Method and apparatus for analyzing and predicting future degradation of gas turbine engine components and determining if components made from counterfeit materials

Abstract

A method for a component of a gas turbine engine includes obtaining images of the component for a plurality of sequential inspection time periods, at least a portion of the images depicting degradation of the component. A first machine learning model (MLM) includes an input layer, an output layer, and a latent space therebetween. The latent space includes a plurality of clusters of nodes corresponding to respective image features. The method includes, for a plurality of the inspection time periods, utilizing the first MLM to determine which one or more of the clusters of the latent space the image most closely aligns with. The method also includes, for a plurality of the inspection time periods, utilizing a second MLM, which is a different type of MLM than the first MLM, to predict, based on the latent space of the first MLM, a degradation of the component that will occur prior to a subsequent one of the inspection time periods. The method includes determining a difference between the prediction and the image of the component from said subsequent one of the inspection time periods, and providing a notification to a user based on the difference. A system for analyzing degradation of a component of a gas turbine engine is also disclosed.