In paint manufacturing, pH, color and viscosity specified by clients are challenging to meet because of irreducible variability in production conditions, thereby reducing output consistency. A substantial percentage of the stock mixtures requires multiple rounds of adjustments to meet requirements, resulting in significant waste, additional cost, profit reduction, lower productivity and delivery delays. Our AI solution solves these issues.
Cognistx uses a variational deep learning model designed for large scale, real-world paint and coatings production. Chosen from >50 alternative models, our algorithm accommodates highly dimensional and non-linear sequential data on >2000 raw material and multiple plant conditions.
Our architecture accommodates irreducible test errors, detection limits as well as highly dimensional, non-linear sequential data on thousands of raw materials and production conditions.
By improving the manufacturing process, our AI solution is expected to save between $500K - $1M per plant, annually.
Ingest:
Raw material properties, plant conditions and final product requirements data is input into the system
Analyze:
Cognistx data engine cleans and transforms the data
Predict:
Machine learning engine suggests optimal raw material mix for human feedback
Learn:
Self-tuning engine remembers and automatically improves model from human feedback
Measure:
System dashboard displays real-time production performance matrix for human inspection
The Result:
Improved Quality Control, Reduced Cost, Increased Production Capacity for Plants
Ready for AI? Just email Sanjay Chopra to set up an introductory call.
