Corrosion Modeling and Machine Learning Tools to Promote Corrosion-Informed Design to Reduce Ship Maintenance

Triton Systems and their highly qualified team propose to develop an algorithm in Phase I to transform raw data formats containing corrosion data into information-rich features for machine learning. This may include data on relevant operational environments, materials databases, failure analyses, inspection reports, documented “lessons learned”, and results of past maintenance practices. Once literature and experimental resources have been mined for relevant data, Triton will develop a computational algorithm in Phase II that trains on this data and outputs optimal corrosion-control measures for both new ship designs and improved maintenance for legacy vessels. The Phase II database will be able to identify prioritization of features including structural, chemical, physical properties, and AM-related processing-microstructure-property phenomena relative to corrosion. These efforts will ensure reduced ship maintenance for Navy ships, saving time and money for existing ships and new designs alike.   Beyond Phase II, Triton will transition the optimized computational/informatics handling engineering tools for commercialization in ML utilization through original equipment manufacturers and other partnering agreements. This design tool will provide usefulness to DoD warfare centers in addition to production facilities. The design tool, having trained on data from a marine environment, will also be applicable to offshore structures such as oil and gas platforms.