轴流泵叶片动力特性预测模型研究

发布时间：2018-02-12 03:57

本文关键词： 轴流泵叶片 Bezier曲线参数化建模数值分析 ANN代理模型　出处：《西华大学》2012年硕士论文　论文类型：学位论文

【摘要】：综合升力法、Bezier曲线、三维建模、计算流体力学（C FD）、人工神经网络等理论知识，研究一种轴流泵叶片设计的新方法，即：改进轴流泵叶片设计理论及其建模方法，采用升力法、选用NACA翼型对轴流泵叶片进行初始设计。对所得叶片各截面翼型点进行坐标转换，将其转换为空间坐标点。利用Bezier曲线对叶片翼型坐标点进行曲线拟合，从各截面翼型坐标点中选取四点作为型值点，反求Bezier曲线控制点，从而可得各相应翼型曲线。将各曲线转换为UG曲线表达式，导入UG生成叶片翼型骨线，结合其它结构参数，实现轴流泵参数化建模，增强了叶片的可控性，并使叶片具有最少的控制参数。采用VB编程，实现输入轴流泵设计参数、选择叶片翼型后，程序自动对轴流泵进行初始设计，并输出轴流泵结构参数及叶片UG建模曲线。利用UG布尔运算功能建立轴流泵叶轮处流域模型，利用UG WAVE技术建立轴流泵喇叭管及导叶处流域模型，装配形成轴流泵整体流场模型。同时为提高轴流泵及其流域的建模效率，对UG进行二次开发，以实现轴流泵及其流域的自动建模。对轴流泵流场内外壁面及其进出口面进行命名，将轴流泵流场模型、叶片翼型参数及命名的面导入到ANASYS WORKBENCH中，将ANASYS高性能分析技术与UG参数几何模型相结合，达到与UG双向共享参数化模型，实现CAD与CAE协同仿真[1]。结合流体力学相关知识计算轴流泵各边界条件，利用Workbench的Fluent模块选择标准模型，对轴流泵进行三维流体动力学分析。根据所得轴流泵流场数值分析结果，建立轴流泵扬程H的表达式，以叶片已知型值点为输入参数，，扬程H为输出参数，采用DOE技术生成叶片翼型坐标点与轴流泵性能参数之间的关系样本。采用MATLAB的ANN工具箱建立轴流泵叶片动力特性预测代理模型，即建立叶片翼型坐标点与轴流泵性能参数之间的BP神经网络模型，并利用以上样本对其进行网络训练和检测，从而形成具有一定精度的轴流泵叶片动力特性预测模型。将该网络模型以C语言形式输出，生成.dll文件，并将其集成到轴流泵辅助设计系统中。通过该设计方法，可实现叶片翼型曲线型值点到扬程的非线性映射及叶片动力特性的实时预测，提高轴流泵设计效率和水力性能。
[Abstract]:A new design method of axial flow pump blade is studied by synthesizing the theoretical knowledge of Bezier curve, 3D modeling, computational fluid dynamics (CFD) and artificial neural network (Ann). The design theory and modeling method of axial flow pump blade are improved, and the initial design of axial flow pump blade is carried out by using lift method and NACA airfoil. The Bezier curve is used to fit the coordinate points of the blade airfoil, four points are selected from the coordinate points of the airfoil of each section as the type value points, and the control points of the Bezier curve are obtained. Thus, the corresponding airfoil curves can be obtained. The curves are converted into UG curve expressions, the blade wing bone lines are generated by UG, and the parametric modeling of the axial flow pump is realized by combining other structural parameters, and the controllability of the blades is enhanced. With VB programming, the design parameters of axial flow pump are inputted. After selecting the blade airfoil, the program automatically designs the axial flow pump, and outputs the structural parameters of the axial flow pump and the UG modeling curve of the blade. The watershed model of axial flow pump impeller is established by using UG Boolean operation function, and the watershed model of axial flow pump horn tube and guide vane is established by UG WAVE technology. In order to improve the modeling efficiency of axial flow pump and its watershed, UG was redeveloped. In order to realize the automatic modeling of the axial flow pump and its watershed, the inner and outer wall surface of the axial flow pump field and its inlet and outlet surface are named, and the flow field model, blade airfoil parameters and named surface of the axial flow pump field are imported into ANASYS WORKBENCH. The high performance analysis technology of ANASYS is combined with the geometric model of UG parameters to achieve the bidirectional sharing parametric model with UG, and the cooperative simulation between CAD and CAE is realized [1]. The boundary conditions of axial flow pump are calculated by combining the relevant knowledge of fluid mechanics. The standard model of axial flow pump is selected by Fluent module of Workbench, and the three-dimensional hydrodynamic analysis of axial flow pump is carried out. According to the result of numerical analysis of flow field of axial flow pump, the expression of head H of axial flow pump is established, and the point of known type of blade is taken as input parameter. The head H is the output parameter and the DOE technique is used to generate the sample of the relationship between the coordinate points of the blade airfoil and the performance parameters of the axial flow pump. The prediction agent model of blade dynamic characteristics of axial flow pump is established by using ANN toolbox of MATLAB, that is, the BP neural network model between the coordinate point of blade airfoil and the performance parameters of axial flow pump is established, and the above samples are used for network training and detection. Thus a prediction model of the blade dynamic characteristics of axial flow pump with certain precision is formed. The network model is output in C language, and the .dll file is generated and integrated into the axial flow pump aided design system. Through this design method, the nonlinear mapping from the curve point of blade airfoil to the head and the real-time prediction of blade dynamic characteristics can be realized, and the design efficiency and hydraulic performance of axial flow pump can be improved.
【学位授予单位】：西华大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：TH312

【参考文献】