摘要
在叶片气动设计中,叶片叶尖部分翼型的升阻比和失速迎角对叶片的气动性能有很大影响。根据不同样本叶尖翼型的几何特征和气动特征,利用SOM神经网络对其进行分组,形成系统的叶尖翼型专家数据库。在此基础上,采用置信度推理法建立了翼型几何参数与气动参数之间的关系,并给出了优化方向,由此产生优化设计外形。研究结果表明:SOM神经网络能够有效地区分有相同特征的一类翼型,分类灵活,可以为风机叶片设计工作提供方向性指导;最终得到的设计翼型与基准翼型相比,有效地提高了升阻比和失速迎角,具有较优的综合气动性能。
An optimization system for wind turbine blade based on Self-Organizing Map（SOM） was deve-loped. It is important for airfoil design in blade tip section to improve the lift-to-drag curve and delay the occurrence of stall phenomenon. With the aerodynamic characteristics calculated by N-S equation and the relevance built between the geometry and aerodynamic characteristics, an expert database was built by SOM method,which can automatically give the optimization direction with the design target according to the Certainty Factor inference method. Based on reference geometry and optimize direction mentioned a- bove, the network gives a large number of optimized geometry. As a test sample,the airfoil $828 was im-proved through optimization direction and an optimized airfoil was developed. Furthermore, Blade-Ele-ment Momentum（BEM） theory is introduced to evaluate the output power performance of the optimized blade. The present method is proven to be useful for improving the performance of wind turbine blades.
出处
《力学季刊》
CSCD
北大核心
2012年第4期672-678,共7页
Chinese Quarterly of Mechanics
基金
上海市科委科研计划项目(10521100403)资助
关键词
风机叶片
神经网络
优化
翼型
wind turbine blade
SOM
optimization
airfoil
