【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。更多还原

【Abstract】 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.更多还原