Machine Learning of Thermal Spray Manufacturing “big data” for SMART Component Specifications

A majority of aerospace components, particularly aircraft engines, have coatings applied to them to both enhance functionalities and enable sustainment (multiple overhaul and repair cycles).  Coatings provide a platform to sustainment and as such is the key step during overhaul operations at DoD depots.  Thermal spray is a key process for both OEM and repair/overhaul sustainment operation.   Thermal spray processes are highly complex involving some 100 or more variables. The qualification of these coatings through specifications can at many times be subjective with large variabilities.    ReliaCoat has developed and transitioned new approaches to both process and product evaluation which are both synchronous and quantitative.  Concurrently, ReliaCoat has also deployed these concepts at USAF logistic centers in a production friendly automated format to allow ease of data collection.  Through the proposed AFWERX SBIR, ReliaCoat seeks to understand the data that is being generated in manufacturing, connect the dots through a priori knowledge and machine learning tools and validate them so as to enhance qualifications and specifications enabling 21st century transformation of these sustainment capabilities.