Soft-sensor Modeling Method Based on Support Vector Machine

A soft sensor method based on support vector machine is presented, and a soft-sensor model for biomass estimation in penicillium fermentation process is developed by this method.The effects of parameter adjusting and selection of kernel functions on model quality are analyzed by simulation experiments.With data collected from real plant,models based on some other technologies are also presented.Experiments show that support vector machine method is superior to traditional modeling methods in theory.