Position parameters optimization of surface piercing propeller by artificial neural network

Area of Science:

• Naval Architecture and Marine Engineering
• Computational Fluid Dynamics
• Optimization Algorithms

Background:

• Surface-piercing propellers are critical for marine propulsion.
• Optimizing their performance requires understanding complex hydrodynamic interactions.
• Existing methods may be time-consuming or lack accuracy in predicting performance parameters.

Purpose of the Study:

• To investigate key factors influencing surface-piercing propeller performance.
• To develop accurate predictive models for propeller performance.
• To optimize propeller position parameters for enhanced efficiency.

Main Methods:

• Utilized Artificial Neural Networks (ANN) for nonlinear model identification.
• Employed Non-Dominated Sorting Genetic Algorithm II (NSGA-II) for optimization.
• Collected experimental data from IUST's HYDROTECH center for training and validation.

Main Results:

• ANN achieved a mean error of 7.5e-5 for trained data and 1e-4 for verified/test data.
• Optimization yielded a 9.7% relative error for thrust coefficient and 7.5% for torque coefficient.
• Validated the effectiveness of ANN and NSGA-II in predicting and optimizing propeller performance.

Conclusions:

• The combined ANN and NSGA-II approach provides accurate performance predictions.
• This methodology offers a cost-effective and time-saving solution for propeller optimization.
• The study demonstrates a viable method for improving surface-piercing propeller efficiency.