Compared with the greatest possible way Bizet methods to estimate the failure rate of the existence of surveillance data from the first type of samples failure

Abstract

AbstractThe – Chart and R-Chart are based on assumption that the Sample drawn from the process are normally distributed. When the normality assumption is not valid , there is several different actions , such that , increasing the size of Samples drawn from the process until the distribution of sample mean is considered normal , using Box- Cox power transformation on the original data to yield an approximate normal distribution, or use Skewness correction (SC). In this paper the Simulation is used to compare artificial neural network (ANN) technique for monitoring the process variability and Classical method of Statistical process Control Chart ( – chart and R-chart) . We conclude that when the process distribution is normally distributed or in Some neighbor- hood weibull , Gamma and Lognormal ,Simulation Show that ANN technique have type I error is less than Type I error of - Chart and R- Chart. There for ANN is more efficient than - Chart and R- Chart for symmetric and non-symmetric distributions at different skewed process distribution .