基于随机森林算法的大侧斜螺旋桨图谱数字化表达研究  

Research on Digital Representation of High-skew Propeller Diagram Based on Random Forest Algorithm

作　　者：李亮[1,2,3] 陈奕宏[2,3,4] 肖裕程 王超[1] 强以铭 LI Liang;CHEN Yihong;XIAO Yucheng;WANG Chao;QIANG Yiming(Harbin Engineering University,Harbin 150001,China;China Ship Scientific Research Center,Wuxi 214082,China;Taihu Laboratory of Deepsea Technological Science,Wuxi 214082,China;Zhejiang University,Hangzhou 310027,China)

机构地区：[1]哈尔滨工程大学,哈尔滨150001 [2]中国船舶科学研究中心,无锡214082 [3]深海技术科学太湖实验室,无锡214082 [4]浙江大学,杭州310027

出　　处：《中国造船》2024年第3期40-49,共10页Shipbuilding of China

摘　　要：针对大侧斜螺旋桨INSEAN E1619的图谱水动力数据,利用三次样条插值算法对图谱数据进行大规模扩充,基于多项式线性方法和随机森林法开展图谱螺旋桨推力系数和转矩系数的回归分析,通过k折叠交叉验证和随机网格搜索方法优化模型超参数,获得适合大侧斜螺旋桨图谱的高精度随机森林代理模型,它在扩充后的图谱数据集上表现更佳,螺旋桨推力系数、转矩系数和效率的平均预测误差仅为0.8%,0.8%和0.4%,可为更高精度的数字化螺旋桨图谱构建提供方法和参考,螺旋桨图谱的高精度数字化表达对实现图谱设计法的自动化和创新发展有重要影响。The high-precision digital representation of propeller diagram has a significant impact on the automation and innovation development of diagram design.The hydrodynamic data of high-skew propeller(INSEAN E1619)diagram is chosen as the object,which has been massively and effectively expanded by using cubic interpolation algorithm.The analysis of the thrust and torque coefficient of the propeller diagram was carried out by adopting the polynomial regression method and the random forest regression method.The hyperparameters of the model were optimized by k-fold cross validation and random search CV method,and the high-precision surrogate model of random forest for the high-skew propeller diagram was obtained.The results show that the random forest model performs better on the expanded diagram data set,and the average prediction errors of propeller thrust,torque and efficiency are only 0.8%,0.8% and 0.4% respectively,which can provide a method and reference for the construction of more accurate digital propeller diagram.

关 键 词：大侧斜螺旋桨 水动力 图谱 随机森林 数字化表达 

分 类 号：U664.33[交通运输工程—船舶及航道工程]