Abstract
The double exponentially weighted moving average (dEWMA) controller is a popular run-to-run controller for a drifted process. It uses the information collected from past process data to adjust the process parameters for the later runs. The dEWMA controller can guarantee long-term stability under suitably fixed discount factors and fairly regular conditions. However, it usually requires a large number of runs to bring the process output to meet its target and, thus, may leave the output out of its specification at the beginning of the first few runs. Hence, dEWMA controller is not suitable for the processes with short production runs. To reduce the possibility of high rework rate, we propose an enhanced dEWMA controller, which we refer to as the initial intercept iteratively adjusted (IIIA) controller, to further eliminate the off-target (nonrandom biases) of the process output. We derive an analytic expression of the process output of the III A controller, and present several examples to show that the IIIA reduces the process mean square error significantly, as well as the rework rate. © 2005 IEEE.