Skip to content

Glass Worldwide - Batch Tracking With Machine Learning and Deep Neural Networks

水曜日, 10月 13, 2021

Read our latest article in Glass Worldwide where Matt Gillott explains how AMETEK Land is focusing on better monitoring for batch tracking by implementing image processing with neural networks.

Glass Worldwide - September/October 2021"To provide a glass product of uniform quality, it is vital to ensure a consistent melt of the batch. This requires monitoring of the furnace to prevent issues such as batch entering the refining zone. However, getting a good picture of the processes inside the furnace can be difficult – flames can obscure the view, cameras can overheat, and making sense of the image can be a problem for untrained operators. To achieve a clearer image of the inner workings of a glass melt furnace, AMETEK Land is implementing cutting-edge machine learning and deep neural network technology, working alongside its high-resolution infrared borescope cameras.

AMETEK Land’s IMAGEPro-Glass advanced imaging processing software, combined with high-resolution near infrared borescope thermal imagers for verification and thermal profiling, could easily support operators in adjusting the batch and foaming line identifying the best batch pattern to avoid glass defect and get the proper pull rates. It provides a top 2D view of the melting zone, based on a grid analysis where the critical batch-free location in front of the furnace is monitored. The current model for batch coverage in IMAGEPro-Glass runs well, providing a grid with batch analysis of the melting zone for the users. The next step is to focus on better monitoring for batch tracking, implementing image processing with neuronal networks."

Read the full article in Glass Worldwide September / October 2021