Sliding-window Recursive PLS Based Soft Sensing Model and Its Application to the Quality Control of Rubber Mixing Process

摘要

Rubber industry requires highly the quality control of rubber mixture. In order to overcome the shortage of PLS algorithm in rubber mixing process with complex nonlinearity and time-variance, an adaptive soft sensing model based on sliding-widow recursive PLS (RPLS) is presented to build a prediction model for the Mooney-viscosity of rubber mixture. The improved RPLS model can adaptively adjust the structures and parameters of PLS model according to on-line monitoring data and take characteristics of batch-wise data updated with the process changes and time-variant tracking capabilities. The application results show that the adaptive model has stronger tracking ability and higher precision than the traditional PLS model. The results also verify its effectiveness.