S. A. Mulema, A. Carrión García
Quality characteristics describing products or processes are usually correlated and autocorrelated, and require a multivariate approach for monitoring and controlling their performance. In this paper, we analyse this situation, developing a multivariate statistic model to monitor this type of processes. To achieve this, a MEWMA chart was used and adapted. The autocorrelation structure in the data was adjusted using an autoregressive model. With this model, our proposed MEWMA chart was built. Chart performance was analysed through a simulation of a bi-variate cross-correlated autoregressive time series, which was adjusted to the MEWMA chart. Its efficiency was compared with a MEWMA chart adjusted to the residuals of the series, using the ARL statistic. The results show that the proposed MEWMA chart is faster in detecting shifts in the process mean than the other methodology, and the larger the autoregressive coefficient, the faster the difference
Keywords: MEWMA, control chart, autocorrelation, AR model
Scheduled
PO-1 Poster Session
September 4, 2019 10:40 AM
Multifunctional room. Carbonell building