Q3 Predictive Maintenance
Project description
Is it possible to recognize patterns in the machine's workflow, and if so is it possible to detect where the machine has produced an error?
Context
Q3 produces plastic products that require molds to be manufactured, such as bottles, containers, and more. These mold injecting machines sometimes fail due to stress, overheating, misalignment, and a variety of other causes. Our goal was to figure out within the time of a report, where the error has happened in the workflow of the machine.
Results
We have developed a system that can recognize pattern KPIs inside of the workflow of the machine. These patterns are then categorized and given a certainty score, allowing us to visualize where the algorithm believes the error has happened with a certain accuracy.
About the project group
This was a semester project for the Smart Industries group. It was about finding ways to recognize error patterns in shot/time graphs given by a mold injection machine. In the end we have successfully made an algorithm that predicts, within a range, where the error should have happened in the machine lifetime.