiments results show that the obtained soft sensor is with high fitting precision and generalization capacity,which well satisfies the control requirement of lysien fermentation procedure.更多还原