Optimization Of Characteristic Parameters Of Jet Pump Throat And Optimal Value Prediction

As a mechanical equipment that uses the jet to suck fluid,jet pump has a wide range of applications because of its simple and reliable mechanical structure.The problem of low pumping efficiency caused by the unsatisfactory design of the characteristic parameters of the throat section has been widely concerned.In order to solve this problem,this paper optimizes the structure of the inclination angle of the jet pump throat based on numerical calculation and visualization experiments.In order to further explore the coupling relationship between its feature parameters,this study creatively introduces machine learning technology to better solve the above problems.In this study,the inclination structure of the jet pump throat was optimized.Firstly,a mathematical model of the pumping process of the jet pump is established.Then,the accuracy of the mathematical model is verified by visual experiments.Finally,the numerical simulation of 12 sets of jet pumps under different working conditions is carried out,and the velocity field,pressure field and momentum correction coefficient of the flow process are systematically analyzed,and the optimization value of the inclination angle of the throat pipe is preliminarily explored,and its influence mechanism is studied.The results show that:（1）the numerical calculation results are similar to the experimental results,and the maximum error is only 2.13%.（2）Compared with the general situation,changing the inclination angle of the throat pipe can obtain a 13%improvement in pumping performance.（3）The linkage design of the inclination angle and length ratio of the throat pipe can save 28.6%of consumables while improving the pumping efficiency.In fact,the enhancement effect is achieved by the coupling enhancement of the inclination angle of the throat and the length of the throat,so it is necessary to explore the above-mentioned two-factor coupling linkage mechanism,and it is necessary to introduce machine learning technology to predict the complex relationship of multiple parameters.In this study,machine learning technology is creatively introduced to solve the complex coupling relationship between design parameters,and machine learning technology is used to adaptively determine the optimal parameters,thereby improving pumping efficiency.In this study,a total of 804 groups of working conditions composed of four characteristic parameters of jet pump pipe diameter,pipe length,pipe inclination angle and flow ratio parameters were numerically calculated.Five machine learning models are used to determine the coupling relationship between each parameter and pumping efficiency to achieve the prediction of optimal solution.The results show that:（1）the evaluation indexes of XGBoost(MSE=3.7574×10^-5,MAE=4.6×10^-3,R²=0.9785)are better than other models,and the prediction effect is the best,the error between the predicted results and the experimental results is only 3.31%.（2）The importance of the four features to model construction was 62.5%,8.5%,10.6%and 18.4%,respectively.（3）In the conventional optimal feature value interval,the optimal values of the optimized four features can bring 105.9%,12.2%,12.4%and 119.8%improvement in pumping efficiency,respectively,and the comprehensive increase is as high as 120.1%.In general,this study innovatively integrates machine learning and numerical computing to explore the concept of multi-factor coupling mechanism,and uses the above concept to improve the pumping efficiency of jet pumps,and obtain great performance improvement.This shows that this study not only has practical significance to solve the current production problems,but also has the development significance of combining the field of machine learning and numerical simulation computing.