基于滑动窗的RPLS在线校正模型在橡胶混炼质量控制过程中的应用

Application of on-line corrective model based on sliding window of RPLS for quality control process of rubber mixing

在传统偏最小二乘法(PLS)建模的基础上,采用滑动窗方法,用递推偏最小二乘法(RPLS)根据在线数据自适应调整PLS混炼模型的结构和参数,建立橡胶混炼质量控制的RPLS在线校正模型。仿真与试验结果显示,该模型能自适应跟踪混炼的过程变化,门尼粘度预测误差为±3,符合生产要求。

On the basis of traditional partial least squares （PLS） modelin method and recursive partial least squares （RPLS）,which adaptively adjusted t meters of PLS model for mixing according to on-line data, were used to build g, the sliding window he structure and para- RPLS on-line renewa model for the quality control of rubber mixing. It was verified with experimental data that the mode could adaptively track the variation of mixing process, with the predictive error ±3 of Mooney viscosi ty,which met production requirement.