Prediction of machine configuration for zero defects
BSH is a global manufacturer of household appliances, being a national and European leader with more than 40 manufacturing plants around the world, as well as having a significant market share worldwide. Its production focuses on the commercial brands Bosch, Siemens, Balay, Gaggenau and Neff.
BSH was faced with the need to anticipate, well in advance, the reconfiguration and recalibration of a machine critical to its manufacturing processes in order to minimise machine downtime and avoid manufacturing defects.
In this way, the operator in charge will receive relevant information on how to reconfigure the machine and thus improve both the efficiency and the OEE rate of the production line and the factory.
- Significant increase in OEE.
- Reduction of scrap and defects due to machine set-up errors.
- Greater independence of the machine operator at any given time, and his experience.
- Substantial increase in production line efficiency.
- Gaining new insights into machine operation through data cleansing and modelling. It is now possible to know much better what factors actually impact on whether the outcome of the manufacturing process is correct or incorrect.
- Possibility of defining much more efficient manufacturing strategies.
Automatic processing of the following data sources:
- Production Order Data.
- Material Batch Data.
- Machine Process Data.
- Machine Pre-Process Data.
- Maintenance Data.
- Predictive Models able to anticipate to the Operator the best possible machine configuration in each new situation.
The Technology used for this Solution is based on:
- Machine Learning.
- Deep Learning (Neural Networks).
- Python, R.
- AWS S3 Storage.
- Apache Zeppelin / PySpark.
- RStudio Server / SparkR.
Javier Chasco Echevarría
I4.0 Responsible at BSH Electrodomésticos, Esquiroz (Navarra)
“The Machine Configuration Prediction project to achieve Zero Defects on our automatic door line was the first Data Analytics project we have tackled in the factory. During this time, we have come a long way together with PredictLand, not only in solving the project but also in getting to know a world that was new to us.
In this case, the complexity of the project is very high, this has never been a barrier for PredictLand.
What I appreciate most about PredictLand is the approach to customer service. Our feedback and opinions are always taken into account in the developments“