关键词: 稳态自回归滑动平均控制图, AR(p)模型, 状态空间模型, 稳态卡尔曼滤波器, Grubbs调和规则, 自相关过程

Abstract: To solve two problems of the auto-correlated phenomenon in the process output and the process mean was usually off-target due to the frequent adjustment of machine at the setup phase existed in multi-products and low volume production mode,a new integrated method with the modified steady-state Auto Regressive and Moving Average (ARMAST) chart as monitoring technology and Grubbs'harmonic rule based on steady-state Kalman filter as adjusting technology was proposed.Aiming at the problem that the steady-state variance of ARMAST chart was hard to solve under high-order ARMA model,the solving method on generally arbitrary order stationary AR model was established to expand the applicable scope of steady-state variance.For the low efficiency of ARMAST chart parameters'design method,the new method was designed based on conditional probability to monitor the efficiency better.After the early warning of charts,the production process should be adjusted according to Grubbs'harmonic rule.The simulation experiments and case studies were conducted to demonstrate the validity of these methods.

Key words: steady-state auto regressive and moving average chart, AR(p) model, state-space model, steady-state Kalman filter, Grubbs'harmonic rule, autocorrelated processes