Summary: | 碩士 === 國防大學理工學院 === 造船及海洋工程碩士班 === 104 === This research is to employ a self-propulsion model test system of waterjet propulsion in order to investigate the performance of the waterjet model. This work adopted the Design of Experiments (DOE) by Taguchi’s method, Load Cell used to measure thrust, Laser Doppler velocimetry (LDV) and 3D printing technique to investigate the performance of a waterjet propulsion system. In this study, three controllable factors are considered, such as the setting angle of stator, the numbers of blade on rotor and stator respectively. Furthermore, this work consists of applying Taguchi method which is used to deal with a robust design, ANOVA which is used to analyze the variances in experimental data by the use of Signal-to-Noise ratio for investigating the contribution of factors and interaction between two factors. Then, Artificial Neural Network (ANN) which is an implement to construct a non-linear model, and Adaptive Particle Swarm Optimization (APSO) which is a technique to find out the best combination of levels for factors have to be completely performed in this study. Moreover, it has been confirmed that the waterjet propulsion system conducted by using the DOE method and the ANN modeling technique through the Radial Basis Function Network (RBFN) in this research can achieve higher thrust and efficiency.
|