军事知识服务平台
基于径向基神经网络与多目标粒子群优化算法的全贯流泵导叶优化设计
Optimization design of guide vanes for full tubular pump based on radial basis function neural network and multi-objective particle swarm optimization algorithm
摘要 为提高全贯流泵的水力性能,对泵的导叶体进行优化设计研究.通过Plackett-Burman 筛选试验,筛选出对全贯流泵扬程、效率影响显著的4 个几何参数:导叶扩散角Φ,叶轮与导叶叶片轴向间距 s,导叶叶片高度 h,导叶叶片外缘包角 φ5.基于最优拉丁超立方试验建立数据库,构建了径向基函数(radial basis function,RBF)神经网络模型,以确定目标函数与几何参数之间的非线性关系,并通过误差分析验证了近似模型的可靠性.运用多目标粒子群优化算法(multi-objective particle swarm optimization,MOPSO)进行全局寻优,获得Pareto 前沿解,进而选取Pareto 前沿上效率最大的样本进行数值计算,得到优化后的泵模型.计算结果表明:优化后的全贯流泵与初始模型相比,在各工况下的水力性能均有所提升,在设计工况点效率提升 2.65 个百分点.基于熵产理论分析,优化后的泵各熵产值均显著减小.内部流动分析表明,优化后导叶高能耗区域明显减少,熵产极值部位得以消除,水力性能显著提高,运行稳定性得到改善,优化效果明显,验证了全贯流泵智能优化设计的有效性.
Abstract:
In order to improve the hydraulic performance of a full tubular pump,an optimization design study was conducted on the pump′s guide vane body.Through Plackett-Burman screening tests,four geometric parameters that significantly affect the head and efficiency of the pump were identified as fol-lows:diffuser angle of the guide vane(Φ),axial spacing between the impeller and guide vane blades(s),height of the guide vane(h),and wrap angle of the guide vane′s outer edge(φ5).The optimal Latin hypercube design was used to establish a database for a radial basis function(RBF)neural net-work model to determine the nonlinear relationship between the objective function and geometric pa-rameters.The reliability of the approximate model was verified by error analysis.A multi-objective par-ticle swarm optimization algorithm(MOPSO)was then applied for global optimization to obtain the Pa-reto front solution.Then,the sample with the highest efficiency on the Pareto front was selected for nu-merical simulation to obtain a new optimized model.The computational results show that the optimized pump model exhibits enhanced hydraulic performance across all operating points compared to the initial model,with an increase of 2.65 percentage points in efficiency at the design operating point.Based on entropy production theory analysis,the optimized pump shows a significant reduction in entropy pro-duction values at all points.Internal flow visualization analysis further reveals that the high-energy dis-sipation regions in the optimized guide vane are noticeably reduced,and the areas of maximum entropy production are eliminated.The hydraulic performance is significantly improved,and the operational stability is also improved.The optimization effect is obvious,verifying the effectiveness of the intelligent optimization design of full tubular pumps.
刊 名 排灌机械工程学报 ISTICPKU
年,卷(期): 2026, 44(10)
分类号: TH312
机标分类号: TP391.41 TP183 TP273
在线出版日期: 2026年11月6日
基金项目: U20A20292:国家自然科学基金52371322:国家自然科学基金,