Treffer: Blow Molding Process Automation using Data-Driven Tools

With the increasing demand for goods in today’s world with its ever-increasing population, industries are driven to boost their production rate. This makes issues of process optimization, maintenance, and quality control difficult to be carried out manually. Furthermore, most industrial sectors have a plethora of data acquired every day with very little knowledge and ideas to handle it most effectively. Motivated by these considerations, we carry out a study in process automation using data-driven tools for a manufacturing process intended to produce high-quality containers. The first task involved studying the process, building and proving a hypothesis through data collection from experiments and the production line. Our hypothesis was the linear correlations between various sensor variables from our physical understanding of the process. We developed an automated process flow, i.e., a so-called Digital Twin (a numerical replication of the entire process) for this process to enhance the analytical and predictive capabilities of the process. This showed similar predictions when compared to static models. An automated in-line quality control algorithm was also built to remove the manual component from this task, using state-of-the-art computer vision techniques to utilize the power of data in the form of images. Lastly, to further provide the process engineers with predictive power for maintenance we carried out a few proof of concept projects to show the competence of such tools in minimizing costs and improving efficiency on the shop floor. All studies carried out showed great results and have immense potential to methodically use data to solve some of the pressing problems in the manufacturing sector.