Using a digital twin for sensitivity analysis to determine sensor placement in a roll-to-roll manufacturing web-line

This article discusses how to determine the optimal number, type, and placement of sensors in a manufacturing line to maximize the accuracy of a digital twin, using a roll-to-roll manufacturing process as an example.

Specifically, the article covers:

• The challenge of balancing cost and accuracy when deciding where to place sensors
• Using sensitivity analysis, particularly Shapley sensitivity analysis, to identify which variables have the greatest impact on desired outputs
• Leveraging MapleSim digital twin models and Monte Carlo simulations to analyze sensitivities and probability distributions
• Optimizing sensor selection by substituting cheaper RPM sensors where possible
• The AWS architecture and tools (TwinFlow, AWS Batch, etc.) used to run scalable sensitivity analyses
• Exploration of using generative AI (Amazon Bedrock) to augment traditional sensitivity analysis methods