Shape optimization approach for cambered otter board using neural network and multi-objective genetic algorithm

The shape optimization approach of the cambered otter board has been performed by the integration of the neural network model and the multi-objective genetic algorithm (MOGA). Because the excellent performance of an otter board is expressed by great lift and less drag force, in this study the lift and drag coefficients were chosen as objective functions to obtain the optimal otter board. The Bézier curve represented the cambered otter board as a simple structure with five control points resulting in the six coordinates, which were adopted as the design variables. The hydrodynamic characteristics of twenty-five otter board models were calculated in a two-dimension computational fluid dynamics (CFD) analysis at an attack angle of 20°. The implicit fitness function in the MOGA algorithm was then obtained by the backpropagation neural network model based on the estimated results of CFD calculation. A set of thirty optimal otter board models were extracted in the optimal solutions of the MOGA, and two optimal models were selected to verify the feasibility of the approach by hydrodynamic and visualization experiments with a comparative hyper-lift trawl door (HLTD). The model 1 showed greatest lift-to-drag ratio before the attack angle of 30° as a high lift-to-drag ratio otter board, and the model 2 showed a large lift coefficient and lift-to-drag ratio than the HLTD before the attack angle of 25° as a large lift force otter board. Through the flow distribution around the model 2, it is observed that the flow separation on the suction side is prevented as a result of less drag owing to the modified shape. In summary, the shape optimization approach is efficient in designing optimal otter board to satisfy supposed needs in otter trawling.