Ferreira Ponciano Ferraz Patrícia, Araújo E Silva Ferraz Gabriel, Leso Lorenzo, Klopčič Marija, Rossi Giuseppe, Barbari Matteo
Department of Agricultural Engineering, Federal University of Lavras (UFLA), Lavras, Minas Gerais 37200-900, Brazil.
Department of Agriculture, Food, Environment and Forestry, University of Florence, Via San Bonaventura, 13-50145 Florence, Italy.
Animals (Basel). 2020 Feb 22;10(2):351. doi: 10.3390/ani10020351.
The bedding materials used in dairy cow housing systems are extremely important for animal welfare and performance. A wide range of materials can be used as bedding for dairy cattle, but their physical properties must be analysed to evaluate their potential. In the present study, the physical properties of various bedding materials for dairy cattle were investigated, and different fuzzy clustering algorithms were employed to cluster these materials based on their physical properties. A total of 51 different bedding materials from various places in Europe were collected and tested. Physical analyses were carried out for the following parameters: bulk density (BD), water holding capacity (WHC), air-filled porosity (AFP), global density (GD), container capacity (CC), total effective porosity (TEP), saturated humidity (SH), humidity (H), and average particle size (APS). These data were analysed by principal components analysis (PCA) to reduce the amount of data and, subsequently, by fuzzy clustering analysis. Three clustering algorithms were tested: k-means (KM), fuzzy c-means (FCM) and Gustafson-Kessel (GK) algorithms. Furthermore, different numbers of clusters (2-8) were evaluated and subsequently compared using five validation indexes. The GK clustering algorithm with eight clusters fit better regarding the division of materials according to their properties. From this clustering analysis, it was possible to understand how the physical properties of the bedding materials may influence their behaviour. Among the materials that fit better as bedding materials for dairy cows, (Cluster 6) can be considered an alternative material.
奶牛养殖系统中使用的垫料对动物福利和生产性能极为重要。多种材料可作为奶牛的垫料，但必须分析其物理特性以评估其潜力。在本研究中，对各种奶牛垫料的物理特性进行了调查，并采用不同的模糊聚类算法根据其物理特性对这些材料进行聚类。总共收集并测试了来自欧洲各地的51种不同垫料。对以下参数进行了物理分析：堆积密度（BD）、持水能力（WHC）、充气孔隙率（AFP）、整体密度（GD）、容器容量（CC）、总有效孔隙率（TEP）、饱和湿度（SH）、湿度（H）和平均粒径（APS）。这些数据首先通过主成分分析（PCA）进行分析以减少数据量，随后进行模糊聚类分析。测试了三种聚类算法：k均值（KM）、模糊c均值（FCM）和古斯塔夫森-凯塞尔（GK）算法。此外，评估了不同数量的聚类（2 - 8个），随后使用五个验证指标进行比较。就根据材料特性进行划分而言，具有八个聚类的GK聚类算法拟合效果更好。通过这种聚类分析，可以了解垫料的物理特性如何影响其性能。在更适合作为奶牛垫料的材料中，（第6组）可被视为一种替代材料。
