Soft-sensor modeling of foot-and-mouth disease vaccine suspension culture based on relevance vector machine

Abstract

The suspension culture of animal cells in vitro involves complex biochemical reaction and metabolism of a large number of cells. The on-line detection of some biological parameters (such as glucose concentration, lactic acid concentration and cell density and so on) is very important and difficult to achieve. The off-line analysis also will lead to the delay of the system regulation and control. In order to estimate the key parameters of suspension culture of foot-and-mouth disease (FMD) vaccine by a more precise method, a soft-sensor method based on relevance vector machine (RVM) is adopted in this paper. Firstly, the auxiliary variables of the model are determined by using the uniform incidence degree algorithm, and the dominant variables are determined by characteristics of the FMD vaccine during the preparation process, then the soft-sensor model is established by using the RVM algorithm. Compared with SVM, the simulation results show that the RVM algorithm can solve the problem of real-time detection of biological parameters in the animal cell culture effectively. It can also reduce the complexity of the model and has a certain practical value.