LAUSR

Abstract This work proposes a method for the statistical monitoring of the weekly weight of shrimp in a fish farm. The design of control charts for shrimp growth presents several challenges, the main ones being the presence of trend and autocorrelation. Several control charts are proposed, the most efficient being a chart based on the residuals of a second order autoregression whose parameters vary with the week. A control chart with less computational complexity is also provided. It is simply based on growth rates and works well at detecting negative weight changes. Real data from a large shrimp farm has been used for both the design and the evaluation.