Machine Learning-Based Quality Improvement for Thermal Energy Cutting Processes

Quality control in manufacturing processes is a challenge for many industries. Historically, quality control has relied on time-consuming manual or semi-manual approaches, where large amounts of valuable data is lost. In the case of thermal energy cutting processes, they outperform mechanical cutting processes. However, there are defects that require post finishing in the process. These secondary operations can add significant time to the overall process and impact the overall production throughput.

The team will develop and validate a digital tool that collects process and quality data, and uses the data directly to provide process control feedback to the operator and process planner in assisting part quality control for a manufacturing process. The proposed solution is to use sensor technology with AI/machine learning algorithms that will incrementally train a predictive model that can be consistently updated to predict the quality measures according to the input process parameters. In addition, the team is collecting data from various types of materials and thicknesses to build a database.

This proposed work offers a low-cost digital solution to close the quality control loop for manufacturing processes. By leveraging advancement in sensor technologies and advanced data analytic methods, significant costs can be saved through reducing the requirement of rework and defects. Consequentially, production throughput can be increased, and product quality can be better managed. This can contribute to MxD’s core mission to enhance US competitiveness through transforming American Manufacturing to digital manufacturing.