FreshTwin
Determining the freshness and shelf life of fruit along the supply chain through digital twins and an EPCIS 2.0 repository.
The project
Fresh fruit loses quality continuously on its way from the producer to retail, and how much shelf life is left in a batch has so far been hard to determine reliably. Inspection at goods receipt is mostly manual and turns out differently depending on who inspects.
FreshTwin combined innovative measurement methods with artificial intelligence to make the actual quality characteristics of fresh food measurable and predictable. Hyperspectral cameras, a handheld spectrometer and image recognition capture the internal and external characteristics of a piece of fruit. That data produces a digital twin for each product, working with hybrid grey box models, meaning a combination of established process knowledge and machine learning.
The approach was trialled on grapes, apples and mangoes, in laboratory tests and in practice at fruit producers and in retail. The goal was a more accurate assessment of quality and shelf life, so that producers and traders can make better decisions and less produce is lost unnecessarily.
The contribution from benelog
In the project we were responsible for the data repository that brings together the results of the measurements, the models and the supply chain data.
A digital twin is only as good as the data that flows into it, and that data arises in many places: measurements from the analytics partners, product-accompanying information and events along the supply chain. For a coherent picture to emerge from it, a shared foundation that conforms to the standard is needed.
For that we provided a modular data repository conforming to GS1 EPCIS 2.0, with OpenEPCIS as the central hub. Through EPCIS Capture and subscriptions, measurement and supply chain data flowed in both directions, so the digital twins could be supplied with new data continuously and their results fed back. Together with the partners we also worked out the target architecture of the solution and took part in validating it in practice.
