Diagnosing and monitoring hull corrosion with AI
The DNV artificial intelligence research center is combining an image-based tool - which takes 2D images as input – with a finetuned deep-learning algorithm, to predict areas of corrosion and coating breakdown, and assess GOOD, FAIR and POOR levels directly in proportion to the number of pixels.
The next phase will involve augmenting the solution with Intel RealSense sensors to obtain exact dimensions: a depth sensor to pinpoint the 3D position of corrosion areas, and a tracking sensor to navigate locations of the areas.
The third stage is to convert current periodical inspections to predictive inspection. To achieve this, it is vital to track and understand the root causes of corrosion evolution over a vessel’s lifetime. DNV views this as a long-term research goal once historical data is available.
The benefits
DNV is focused on creating new solutions leading the way to digitally-enabled inspections.
Corrosion.ai will enable more efficient monitoring and inspection of the vessel asset remotely. This will reduce inspection costs and risks, especially in complex areas by using unmanned vehicles such as drones and mobile robots. Images of the corrosion can be automatically captured and securely hosted and protected on DNV’s Veracity platform.
The solution will also give the shipowner a predictive tool to help make data driven decisions on the maintenance optimization and coating performance, potentially saving them hefty costs down the line.
This type of technology could become standard practice, much like the automatic assessments that are carried out in other industries where corrosion is commonplace.