Joint Research Conference
June 24-26, 2014
Process monitoring using multivariate observations remains an important and challenging problem in statistical process control (SPC). Although the multivariate SPC has been extensively studied in the literature, the challenges associated with designing distribution-free control schemes are yet to be addressed well. In this talk, I'll introduce a new nonparametric methodology for monitoring location parameters when a sufficiently large reference dataset is unavailable. The key idea is to construct a series of conditionally distribution-free test statistics in the sense that their distributions are free of the underlying distribution given the empirical distribution functions. The success of the proposed method lies in the use of data-dependent control limits, which are determined based on the observations on-line rather than decided before monitoring. The conditional probability that the charting statistic exceeds the control limit at present given that there is no alarm before the current time point can be guaranteed to attain a specified false alarm rate. Our theoretical and numerical studies show that the proposed control chart is able to deliver satisfactory run-length performance for any distributions with any dimension. It is also very efficient in detecting multivariate process shifts which occur when the process distribution is heavy-tailed or skewed.