Batch Weighing Methodology Using Machine Learning & Monte Carlo Simulation (2026-US-30MP-2982)

Batch weighing is a common practice in the manufacture, research, development, and handling of product. Counting individual parts can be a time-consuming and inefficient process, and the ability to batch weigh can save time and money. The main downside of batch weighing is the potential risk of error in the estimated quantity due to tolerance and noise stacking.

The methodology highlighted in this presentation aims to directly address and alleviate this risk by quantifying it using Monte Carlo simulation and linear discriminant analysis, a supervised machine learning modeling approach. The final model can be used to inform the user of the specific risk associated with each batch based on weight and reduce the potential for misclassification. The presentation also discusses guidelines for applying the methodology and remedial methods for certain issues that may arise during its application, using a case study to help illustrate the method’s benefits.