Powered by FOSS
near infrared (NIR) analysis test samples for ANN calibrations

European Grain Network study supports ANN calibration in NIR analysis

White paper from FOSS outlines the results of a five-year ring test study that was conducted with members of the European Grain Network.

The European Grain Network (EGN) has been collaborating on research for more than 15 years. For this particular study, the group wanted to investigate the differences between master reference laboratories in order to make any necessary calibration adjustments and inform the members.

The validation uses samples from the actual harvest in different countries and applies the reference methods valid in each of the countries. In the last five years of annual ring tests, 24 organisations from 17 countries participated.

ANN calibration for wheat and barely

The ring test consists of roughly ten samples each of wheat and barley. The samples come from different origins. When enough samples have been collected from different networks, they are cleaned, homogenised, divided and shipped to the participants for analysis. Ten to 20 networks participate in the ring test each year, which provides a reliable foundation for the results.

The results from the 2011 ring test showed that the reproducibility of the ANN method is better than the reference method for protein and nearly the same for moisture.

The robustness of the ANN calibration manifests in its long term stability. The key to the superior stability is two-fold: the extremely large calibration database covering virtually all possible variations and the ANN technology that captures all variations in an optimum way.

Conclusion: high performance ANN prediction model

The high performance of the ANN prediction model is demonstrated by results from the five-year European study. The ANN prediction model is approved as a European standard for the simultaneous prediction of protein and moisture content in whole grain wheat and barley. The stability of the model also holds great promise as a platform for effective grain analysis in the future.

Download the white paper.

0 comments