Predictive modelling of instant whole milk powder functional performance across three industrial plants

Depree, Nick; Prince-Pike, Arrian; Young, Brent R.; Wilson, D.I.

doi:10.1016/j.jfoodeng.2019.01.011

Cited by 7 publications

(1 citation statement)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Contrary to this, PLS ban be more easily interpreted to reflect fundamental changes that could be understood by operators and process engineers. Therefore, PLS is recommended for use in the current industry [36].…”

Section: Artificial Neural Network Modelsmentioning

confidence: 99%

Effects of Morphology on the Bulk Density of Instant Whole Milk Powder

Ding

Boiarkina

et al. 2020

Foods

View full text Add to dashboard Cite

The chemical and physical properties of instant whole milk powder (IWMP), such as morphology, protein content, and particle size, can affect its functionality and performance. Bulk density, which directly determines the packing cost and transportation cost of milk powder, is one of the most important functional properties of IWMP, and it is mainly affected by physical properties, e.g., morphology and particle size. This work quantified the relationship between morphology and bulk density of IWMP and developed a predictive model of bulk density for IWMP. To obtain milk powder samples with different particle size fractions, IWMP samples of four different brands were sieved into three different particle size range groups, before using the simplex-centroid design (SCD) method to remix the milk powder samples. The bulk densities of these remixed milk powder samples were then measured by tap testing, and the particles’ shape factors were extracted by light microscopy and image processing. The number of variables was decreased by principal component analysis and partial least squares models and artificial neural network models were built to predict the bulk density of IWMP. It was found that different brands of IWMP have different morphology, and the bulk density trends versus the shape factor changes were similar for the different particle size range groups. Finally, prediction models for bulk density were developed by using the shape factors and particle size range fractions of the IWMP samples. The good results of these models proved that predicting the bulk density of IWMP by using shape factors and particle size range fractions is achievable and could be used as a model for online model-based process monitoring.

show abstract

Section: Artificial Neural Network Modelsmentioning

confidence: 99%