基於船舶減阻之小水線面雙體船水下胴體設計優化

Abstract

本研究藉由深度神經網路模型尋找優化的22.5公尺水下胴體船形設計，藉此降低具穩定翼小水面雙體船在平浮條件下F_r " = 0.8" 的靜水阻力。本研究將小水線面雙體船的阻力線性化為浮筒、支架和穩定翼阻力，藉此簡化整體阻力的計算。本研究使用4個船形參數定義水下浮筒外形，分別為浮筒前段長度、浮筒後段長度、浮筒入水角及浮筒出水角。本研究使用計算流體力學軟體STAR-CCM+預測水下胴體完全沒水阻力，以及計算穩定翼在不同攻角下的升阻力。接著使用MATLAB數學軟體與1400個CFD阻力預測結果訓練深度神經網路模型，通過K-fold交叉驗證確保模型的穩定性，並尋找優化的船形參數組合。本研究所提出的深度神經網路模型包含五個隱藏層，每個隱藏層的神經元數量分別為6、8、9、8和7。本研究建議的優化設計參數為浮筒前段長度7.8 m，浮筒後段長度6.8 m，前段角度10°，後段角度35°。本研究發現相較於原始船形，優化設計船形的減阻效應主要來自水下胴體產生的孟克力矩較小，使得所需穩定翼的攻角較小。由於小水線面雙體船依靠穩定翼平衡船體生成的孟克力矩，因此穩定翼攻角減少，導致穩定翼阻力大幅降低。使用本研究建議的優化水下胴體設計，總阻力相較於原始船形減少2.2%。

This study employs a deep neuron network (DNN) model to optimize the 22.5 m long pontoon hull form of a small water-plane area twin hull (SWATH) vessel with fin stabilizer for reducing its calm water resistance at F_r " = 0.8" under an even keel condition. The resistance of the target vessel is linearized into three components, i.e., pontoon, strut, and fin stabilizer, to simplify the resistance calculation. Four design parameters, i.e., the length of the fore-body and aft-body, the angle of fore body and aft body, are used to define the geometry of pontoon. The computational fluid dynamics (CFD) software STAR-CCM+ is used to predict the resistance of the underwater pontoon as well as the lift and drag force of the fin stabilizer at different angles of attack. Then, a deep neural network model is trained with 1400 CFD resistance predictions using MATLAB, and K-fold cross-validation is used to ensure the DNN model stability and search for the optimized design parameter set. The proposed DNN model has 6, 8, 9, 8, and 7 neurons in five hidden layers, respectively. The optimized design parameters are the length of the fore-body 7.8 m, the length of the aft-body 6.8 m, fore body angle 10˚, and the aft body angle 35˚. This study finds that the resistance reduction of the optimized design compared to the baseline design is mainly due to the small angle of attack of fin stabilizers where the optimized pontoon results in a small Munk moment to be balanced by the fin stabilizer. The optimized pontoon design is able to reduce the resistance by 2.2% compared to the baseline design.