Multivariate Nonparametric Control Chart Based On Covariance Matrix Test

In this paper,we propose a new multivariate nonparametric control chart for monitoring the covariance matrix,which combines a test for testing the equality of two covariance matrices with the traditional exponentially weighted moving average control chart.The proposed control chart requires only a small number of historical samples.It is distribution-free and easy to design and implement.We compare the proposed method with the existing MEWMC control chart.Numerical simulation shows that the proposed control chart is superior to the MEWMC control chart.Finally,We present the proposed control chart by using a real data from a white wine production process.