基于HBA-SVR混合模型的斜式轴流泵变角性能预测  

Variable Angle Performance Prediction of a Slanted Axial-flow Pump Based on HBA-SVR Model

作　　者：郑海生 周佩剑 肖刚[1,2] 牟介刚 项春[3] 钱亨 ZHENG Haisheng;ZHOU Peijian;XIAO Gang;MOU Jiegang;XIANG Chun;QIAN Heng(College of Metrology and Testing Engineering,China Jiliang University,Hangzhou,Zhejiang 310018,China;College of Mechanical Engineering,Zhejiang University of Technology,Hangzhou,Zhejiang 310023,China;Zhejiang University of Water Resources and Electric Power,Hangzhou,Zhejiang 310018,China)

机构地区：[1]中国计量大学计量测试工程学院,浙江杭州310018 [2]浙江工业大学机械工程学院,浙江杭州310023 [3]浙江水利水电学院,浙江杭州310018

出　　处：《计量学报》2025年第2期190-197,共8页Acta Metrologica Sinica

基　　金：中国博士后科学基金面上资助项目(2022M712816);浙江省公益技术研究计划/工业(LGG21E090003);浙江省“尖兵”“领雁”研发攻关计划(2023C01022,2022C02035)。

摘　　要：针对斜式轴流泵不同叶片角度下性能曲线获取难、耗费成本高的问题,提出了基于混合蝙蝠算法-支持向量回归模型(HBA-SVR)斜式轴流泵性能预测方法。在标准蝙蝠算法中加入方向加速策略和变异策略优化支持向量回归,利用斜30°轴流泵运行数据训练模型,并应用于斜式轴流泵变角性能预测。扬程、效率平均相对误差分别为1.49%、0.41%,收敛时间分别为15.47 s、18.78 s,相较于标准蝙蝠优化支持向量回归预测结果,收敛时间分别减少了122.11%、103.62%。对比PSO、GA、BA优化SVR,扬程预测误差分别降低了29.53%,70.46%,131.54%,效率预测误差分别降低了7.31%,9.75%,19.51%。结果表明所提出模型能快速、有效预测斜式轴流泵变角性能。To address the difficulty and cost associated with obtaining performance curves for different blade angles,a performance prediction method for slanted axial-flow pump based on the hybrid HBA-SVR model is proposed.The standard bat algorithm is enhanced with directional acceleration and variation strategies to optimize the SVR.The model is trained by using the operating data of the 30°slanted axial-flow pump,and applied to the variable angle performance prediction of the slanted axial-flow pump.The average relative errors for head and efficiency are reduced to 1.49%and 0.41%respectively,with convergence times of 15.47 s and 18.78 s.When compared to the results of standard bat optimization support vector regression prediction,the convergence times are reduced by 122.11%and 103.62%respectively.Moreover,compared to PSO,GA,and BA optimized SVR,the head prediction errors are reduced by 29.53%,70.46%,and 131.54%respectively,and the efficiency prediction errors are reduced by 7.31%,9.75%,and 19.51%respectively.The results indicate that the proposed model effectively predicts the variable angle performance of slanted axial-flow pump.

关 键 词：流量计量 斜式轴流泵 支持向量回归 蝙蝠算法 叶片安放角 变角性能预测 

分 类 号：TB937[一般工业技术—计量学]