Soft sensor modeling of feed liquid viscosity control for PVC gloves based on BP neural network

In the plastics Industry, feed liquid viscosity is always a vital input factor to the quality of final products, but difficult to realize real time measurement. Thus, in this paper, a data-driven soft sensor was developed to help control the viscosities of feed liquid in the production of PVC gloves which contribute a lot to the final quality and rating of gloves on the basis of literature review and study. BP neural network was selected to build the MIMO control model after discussing the methods in data pre-processing. The result shows that the inverse quality model has good performance in deciding the input values of feed liquid viscosity.