Research On Numerical Simulation Of High-precision Flow Around And Intelligent Real-time Calculation Strategy

In the field of tall buildings design,circular cylinder and rectangular cylinder are the most widely used foundation structures.Because of the sensitivity of cylinder structure to wind load,wind-induced vibration effect has become a problem that must be considered in the design process of tall building.Under the action of high-speed wind load,the flow field of circular and rectangular cylinders has become one of the most important research topics in civil engineering and related fields.In this paper,the GSM-CFD algorithm combined with the gradient smoothing method and computational fluid dynamics is used to simulate the flow around two-dimensional circular and rectangular cylinders.The artificial neural and network programs are edited by Python language and Tensorflow system.The data of the simulation results of the flow around the cylinder after finishing are trained to achieve the two-dimensional cylinder under different conditions The purpose of flow field parameter prediction mainly includes the following aspects:First of all,a two-dimensional circular cylinder flow model is established.The GSM-CFD algorithm combined with computational fluid dynamics and gradient smoothing method is used to simulate the flow around the circular cylinder.The effects of the diameter of the circular cylinder and Reynolds number on the coefficient of lift,coefficient of drag and Strouhal number are analyzed.The results of numerical simulation are compared with those of literature,which verify the accuracy and stability of GSM-CFD algorithm in fluid analysis.Secondly,based on GSM-CFD algorithm,the numerical simulation and analysis of flow around rectangular cylinder with different width thickness ratio,different size and different flow angle of attack are carried out.The characteristics of coefficient of lift,coefficient of drag and Strouhal number of two-dimensional rectangular cylinder with Reynolds number,width thickness ratio and flow angle of attack are obtained,and the different Reynolds number is further analyzed.The influence of number,width thickness ratio and angle of attack of flow field on the amplitude and frequency variation of flow around a two-dimensional rectangular cylinder is studied.The curve of resistance coefficient variation with width thickness ratio obtained by numerical simulation is compared with the results of relevant literature,which verifies the correctness and reliability of numerical simulation.Finally,a neural network model is built to predict the coefficient of drag of two-dimensional rectangular cylinder flow.Combined with the characteristics of coefficient of drag changing with the flow angle of attack under different width thickness ratio,the coefficient of drag of rectangular cylinder flow around different width thickness ratio and different flow angle of attack is trained by artificial neural network,and the coefficient of drag of rectangular cylinder flow around is trained by the trained neural network model.The neural network prediction curve of coefficient of drag changing with the angle of attack of flow field under different width thickness ratio is obtained,and compared with the numerical simulation results,the correctness of neural network prediction is verified.