基于RBF神经网络与NSGA-Ⅱ算法的渣浆泵多目标参数优化
发布时间：2018-12-07 08:36
【摘要】：由于渣浆泵普遍存在扬程低于设计扬程、效率低、磨损严重等问题,该文选取比转速为75的离心式渣浆泵为研究对象,运用商用CFD求解软件Flunet,选取RNG k-ε湍流模型与欧拉两相流模型对其内部流动进行计算。以离心式渣浆泵的效率、高效区作为优化目标,结合Plackeet-Burman筛选试验,将渣浆泵叶片的进口安放角、出口安放角与叶片包角作为优化变量。采用均匀试验设计安排样本空间,利用RBF神经网络拟合优化变量与优化目标间的映射关联,基于NSGA-Ⅱ遗传算法进行多目标寻优。针对优化所得的Pareto解集,选取其中效率最优个体和高效区最优个体与优化前初始模型进行对比:分析了上述3个个体的通过数值模拟得到的性能曲线之间的差异,得到效率最优与叶片进、出口安放角、叶片包角为21.76?、23.43?、145.56?,高效区最优时为19.38?、22.68?、116.71?。通过试验验证,优化后个体性能得到显著提升,效率最优个体的效率较初始个体的效率提高了3.81%,高效区最优个体较初始个体高效区范围提高了4.33%。给出并分析了上述3个个体在叶轮流道中间剖面上固相相对速度矢量及湍动能分布、叶片工作面、叶轮后盖板的固相浓度分布差异。优化结果表明,该优化方法使叶轮的水力特性得到改善,提高了离心式渣浆泵的性能。
[Abstract]:Due to the problems of lower lift, lower efficiency and serious wear, the centrifugal slurry pump with 75 specific speed is selected as the research object, and the commercial CFD software Flunet, is used to solve the problem. RNG k- 蔚 turbulence model and Euler two-phase flow model are selected to calculate the internal flow. Taking the efficiency and high efficiency of centrifugal slurry pump as the optimization objective and combining with the Plackeet-Burman screening test, the inlet placement angle, the outlet placement angle and the blade envelope angle of the slurry pump blade were taken as the optimization variables. The uniform test design is used to arrange the sample space, and the RBF neural network is used to fit the mapping association between the optimization variables and the optimization objectives, and the multi-objective optimization is carried out based on the NSGA- 鈪，
本文编号：2366916
[Abstract]:Due to the problems of lower lift, lower efficiency and serious wear, the centrifugal slurry pump with 75 specific speed is selected as the research object, and the commercial CFD software Flunet, is used to solve the problem. RNG k- 蔚 turbulence model and Euler two-phase flow model are selected to calculate the internal flow. Taking the efficiency and high efficiency of centrifugal slurry pump as the optimization objective and combining with the Plackeet-Burman screening test, the inlet placement angle, the outlet placement angle and the blade envelope angle of the slurry pump blade were taken as the optimization variables. The uniform test design is used to arrange the sample space, and the RBF neural network is used to fit the mapping association between the optimization variables and the optimization objectives, and the multi-objective optimization is carried out based on the NSGA- 鈪，
本文编号：2366916
本文链接：https://www.wllwen.com/jixiegongchenglunwen/2366916.html
