Animal and Grassland Research and Innovation Centre, Teagasc Moorepark Fermoy, Co. Cork, Ireland P61 C996; Department of Animal and Dairy Sciences, University of Wisconsin-Madison, Madison, 53706.
Animal and Grassland Research and Innovation Centre, Teagasc Moorepark Fermoy, Co. Cork, Ireland P61 C996.
J Dairy Sci. 2022 May;105(5):4156-4170. doi: 10.3168/jds.2021-20464. Epub 2022 Mar 2.
The aims of this research were (1) to develop a model to simulate a herd of cows and quarter milk flowrates for a milking and derive quarter and udder milking durations and box duration (i.e., the time a cow spends inside the robot) for a group of cows milked with an automatic milking system (AMS); (2) to validate the simulation by comparing the model outcomes with empirical data from a commercial AMS dairy farm; and (3) to apply teatcup removal settings to the simulation to predict their effect on quarter and cow milking duration and box duration in an AMS. For model development, a data set from an AMS farm with 32 robots milking over 1,500 cows was used to fit the parameters to the variables days in milk, parity, and milking interval, which were subsequently used to create a herd of cows. A second data set from 2019 from an AMS farm with 1 robot milking 60 cows that contained quarter milk flowrates (at 2-s intervals) was used to extract the parameters necessary to simulate quarter milk flowrates for a milking. We simulated a herd of cows, and each was assigned a parity, days in milk, milking interval, and milk production rate. We also simulated milk flowrates every 1 s for each quarter of each cow. We estimated quarter milking duration as the total time that flowrate was greater than 0.1 kg/min after a minimum of 1 min of milk flow. We incorporated a randomly sampled attachment time for each quarter and calculated cow milking duration as the time from the first quarter attached to the last quarter detached. We included a randomly sampled preparation time which, added to cow milking duration, represented box duration. For simulation application, we tested the effect of quarter teatcup removal settings on quarter and cow milking duration. The settings were based on absolute flowrate (0.2, 0.4, and 0.6 kg/min) or a percentage of the quarter's 30-s rolling average milk flowrate (20, 30, and 50%). We simulated over 84,000 quarter milkings and found that quarter milking duration (average 212 s) had a mean absolute percent error (MAPE) of 7.5% when compared with actual data. Simulated cow milking duration (average 415 s) had a MAPE of 8%, and box duration (average 510 s) had a MAPE of 12%. From simulation application, we determined that quarter milking duration and box duration were reduced by 19% (209 vs. 170 s) and 6.5% (512 vs. 479 s), respectively, when increasing the teatcup removal flowrate from 0.2 to 0.6 kg/min. Quarter milking duration and box duration were 7% (259 vs. 241 s) and 3% (590 vs. 573 s) longer respectively by using a teatcup removal setting of 20% of the quarter's rolling average milk flowrate, compared with 30%. Both results agree with previous research. This simulation model is useful for predicting quarter and cow milking and box duration in a group of cows and to analyze the effect of milking management practices on milking efficiency.
（1）开发一个模型，模拟牛群，并为挤奶导出每个乳区的奶流量，以及为一组使用自动化挤奶系统（AMS）挤奶的奶牛导出挤奶时间和牛舍时间（即奶牛在机器人内的时间）；（2）通过将模型结果与商业 AMS 奶牛场的经验数据进行比较来验证模拟；（3）将奶嘴移除设置应用于模拟，以预测它们对 AMS 中每个乳区和奶牛挤奶时间和牛舍时间的影响。为了进行模型开发，使用了来自一个具有 32 个机器人、为 1500 多头奶牛挤奶的 AMS 农场的数据集来拟合变量天数、胎次和挤奶间隔的参数，随后这些参数被用来创建一个牛群。第二个数据集来自 2019 年一个具有 1 个机器人、为 60 头奶牛挤奶的 AMS 农场，该数据集包含每个乳区的奶流量（每隔 2 秒记录一次），用于提取模拟每个乳区奶流量所需的参数。我们模拟了一个牛群，并为每头奶牛分配了胎次、天数、挤奶间隔和产奶率。我们还模拟了每头奶牛每个乳区每 1 秒的奶流量。我们将奶流量大于 0.1kg/min 的总时间估计为每个乳区的挤奶时间，前提是奶流量至少持续 1 分钟。我们为每个乳区随机抽样了一个连接时间，并将奶牛的挤奶时间计算为从第一乳区连接到最后一个乳区断开的时间。我们包括了一个随机抽样的准备时间，将其添加到奶牛的挤奶时间中，代表牛舍时间。在模拟应用中，我们测试了奶嘴移除设置对每个乳区和奶牛挤奶时间的影响。这些设置基于绝对流量（0.2、0.4 和 0.6kg/min）或每个乳区 30 秒滚动平均值奶流量的百分比（20、30 和 50%）。我们模拟了超过 84000 次的每个乳区挤奶，发现与实际数据相比，每个乳区的挤奶时间（平均 212 秒）的平均绝对百分比误差（MAPE）为 7.5%。模拟的奶牛挤奶时间（平均 415 秒）的 MAPE 为 8%，牛舍时间（平均 510 秒）的 MAPE 为 12%。从模拟应用中，我们确定当将奶嘴移除流量从 0.2 增加到 0.6kg/min 时，每个乳区的挤奶时间和牛舍时间分别减少了 19%（209 秒对 170 秒）和 6.5%（512 秒对 479 秒）。当使用奶嘴移除设置为每个乳区滚动平均奶流量的 20%时，每个乳区的挤奶时间和牛舍时间分别比使用 30%的奶嘴移除设置长 7%（259 秒对 241 秒）和 3%（590 秒对 573 秒）。这两个结果都与之前的研究一致。该模拟模型可用于预测一组奶牛的每个乳区和奶牛挤奶时间和牛舍时间，并分析挤奶管理实践对挤奶效率的影响。
