Accuracy of a real-time location system in static positions under practical conditions: Prospects to track group-housed sows
Accuracy of a real-time location system in static positions under practical conditions: Prospects to track group-housed sows
复制标题
DOI：
10.1016/j.compag.2017.09.020
复制
发表时间：
2017-11
期刊：
影响因子：
--
通讯作者：
Maike K. Will;K. Büttner;Tobias Kaufholz;C. Müller-Graf;T. Selhorst;J. Krieter
中科院分区：
文献类型：
--
作者：
Maike K. Will;K. Büttner;Tobias Kaufholz;C. Müller-Graf;T. Selhorst;J. Krieter
动物之间的社会互动可能会影响疾病的传播途径。因此，实时定位系统的使用对畜牧场和研究机构变得越来越重要，因为这项技术有助于同时获得大量动物的位置并自动对它们进行评估。因此，该项目的目的是具体说明实时定位系统在实际条件下关于未来可能的应用的准确性。在实践中，耳标已经证明了它们的价值，因为猪操纵并因此破坏了长期应用于它们的其他物品。因此，使用实时定位系统来提供集成在耳标中的发送单元。耳标是在母猪的妊娠栏上以静态姿势测试的。每个静态位置进行5分钟的测量，而数据每秒传输一次(1赫兹)，这导致每个位置有300个数据点。金属笔设备导致位置丢失或有噪音。平均而言，9%的数据丢失发生并被插入以进行后续数据评估。利用Haar小波对图像进行降噪处理。在由欧几里得误差和滤波信号的方差误差组成的误差大小的帮助下，对滤波器设置进行评级。当只保留X轴的29个最大系数和Y轴的20个最大系数，而其他所有系数都设置为0时，可以实现最优的滤波器设置。此外，对所有耳标的平均系数数和每个单个耳标的最优个别过滤是否导致显著不同的过滤结果进行了at-test检验。T检验的P值分别为0.15(X坐标)和0.18(Y坐标)，因此不显著。因此，可以将平均过滤器设置应用于所有耳标。用欧几里得距离描述的测量数据的精度中值在过滤前为2.7亿米，过滤后提高到2.0亿米。考虑到这一系统调查的结果，它表明该系统可能有助于后续关于动物行为、运动特征或社会网络的研究，以揭示疾病可能的传播途径。
Social interaction between animals may influence disease transmission paths. Therefore, the usage of real-time location systems gains in importance for livestock farms and research institutes as this technology helps to simultaneously obtain positions of a large number of animals and to evaluate them automatically. Thus, the aim of the project was to specify the accuracy of the real-time location system under practical conditions with regard to a possible future application. In practice, ear tags have proven their worth because pigs manipulate and therefore destroy other objects applied to them in the long term. Therefore, a real-time location system was used providing the sending unit integrated in an ear tag. Ear tags were tested in a sows’ gestation stall in static positions. Measuring took place for 5 min per static position, whereas data was transmitted once per second (1 Hz) which led to 300 data points per position. Metal pen equipment led to lost or noisy positions. On average, 9% of data losses occurred and were inserted for the following data evaluation. A Haar wavelet was applied to reduce the noise. Filter settings were rated with the help of an error size consisting of the Euclidean error and an error for the variance of the filtered signal. An optimal filter setting could be achieved when only the 29 largest coefficients for the X axis and 20 largest coefficients for the Y axis were kept while all others were set to 0. Additionally, at-test was performed to test whether an averaged number of coefficients over all ear tags and an optimal individual filtering of each single ear tag resulted in a significantly different filter result. P-values of thet-test were 0.15 (X coordinate) and 0.18 (Y coordinate) and therefore not significant. Thus, an averaged filter setting can be applied to all ear tags. The median accuracy of measured data described as Euclidean distance was 2.7 m before filtering and improved to 2.0 m after filtering. Considering the results of this system investigation, it shows that the system may be helpful for ensuing studies regarding e.g. animal behaviour, movement profiles, or social networks to uncover possible transmission paths for diseases.