Axial Flow Pump Blade Dynamic Forecast Model Research

Integrating lift method, Bezier curve,3D modeling, computational fluid dynamics (CFD),ANN theory and so on, research a new method of axial flow pump blade design, that is:Improving the design theory and modeling method of blade of axial flow pump, use the liftmethod and NACA airfoil to design the initial blades of axial flow pump,and then changeblades each section airfoil point coordinate to space coordinates. Using Bezier curve to fit bladeairfoil coordinates points, select four point as data point from each section airfoil coordinatespoints, and then reverse Bezier curve control point, thus we can achieve the correspondingcurve of wing. We put UG curves expression which change from each section curves into UG togenerate blade airfoil bone line, and combining with other structural parameters realizeparametric modeling of axial flow pump,and then enhance the controllability of blade when theblade has minimal control parameters. Using VB programming,realizate the automatic designof initial blade after inputing axial flow pump and choosing blade airfoil, and output thestructure parameters of axial flow pump and the UG modeling curve of blade.By using UG Boolean function establish the model of axial flow pump impeller drainagebasin, and by using the UG WAVE technology establish the model of axial flow pump guidevane and bell mouth drainage basin, and then by assembly form axial flow pump overall flowfield model. At the same time for improving modeling efficiency of axial flow pump and itsdrainage basin, carried out two times development on UG, so as to realize axial flow pump andits drainage basin automatic modeling. Named axial flow pump flow field inside and outsidewall、import and export surface,put axial flow pump flow field model、blade airfoilparameters and named surface into ANASYS WORKBENCH, and combine ANASYS highperformance analysis technology and UG parameter geometric model to achieve bi-directionalshared parameter model, and then realizes the CAD and CAE collaborative simulation.Combining fluid mechanics knowledge to calculate axial flow pump various boundaryconditions, and use the Workbench Fluent module selecting criteria model to analysis fluiddynamics of axial flow pump. According to the numerical analysis results of flow field ofaxial flow pump to establish the expression of axial flow pump head H, with the blade knowndata points as input parameters and H as output parameters, using DOE technology generatethe relationship sample between blade airfoil coordinates and axial flow pump performanceparameters.Using MATLAB ANN toolbox establish axial flow pump blade dynamic characteristicprediction model, namely establish BP neural network model between blade airfoil coordinatespoint and axial flow pump performance parameters, and use the above sample to train and test the network, thus form a certain accuracy of axial flow pump blade dynamic predictionmodel.We integrate the.dll file outputed neural network model as C way into the axial flowauxiliary design system. In the axial flow pump blade design process achieve real-timeperformance prediction to improve the efficiency of axial flow pump design and usingperformance.Through the design way，we can realize the nonlinear mapping from blade airfoil curvepoints to lift and the real-time forecast of blade dynamic, which can improve the designefficiency of axial flow pump and hydraulic performance...