Acta Agriculturae Scandinavica Section A – Animal Science · 1993 · 29 citations · 11 references
ComputingDairy ScienceForecastingMilk QualityTime Series AnalysisBayesian FilteringKalman FilteringMilk ProductionSomatic Cell CountState-space ModelsTime Series ForecastingBulk Tank Somatic Cell CountDairy Herd MonitoringFilteringHealth Status MonitoringMilk Quality ControlSmoothingSomatic Cell Count AssessmentSomatic Cell Count Estimation
Abstract High amounts of somatic cells in bulk tank milk from dairy herds indicate poor milk quality and low health status. Bulk tank somatic cell counts (BTSCC) are therefore regularly used to encourage good milk quality through price penalties and to detect mastitis problems in the dairy herd. The use of BTSCC for monitoring is complicated by variation and outliers in the data that may mask the changes that should be detected. The Kalman filter, a recursive algorithm for making short-term predictions, is suggested to reduce the noise in BTSCC data. A multi-process Kalman filter with three components applying to (1) normal variation in the data, (2) outliers and (3) abrupt changes in the level of the data is described with updating equations and an example. Retrospective smoothing of the data is also described. A graphical presentation with the smoothed level of historical BTSCC data and forecasted future BTSCC values, both with confidence intervals, is suggested for monitoring purposes. Key Words: dairy herdmilk qualityshort-term predictionsKalman filter
11