刘正先¹, 韩博², 鲁业明¹
AuthorsHTML:刘正先¹, 韩博², 鲁业明¹
AuthorsListE:Liu Zhengxian¹, Han Bo², Lu Yeming¹
AuthorsHTMLE:Liu Zhengxian¹, Han Bo², Lu Yeming¹
Unit:1. 天津大学机械工程学院，天津 300072；2. 天津大学化工学院，天津 300072
Unit_EngLish:1．School of Mechanical Engineering, Tianjin University, Tianjin 300072, China
2．School of Chemical Engineering and Technology, Tianjin University, Tianjin 300072, China
Abstract_Chinese:针对离心叶轮优化设计中调控参数不明确、难以直接获得最优设计结果的状况, 将单层神经网络算法应用于离心叶轮优化, 并建立了分层优化的算法．根据设计目标将离心叶轮叶片优化设计分为压比和效率两个阶段, 随后分析确定了各目标阶段的有效控制变量参数．最后以初始叶轮为基础, 将该目标优化算法应用于某工业叶轮的改型实例设计, 结果表明：压比目标算法使叶轮达到压比值要求; 一次效率目标算法实现了对叶轮设计点流量的调控; 二次效率目标算法达到了对叶轮非设计工况性能的提升．
Abstract_English:Aiming at the uncertainty of setting the control parameters and the difficulty in getting the optimized results directly in the optimization design process of centrifugal impellers，the monolayer neural network was applied to optimize the centrifugal impeller，and then the hierarchical optimization algorithm was established. Based on the design target，the optimization process of blades for centrifugal impellers was classified into two stages，namely，the pressure ratio stage and the efficiency stage，and after that，the effective control parameters of all stages were analyzed and determined. Finally，the initial impeller was set as the base，and with the application of the optimization algorithm in the modification process of an industrial impeller，it could be found as follows: the pressure ratio objective algorithm can make the impeller reach the target value of pressure ratio; the first-stage efficiency objective algorithm has succeeded in adjusting the mass flow at design point; the second-stage efficiency objective algorithm can improve the impeller performance of off-design mass flow.
Keyword_Chinese:离心叶轮; 目标算法; 单层神经网络; 分级优化设计; 数值计算
Keywords_English:centrifugal impeller; objective algorithm; monolayer neural network; grading optimization design; numerical calculation

PDF全文下载地址:http://xbzrb.tju.edu.cn/#/digest?ArticleID=5760