Neural network integration in the analysis, selection of wageningen’s b-series ship propeller

Abstract

In the preliminary stage, ship design analyzes and evaluates the close correlation and interaction fit between the hull, main engine, and propulsion. Therefore, the process of calculation and selection of the appropriate propulsion device plays the role of ensuring the necessary propulsion to achieve the design speed according to the mission and ensuring the appropriate torque of the main engine to achieve optimal performance. According to the traditional approach, the propeller design method is based on series B - Wageningen's experimental graph to determine the suitable diameter and geometric parameters. This paper presents the method of integrating the neural network algorithm in the preliminary design stage to support selecting the appropriate blade area ratio from input parameters, including the ship length, displacement, design speed, and the number of propeller blades. The neural network principle is to synthesize the reference result from the propeller database's individuals to give the appropriate blade area ratio with the closest probability in the database, taking into account cavitation. In this study, the B-Wagenningen series propeller design database is verified and applied well in practice. On that basis, the propeller geometry parameters are proposed from the neural network algorithm, and the thrust and torque coefficients are calculated and verified based on computational analysis from commercial software.