Applications of Artificial Intelligence (AI) to Marine Propeller Design

Projects Descriptions:

1. Design Support System for Preliminary Design of Marine Propellers

The aim of this project is to study the process of propeller design and build a design decision support system to help designers improve and speed up their work.
The project is a collaborative effort with:

• Volker Bertram, Institut für Schiffbau, Lämmersieth 90, 22305 Hamburg, Germany
• Jürgen Friesch, Hamburg Ship Model Basin, Bramfelder Str 164, 22305 Hamburg, Germany

• Cased-Based Reasoning
• Design Rationale
• Fuzzy Logic
• Neural Networks

A Sample Publications Include:

• Reich, Y. and Bertram, V. and Friesch, J. (1997), The development of a decision support system for propeller design, Proceedings of the 9th International Conference on Computer Applications in Shipbuilding (ICCAS '97).
  (Postscript file, 156K)

2. Using Neural Networks (NN) for Building Models of Propeller Behavior

This project investigates the use of neural networks for creating models that can predict the behavior of maring propellers (Kt, Kq, and eta). The aim of the project is to develop techniques that could assess the reliability of predictions made using the models and relate them to the reliability of the various steps of building these NN models. Thus far, the results are very promissing.

A Sample Publications Include:

• Barai, S. V. and Reich, Y. (1998), Learning reliable marine propeller behavior, Submitted for publication.
  (Postscript file, available soon)