Multivariate Statistical Process Control Using Wavelet Approach

This paper attempts to monitor the mean vector of multivariate processes using a wavelet-based model. In the case of monitoring several related technical specifications, wavelet approach is an attractive contribution to analyze the performance of the multivariate process over time statistically. This advanced approach of signal processing enables more effective process monitoring compared to the traditional methods. The wavelet capability can lead practitioners to a root cause analysis sooner than the traditional schemes when the process shifts to an out-of-control condition. In this paper a new statistic named TMO-WAVE is proposed to analyze the variation of a multivariate process. The capability of the proposed scheme is compared numerically with different methods in this paper. The numerical comparative reports address high capability of the proposed wavelet-based method compared to the models in the literature in terms of average run length (ARL).