基于人工神经网络的船型设计与阻力预报

摘要

The relationship between resistance and hull shape is highly complicated nonlinear. In real design, empirical formula method and CFD method are mainly deployed to predict the ship resistance.Due to the ability of artificial neutral network (ANN) theory for solving highly nonlinear problems,with the adoption of RBF network, the paper developed a new method for hull shape design exceeding the limitations of conventional methods and avoiding the high cost of CFD method, and built the model for resistance forecasting. The real ship data proved the feasibility and practicability of the method.%船舶阻力与船体形状之间的关系是复杂的非线性关系,在实际设计中预报船舶阻力的方法一般主要有经验公式法和CFD方法.从打破传统方法的局限性和避免CFD计算的高成本出发,考虑人工神经网络理论对处理非线性复杂问题具有良好的适应性,采用RBF神经网络,试图为不同设计阶段的需要提出新的船型设计方法,并建立了相应的船型-阻力性能预测模型.通过实际数据验证了此方法的可行性和实用性.