Monitoring livestock production processes by means of statistical control charts can provide an importantsupport for management. The non-stationary and autocorrelated characteristics of most data originatingfrom such processes impede the direct introduction of these data into control charts. To deal withthese characteristics Engineering Process Control strategies can be applied. Stationarity was achieved bymodelling and subtracting the time dependent trend using a non-linear model. Next, the autocorrelationstructure in the residual data is modelled and corrected for by means of an ARMA model. The resultingcorrected stationary and independent residuals are then inserted in the traditional cusum controlscheme. This combined use of Engineering Process Control strategies for modelling the unconventionalstatistical characteristics and Statistical Process Control strategies for constructing the control chart basedon the resulting pre-processed data, is referred to as a Synergistic Control strategy. The developed cusumcontrol chart was tested on data of two layer flocks. In both cases the control chart provided alarms forimportant problems in production and furthermore signalled problems that remained unnoticed by thelayer managers. The amount of false alarms was acceptable. With this control scheme and the schemeof the average egg weight, control procedures for two important output parameters of the productionprocess of consumption eggs are available. Furthermore, this strategy could provide a possible solutionfor other process parameters that also display non-stationarity and autocorrelation.
展开▼
