The fat globule size distribution of raw milk and milk with an increasing degree of homogenization was estimated based on their bulk light scattering properties in the Vis/NIR wavelength range. The particle size distribution (PSD) was approximated as a lognormal distribution, of which the parameters were estimated simultaneously with the fat concentration. This resulted in a good agreement between the estimated PSDs and the reference PSDs obtained by laser diffraction in case of raw and strongly homogenized samples. The accuracy increased if a known fat concentration was incorporated, or when the scattering coefficient and anisotropy factor spectra were used simultaneously as input. For mildly homogenized samples, the lognormal distribution was unable to fit the bimodal PSD correctly and focused on the largest fat globules. In this case, the estimated PSDs provided still relatively accurate information on D90, D32 and the right distribution tail, which contains the largest fat globules.

Industrial relevance

The presented estimation method demonstrates the potential of bulk scattering spectra for determining the PSD and concentration of scattering particles in turbid media. Further development of this technology can lead to new solutions for spectroscopic PSD determination, allowing on-line monitoring systems for a wide range of food and non-food products.

中文翻译：

---

牛奶均质化监控：根据牛奶的散射光谱估算脂肪球大小

根据其在Vis / NIR波长范围内的大量光散射特性，可以估算出生奶和均质程度提高的牛奶的脂肪球尺寸分布。粒度分布（PSD）近似为对数正态分布，其参数与脂肪浓度同时估算。在原始样品和强烈均质样品的情况下，这导致估计的PSD与通过激光衍射获得的参考PSD之间有很好的一致性。如果掺入了已知的脂肪浓度，或者同时使用了散射系数和各向异性因子光谱作为输入，则准确性会提高。对于轻度均质的样品，对数正态分布无法正确拟合双峰PSD，而无法聚焦于最大的脂肪小球。在这种情况下，

行业相关性

提出的估算方法证明了散装散射光谱在确定PSD和混浊介质中散射颗粒浓度方面的潜力。这项技术的进一步发展可以为光谱PSD测定带来新的解决方案，从而允许对各种食品和非食品产品使用在线监控系统。