Machine Learning for Marine Engines

---

Project Team

---

Project

https://sites.psu.edu/mcreu/files/formidable/2/Machine-Learning-PDF-3-1.pdf

---

Project Video

---

Project Abstract

Kazi Nafis
Campus Affiliation: Lehigh Valley
Major: Aerospace Engineering
Anticipated Graduation: May 2023
Mentors: Tracey Carbonetto (Lehigh Valley), Arash Khoshkbar-Sadigh (University Park)
Project Title: Machine Learning for Marine Engines
Machine learning has become an important process in many areas of engineering. As technology throughout the world progresses, the process of automation becomes far more common. Scientists must rely on machines to identify trends, relationships and other critical attributes gathered from large data sets. In engineering processes that depend on programmable logic controllers (PLC) to control the inputs and outputs, the PLC is now recognized as a barrier to optimizing the process. Engineers across the globe are transferring the role of a PLC to the machine learning process in order to further efficiency. This research is focused on a specific application of a four-stroke, diesel-electric, medium-speed Wartsilla engine that powers the Moving Vessel (M/V) Sulphur Enterprise, a U.S.-flagged tanker ported in Tampa, Florida. Marine engineers who crew the engine room on this vessel monitor the performance of all engines on board with emphasis on the main engine. Data is generated on a 24-hour continuous cycle with engineers reviewing the data daily. When timing and phase shift discrepancies are evident, an engineer will either mechanically increase or decrease the fuel intake, change out the valves or the fuel pump itself. Many varying factors make the overall process inefficient. Engineering students
will study the data received from the Sulphur Enterprise and look for trends between the many data points. The students will then create a theoretical program loop that would help create a more dynamic and efficient system. The program will indicate when fuel intake needs to be increased or decreased and which pump needs to be adjusted. The theoretical feedback loop will study the general trends among various inputs such as load (break horsepower), RPM, fuel rack (opening for the fuel delivery system), and pressure to create a predictive system for the adjusting process. The research indicates many trends across all data points which can potentially be used to create an optimizing feedback loop. Further research is needed to design the general structure of a machine learning feedback loop for this process. The purpose and impact of this research is to identify the marine industry as one which extensive efforts into applying machine learning processes would benefit not only the organizations who rely on these vessels to perform optimally but also to make gains in reducing the emissions from large marine vessels.

---

Evaluate this Project

Use this form link to provide feedback to the presenters, and add your project evaluation for award(s) consideration.