09 3 matthias reinelt fast shelf life prediction for milk

1
13TH TAPPI EUROPEAN PLACE CONFERENCE Fast Shelf Life Prediction for Milk – Fast Selection of Right Packaging Contact person: Matthias Reinelt; [email protected] Presented by: Sven Sängerlaub, [email protected] MOTIVATION Avoid long-lasting storage tests Substitute slow tests by quick ones Reduce number of experiments Make quick decisions: right packaging, expected shelf life Accelerate implementation of recyclable packaging materials Allow decisions on every step of the production chain METHODS Expandable Coupled Differential Equations with Reactions Terms, Benchmarking, “Learning Models” Extensive Testing for Model Development Big Data Training of Neuronal Nets Machine Learning Complex dynamics of autoxidation Milk in cartons packagings Expected shelf life of UHT milk (based on development of rancidity) at different OTRs. Points: Experimental data, lines: Simulation Expected shelf life of ESL milk (rancidity) at different OTRs and storage temperature. Points: Experimental data, lines: Simulation; Chemical ageing of UHT-milk (1.5 % fat) stored at different temperatures, determined by chemometrical analysis of IR-spectra Correlation of sensory prediction (taste rancid) from a trained neural network versus. validation data. Quick test of quality parameters by AI-driven pattern recognition Maximum functional barrier overpackaging / underpackaging METHODS Shelf Life [days] PET bottle carton without aluminium Shelf Life [days] OTR of Packaging fat content Poster 9-3 Shelf Life of milk in different packages

Upload: others

Post on 03-Feb-2022

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: 09 3 Matthias Reinelt Fast Shelf Life Prediction for Milk

13TH TAPPI EUROPEAN PLACE CONFERENCE

Fast Shelf Life Prediction for Milk – Fast Selection of Right Packaging

Contact person: Matthias Reinelt; [email protected]

Presented by: Sven Sängerlaub, [email protected]

MOTIVATION

Avoid long-lasting storage tests

Substitute slow tests by quick ones

Reduce number of experiments

Make quick decisions: right packaging, expected shelf life

Accelerate implementation of recyclable packaging materials

Allow decisions on every step of the production chain

METHODS

Expandable Coupled Differential Equations with Reactions Terms, Benchmarking, “Learning Models”

Extensive Testing for Model Development

Big Data

Training of Neuronal Nets

Machine Learning

Complex dynamics of autoxidation

Milk in cartons packagings

Expected shelf life of UHT milk (based on development of rancidity) at different OTRs. Points: Experimental data, lines: Simulation

Expected shelf life of ESL milk (rancidity) at different OTRs and storage temperature. Points: Experimental data, lines: Simulation;

Chemical ageing of UHT-milk (1.5 % fat)stored at different temperatures,determined by chemometrical analysisof IR-spectra

Correlation ofsensory prediction(taste rancid) froma trained neuralnetwork versus.validation data.

Quick test ofquality parametersby AI-driven patternrecognition

Maximum functional barrier overpackaging / underpackaging

METHODS

Sh

elf

Lif

e [d

ays]

PET bottle

carton

without aluminium

Sh

elf

Lif

e [d

ays]

OTR of Packaging

fat content

Poster 9-3

Shelf Life of milk in different packages