Applications of exponential smoothing

Abstract

This dissertation deals with the applications of exponential smoothing to control charts with some consideration given to applications to operational control systems. In order to gain an intuitive feel for the behavior of the mean, variance, fraction defective and number of defects exponentially smoothed control charts, simulations models, which allow for different types of process changes and include plot routines, were developed. These simulation models also compute and plot the Shewhart control charts. The exponential control charts, as well as the run sum and cumulative sum control charts, have appreciably different operating characteristics than the Shewhart control charts. The difference in operating characteristics is examined through the use of sensitivity characteristic curves which show the probability of acceptance after a process change as a function of the sample period. To determine what savings are possible through the use f exponential control charts, several trade off analyses were performed. These analyses show that the use of exponential charts can yield appreciable savings in sample sizes or faster detection of a change in process mean. Two other topics that are treated are adaptive control charts and operational control systems. For the adaptive control chart case, adaptive exponentially smoothed forecasting techniques are modified for use in control charts. For the operational control system case, exponential smoothing techniques for individual data points are utilized in the information processing element of an operational control system.