Monitoring Milking Process Variation Using Model Driven Multivariate Control Charts (MDMVCC)

The main aim of this paper is to define if the milking process is out or in control in different milking parlours (herringbone, parallel and rotary). Therefore, the whole year data from the mentioned milking parlours equipped with herd management software (AfiFarm v. 3.07, Afimilk, Afikim, Israel), were analyzed by using Model Driven Multivariate Control Charts (MDMVCC). In the herringbone milking parlour the average milk yield/cow was the variable with the highest contribution rate in 17 of 31 out-of-control points. In the parallel and rotary milking parlour the multiple cluster attachment was the variable with the highest contribution rate in 12 of 21 and 11 of 22 out-of-control points, respectively. Additionally, values of the milk produced/stall per hour were significantly higher both in herringbone and parallel parlours when the milking process was in control compared to the situation when it was out of control. The same was found for the rotary parlour about the milk produced/h. In all three milking parlours we found a statistically significant higher milk production (p < 0.01) when the system was in control than out-of-control. The MDMVCCs are a potentially useful tool to detect milking process variation in each of the three investigated milking parlours.