针对趋势失控的基于TAR模型的SPC-EPC集成研究

目前在SPC-EPC集成研究中使用线性时间序列模型,该模型对复杂的非线性自相关关系的描述存在偏差并影响最终的控制效果。针对这一问题提出使用一类非线性时间序列模型(即门限自回归模型)描述系统的动态噪声,据此建立基于门限自回归模型的最小均方误差控制器,并进一步建立SPC-EPC集成控制体系。针对在连续生产过程中常见的以趋势形式存在的过程失控,通过实例研究控制器在单独使用和集成控制方法下的控制效果,并与线性控制器相应的结果进行对比,通过模拟研究进一步验证和分析这一集成控制方法的控制效果。研究结果表明,基于非线性时间序列的集成SPC-EPC控制方法,对含趋势形式失控的、复杂的非线性自相关过程,在单独使用控制器调整的基础上可以进一步减小过程变异,模拟研究进一步验证此集成控制方法对不同幅度趋势变异的有效性,并给出不同情况下的集成控制方案。

An Integrated SPC-EPC Study for Checking Assignable Causes Resulting in Trend Based on TAR Model

The linear time series models applied in traditional integrated SPC-EPC study sometimes cause error in modeling complex nonlinear autocorrelationships and affect the final control result.This study presented a method using a kind of nonlinear time series model,that is the threshold autoregressive model(TAR) to describe the dynamic noise of the system,and built an minimum mean square error(MMSE) controller based on this model and further built an integrated SPC-EPC control system.Aimed at checking assignable causes resulting in trend which is common in continuous manufacturing processes,the control result of this controller and integrated control method was analyzed first by examples and also compared with the result of linear controller.Next the result of this integrated method was verified and analyzed further by simulations.The final results indicate that the integrated SPC-EPC method based on nonlinear time series model can reduce more variation in controlling complex nonlinear systems which have assignable causes resulting in trend than that using controller alone.Simulation study verifies that the integrated control method is effective in controlling different sizes of trend,and also gives the integrated control schemes in different situations.