03/01/2024

Improving the prediction of production times through machine learning (AI)

intelliDivide Cutting

Prediction accuracy of the model based on machine learning. The deviation of the prediction from the actual time required is shown in seconds per cutting pattern.

Prediction accuracy of the simulation model used to date. The deviation of the prediction from the actual time required is shown in seconds per cutting pattern.

Previously, the prediction of production time in intelliDivide Cutting was based on a simulation model that could not take sufficient account of the actual conditions on site. As a result, deviations between the prediction and the actual time required could occur.

To improve the predictions, we have now implemented a new machine learning-based model that uses anonymized feedback data from saws connected to tapio. This data is continuously fed into the model to base the prediction of newly calculated cutting patterns on actual, real-world values.

Saws that have already cut over 500 cutting patterns and reported back via tapio benefit from even more accurate results by using individual feedback data to predict production time.

These improvements enable better production planning and more efficient use of resources.