Industry 4.0 research projects

Internet of Things (IoT) for predicting defects in mass-produced manufactured parts

The aim of this project is to develop a software prototype to support Industry 4.0 quality managers using data from 3D measuring machines (measured/scanned) collected on manufactured parts, and by using an AI algorithm to categorize the control card patterns of each part containing the ranges of customer quality requirements.

What we do

We study tool data to measure and control the quality of manufactured products. Manufacturers and production managers need a link between their coordinate-measuring machines (CMMs) and a statistical process control (SPC) system to control and predict production quality. CMM is a coordinate-measuring machine that helps companies increase quality assurance and inspection capacity. It is a 3D portable measuring device used to provide significantly greater flexibility and efficiency in performing quality controls directly on the production floor.

The objectives are to:

• Remove all restrictions associated with the IT infrastructure and implementation costs through the use of a secure cloud-based IoT collection platform;
• Provide an overview of plant performance using part-by-part reports, as well as overall ratios for all manufactured parts;
• Use the measurement history in the database that collects all CMM data, as well as the control card trend graphs generated by the system to predict the trends of certain variables and characteristics, and to anticipate non-compliant and non-standard items;
• Configure special customer settings and control them during production;
• Identify and even predict non-conforming and non-standard functionalities in real time.

View reports: preliminary student work, non-conformity tracking module et tests performed on Azure (in French).

A final year project is underway to test the platform ThingWorks.

Challenges

• Obtain reliable historical data and detailed specifications of the quality expected by clients.
• Predict using minimal CMM data entered after parts have been manufactured.
• Identify the machine learning algorithm suitable for this specific case.

This project is being carried out on Microsoft Azure.