Early warnings from automatic milk yield monitoring with online synergistic control

Abstract

Sensors play a crucial role in the future of dairy farming. Modern dairy farms today are equipped with many different sensors for milk yield, body weight, activity, and even milk composition. The challenge, however, is to translate signals from these sensors into relevant information for the farmer. Because the measured values for an individual cow show nonstationary behavior, the concepts of statistical process control, which are commonly used in industry, cannot be used directly. The synergistic control concept overcomes this problem by on-line (real-time) modeling of the process and application of statistical process control to the residuals between the measured and modeled values. In this study, the synergistic control concept was developed and tested for early detection of anomalies in dairy cows based on detection of shifts in milk yield. Compared with the combination of visual observation and milk conductivity measurements, the developed strategy had a sensitivity of 63% for detecting clinical mastitis. Consequently, this technique could have added value on many farms, as it extracts practical information out of inexpensive data that are already available. As it can be easily extended to other measured parameters, the technique shows potential for early detection of other nutrition and health problems.

Exploring milk shipment data for their potential for disease monitoring and for assessing resilience in dairy farms

“…It would provide an opportunity to detect and possibly mitigate the consequences of disease outbreaks at herd and regional levels at an early stage. The possibilities of syndromic surveilance in order to detect viral diseases such as Bluetongue and Schmallenberg in dairy herds has previously been described using milk recording data (Madouasse et al, 2013;Veldhuis et al, 2016), while milk robot data has been used for early detection of mastitis (Huybrechts et al, 2014), but to the knowledge of the authors exploring the possibilities of milk delivery data has not yet been done. Early detection of outbreaks with endemic viruses, such as Bovine respiratory syncytial virus and Bovine Corona virus, would be desirable in order to reduce the impact of, and to control disease outbreaks.…”

Bayesian integration of sensor information and a multivariate dynamic linear model for prediction of dairy cow mastitis

“…Similar adaptive forecasting has been applied by Huybrechts et al (2014), who used a synergistic control process to adjust lactation curves in an effort to use milk yield as a predictor of clinical mastitis (sensitivity: 0.63). Huybrechts et al (2014) relied heavily on a specific mathematical model for long-term forecasting, whereas the adaptive and short-term nature of the forecasts produced by a DLM allows for a freer description of multiple (non)linear trends that may predict the short-term observations better.…”

“…Huybrechts et al (2014) relied heavily on a specific mathematical model for long-term forecasting, whereas the adaptive and short-term nature of the forecasts produced by a DLM allows for a freer description of multiple (non)linear trends that may predict the short-term observations better. Furthermore, the DLM easily handles missing data because one-step-ahead forecasts are always produced given the available data.…”

Production trend identification and forecast for shop-floor business intelligence

“…The out-of-control observations can be determined precisely with the Dijkstra model and cumulative sum chart of the corrected residuals between the measured and predicted values. Milk yield data from two Automatic Milking System (AMS) farms and one farm with a conventional milking system were used for the case study [16].…”