Conclusion
This concludes our workshop! We have taken a closer look at all the parts required to build a manufacturing data lake using data from a factory and gaining insights: from connecting your machinery and PLCs through an industrial gateway, collecting, streaming and storing the data and then building up the required infrastructure to make this data accessible for analytics tools. We’ve also gained the first few insights based on the data we ingested, visualizing it to get a better look at metrics and to start investigating where defects might be introduced.
What’s next? We have invested quite a bit of time to get the data lake up and running – but of course, having just a single table is but a starting point. For a real world use case, we would now start to add all kinds of other data into the data lake – ERP information on product batches, orders and customers; BOMs, more information on actual product performance in the field etc. so that we can run more interesting analysis. To make our lives and our analysis easier, we would also look at transforming the raw event data into something more accessible, e.g. with views listing all the information on a single productid in one place, or a timeline of the changes in sensor readings by station over time etc. We would also start looking more closely at the actual sensor data, looking to predict errors in the process, SUSPEND states or quality problems based on these early warning signs etc.
The good news is that the infrastructure we have built up today can serve as a foundation for your own experiments going forward! We hope you enjoyed this brief journey – have fun building on this!
