Problem Statement

At a large aircraft manufacturing company, during robotic drilling and Wing assembly operation, key product characteristics were measured only at predefined intervals to reduce cycle time. However, any anomaly detected by this approach was 'reactive to the event' and the production had to be halted till solution to fix anomaly was identified and implemented. Hence a more comprehensive, automated and proactive quality inspection process was required to predict and prevent production line disturbance.


We employed a holistic approach of tracking and processing every aspect of manufacturing process (IoT data). Supervised machine learning algorithm was developed to deduce insights by correlating historical manufacturing key process characteristics (IoT process data) with corresponding key product characteristics (inspection data). The algorithm was then able to detect anomalies and make proactive predictions about the product characteristics.

Key Achievements

  • Fast and easy analysis of IoT and metrology data by manufacturing engineers.
  • The machine learning model highlighted the specific manufacturing parameters driving quality issues.
  • Prescriptive analysis highlighted parameters resulted in 10% reduction in the cost of quality.
  • The results were achieved with in 4 weeks (rather than 6 months) by involving lean six sigma practitioners rather than expensive data scientist.
  • Promoted a culture of data curation and data harvesting within the business.
  • Reduced/eliminating inspection step.
  • Improved yield and production efficiency.