摘要
在原有图谱设计方法的基础上,采用BP(Back-Propagation)人工神经网络模型和遗传算法GA(GeneticAlgorithm),建立了一种船舶螺旋桨优化设计方法。BP人工神经网络模型通过训练可以具备强大的非线性映射能力,以数学解析的形式,较好地提取了海量螺旋桨水动力性能数据特征;GA不依赖于问题的具体领域,对问题的种类有很强的鲁棒性,为计算机辅助船舶螺旋桨优化设计提供了一种通用的多参数优化框架。针对三体消波艇半浸式螺旋桨和沿海巡逻艇螺旋桨的设计实例表明,该方法能快速可靠地搜索到最优解,不仅具有足够的工程精度,而且实用方便,适用性强。
A design optimization method for ship propeller based on BP neural network and genetic algorithm is presented.Through trained, BP neural network can obtain strong nonlinear mapping ability to learn the data characteristics of the hydrodynamic performance of ship propeller as a mathematic analytic form. Genetic algorithm provides a kind of universal framework for multi-parameter optimization which can be used to design propeller assisted by computer.Two design examples, one is a surface piercing propeller （SSP） and the other is a patrol boat propeller, shows that the design method can accurately work out the optimum propeller, and can be taken as a practical and convenient design tool.
出处
《船舶力学》
EI
北大核心
2010年第1期20-27,共8页
Journal of Ship Mechanics
关键词
半浸式螺旋桨
BP人工神经网络
遗传算法
优化设计
surface piercing propeller
BP neural network
genetic algorithm
design optimization
