| 基于RBF神经网络和NSGA-Ⅱ算法的海水淡化高压泵多工况优化设计 |
| |
| 引用本文： | 张德胜,张奇,赵睿杰,祁炳,沈熙.基于RBF神经网络和NSGA-Ⅱ算法的海水淡化高压泵多工况优化设计[J].排灌机械工程学报,2022,40(12):1189-1196. |
| |
| 作者姓名： | 张德胜 张奇 赵睿杰 祁炳 沈熙 |
| |
| 作者单位： | 江苏大学国家水泵及系统工程技术研究中心, 江苏 镇江 212013 |
| |
| 摘 要： | 为提高透平式能量回收一体机的水力性能,采用RBF神经网络、NSGA-Ⅱ遗传算法和数值模拟集成的设计方法,对高压泵进行多工况优化设计.以高压泵三工况点的加权平均效率为优化目标,设计工况点下的扬程为约束条件,结合Plackett-Burman筛选试验,将进口安放角、出口安放角、叶片出口宽度及叶片包角作为优化变量,采用最优拉丁超立方设计试验空间,基于Isight多学科优化平台,通过编写批处理命令集成CFturbo,ICEM,CFX等,搭建智能水力优化平台,实现高压泵的CFD自动预报.基于数值模拟结果,利用RBF神经网络建立目标函数与几何参数之间的非线性关系,并运用NSGA-Ⅱ算法对该模型寻优.研究结果表明:RBF神经网络模型能够准确预测高压泵效率以及扬程与设计变量之间的关系;优化后高压泵与初始方案相比,3个工况点加权平均效率提高了3.38%,在0.8Qd,1.0Qd和1.2Qd工况下效率分别提高了2.21%,3.59%和4.23%;优化后叶轮在设计工况点附近轴功率略有减小;对比优化前后叶轮的流场分布,优化后的叶轮进口低速区减小,内部速度梯度分布更加均匀,流场得到明显改善,能量耗散减小.研究结果可为高压泵多工况水力优化设计提供一定理论依据.
|
| 关 键 词： | 海水淡化高压泵 能量回收一体机 多工况优化设计 RBF神经网络 NSGA-Ⅱ遗传算法 |
| 收稿时间： | 2021-03-21 |
Multi-condition optimization design of seawater desalination high-pressure pump based on RBF neural network and NSGA-Ⅱ genetic algorithm |
| |
| ZHANG Desheng,ZHANG Qi,ZHAO Ruijie,QI Bing,SHEN Xi.Multi-condition optimization design of seawater desalination high-pressure pump based on RBF neural network and NSGA-Ⅱ genetic algorithm[J].Journal of Drainage and Irrigation Machinery Engineering,2022,40(12):1189-1196. |
| |
| Authors: | ZHANG Desheng ZHANG Qi ZHAO Ruijie QI Bing SHEN Xi |
| |
| Institution: | National Research Center of Pumps, Jiangsu University, Zhenjiang, Jiangsu 212013, China |
| |
| Abstract: | In order to improve the hydraulic performance of the turbine type energy recovery integrated machine, RBF neural network, NSGA-Ⅱ genetic algorithm and numerical simulation integrated optimization design method were used to optimize the design of high-pressure pump under multiple working conditions. In order to do that, the weighted average efficiency at multiple operation was taken as the optimization objective, the head under the design condition was taken as the constraint condition, and the four parameters, i.e., blade inlet angle, outlet angle, outlet width and wrap were selected for the Plackett-Burman test design, and the optimal Latin hypercube sampling method was used to choose the design point in the design space. Besides, based on the Isight multidisciplinary optimization platform, an intelligent hydraulic optimization platform is built by writing batch processing commands to integrate CFturbo, ICEM, CFX, which can achieve automatic CFD prediction of high-pressure pumps. Based on the numerical simulation results, the nonlinear relationship between the objective function and geometric parameters was established using RBF neural network, and the NSGA-Ⅱ algorithm was used to optimize the model. The results indicated that the RBF neural network model could accurately predict the relationship between the head, efficiency and the design variables. Compared with the initial scheme, the weighted average efficiency of the optimized high-pressure pump at three points is improved by 3.38%, more specifically, the efficiency under 0.8Qd,1.0Qd and 1.2Qd is increased by 2.21%,3.59%, 4.23%, respectively, and the shaft power under the design condition is slightly lower than the prototype model. Compared with the flow velocity distribution, streamlines, turbulence energy distribution of the optimized and reference impeller, it is found that the low-speed area at the impeller inlet is reduced, the internal velocity gradient distribution is more uniform, the flow field is significantly improved, and the hydraulic loss of energy dissipation is reduced after optimization. The optimization design method presented in this paper can provide theoretical basis for the hydraulic optimization design of high-pressure pumps under multiple operating conditions. |
| |
| Keywords: | seawater desalination high-pressure pump energy recovery integrated machine multi-condition optimization design RBF neural network NSGA-Ⅱ genetic algorithm |
|
| 点击此处可从《排灌机械工程学报》浏览原始摘要信息 |
| 点击此处可从《排灌机械工程学报》下载免费的PDF全文 |
|
