An optimal bivariate Poisson field chart for controlling high-quality manufacturing processes

Shewhart C-chart is a widely accepted control chart for monitoring number of defects in a given process. This chart is based on normal approximations. Normal assumption is, however, impractical in many cases especially for count data. This assumption becomes stronger when correlation between characteristics exists. In this article, we propose an optimal bivariate Poisson field chart to monitor two correlated characteristics of count data for both industrial and non-industrial purposes. Our chart is based on optimization of bivariate Poisson confidence interval and projection of bivariate Poisson data in Poisson field. The performance of our proposed algorithm is presented using both real case study and simulations. The experimental results demonstrate improved performances regarding visualization and false alarm rate.