采用免疫算法的离心压缩机叶轮 多目标优化及性能分析  

Multi-Objective Optimization and Performance Analysis of Centrifugal Compressor Impeller Based on the Immune Algorithm

作　　者：梁璐 宫武旗[1] 刘一彤 王芳[1] LIANG Lu;GONG Wuqi;LIU Yitong;WANG Fang(School of Energy and Power Engineering,Xi’an Jiaotong University,Xi’an 710049,China)

机构地区：[1]西安交通大学能源与动力工程学院,西安710049

出　　处：《西安交通大学学报》2024年第2期56-65,共10页Journal of Xi'an Jiaotong University

基　　金：国家科技重大专项资助项目(2017-Ⅴ-0012-0064)。

摘　　要：为进一步探索高性能离心压缩机的优化设计方法,以某气浮轴承离心制冷压缩机叶轮为研究对象,结合多层前向神经网络,发展了基于非支配近邻免疫算法的叶轮多目标优化策略。首先,采用免疫算法优化神经网络的隐含层结构,以提升其作为代理模型的非线性逼近能力;其次,以近喘振点、设计点、近堵塞点的多变效率最大为优化目标,采用免疫算法进行叶轮全工况性能寻优,并通过数值仿真对优化前、后叶轮的气动性能及流动特性进行了对比分析。仿真结果表明:采用免疫算法进行优化后,叶轮在近喘振点、设计点、近堵塞点的多变效率分别提高了1.8%、1.9%、4%,稳定运行工况范围明显拓宽;对比流场后发现,在90%叶高处,主叶片和分流叶片载荷从前缘至尾缘均明显增加,叶片做功能力增大;分流叶片进口倾角减小使得叶片进口冲击损失降低,流道内泄漏流与主流掺混现象明显减弱,叶轮内部流动更加均匀。研究结果验证了所提多目标优化策略的有效性。In order to further explore the optimal design of high-performance centrifugal compressor,taking the impeller of a centrifugal refrigeration compressor with air bearing as the research object,combined with the multilayer BP neural network,a multi-objective impeller optimization strategy based on the nondominated neighbor immune algorithm is developed.Firstly,the hidden layer structure of the neural network is optimized using the immune algorithm,aiming to improve the nonlinear approximation ability of the neural network as a surrogate model.Secondly,the immune algorithm is applied to optimize the performance of impeller under all operating conditions with the polytropic efficiency of near surge point,design point and near choke point as the optimization target.The aerodynamic performance and internal flow characteristics of the impeller before and after the optimization are compared and analyzed through numerical simulation.The simulation results show that after the optimization based on the immune algorithm,the polytropic efficiency of the impeller increases by 1.8%,1.9%and 4%at the near surge point,design point and near choke point,respectively,indicating that the stable operating range of impeller is widened.The flow field analysis shows that the loads distributed from LE to TE of the main blade and the splitter blade at 90%span increase obviously after the optimization,which reflects improved work capacity of the impeller.The shock loss decreases because of the reduced inclination angle of the splitter blade.The mixing phenomenon between the leakage flow and main flow in the passage is obviously weakened,and the flow field is more uniform,which verifies the effectiveness of the proposed multi-objective optimization strategy.

关 键 词：离心制冷压缩机 多层前向神经网络 非支配近邻免疫算法 气动性能 多目标优化 

分 类 号：TK05[动力工程及工程热物理]