Multiple projects are being conducted by machines in the ISE manufacturing laboratory. Yet all of the data generated by the machines during part fabrication are not being stored. I wanted a method to store real-time streaming data from the machines in a database that was structurally capable of handling large streaming data from manufacturing shop-floors. Proprietary plant historians or relational type databases are not suited for such real-time applications.

What it does

Machine generated data during part fabrication is captured, stored in a scale-able database and finally retrieved by client applications to be compared with original product geometry. Deviations between digital definition vs actual geometry is computed and displayed as a visualization output. This output can be further utilized as a first round qualification of the part.

How I built it

• Obtained a low-cost computing device to interface a 2005 HAAS VF2 to the university ethernet infrastructure. • Setup a PostgreSQL database on a server within the university network to capture data streaming from the machines. • Built python based client applications for a manufacturing intelligence app that directly interfaces with the database.

Challenges I ran into

There were lots of challenges when I began this project. Some of them have been overcome but a few have been highlighted that still needs better solutions. • RS232 ports are slow to communicate with the HAAS machines. The lack of APIs and direct interfaces to the machine’s control system slows down data capture from the machine hardware. It may be solved by acquiring hardware adapters from HAAS. • There is a lack of libraries in python that allowed direct import of CAD geometry. Better software tools may be needed to resolve this issue.

Accomplishments that I'm proud of

• Setting up the next generation PostgreSQL, a much better database suited for capturing streaming time-series data that is easy to scale as the number of machines streaming data is made possible. • Building the python 3D graphical GUIs that interface directly with the database, rather than with the MT-CONNECT agent itself. • Setting up a live LabVIEW application that directly reads in machine generated data and accessible within the network.

What I learned

• Being a manufacturing engineer, learning the importance of manufacturing intelligence and data as an asset rather than an afterthought. • Next generation SQL database beyond that of Access and MySQL.

What's next for Real-Time Digital Verification of Product Manufacturing Data

• Interconnect multiple machines on the NCSU manufacturing laboratory and stream their data to a data-store of some sort. • Store streaming data for community access, either to students or researchers. • Build data models around the data generated during product fabrication.

Built With

Share this project: