Research And Implementation Of Aerodynamic Data Modeling Method And Visualized Interaction System For Aircraft

The establishment of aerodynamic model is the most important part in aircraft engineering.The accuracy of the model will directly affect the control effect of the aircraft and the credibility of the simulation system.With the increasing mobility of modern aircraft,the expansion of flight envelope and the increasing complexity of flight conditions,the problem of nonlinear aerodynamic characteristics becomes more and more obvious.How to establish an accurate aerodynamic model becomes one of the current research topics.Aiming at the nonlinear aerodynamic problems,this paper studies the aerodynamic modeling from two directions of traditional mathematical fitting method and artificial intelligence method,and achieves good results.Finally,based on the previous research,an online aerodynamic data modeling system is constructed,which can provide a more intuitive operation mode to complete the aerodynamic modeling work.The specific research contents are as follows:Firstly,an aerodynamic data modeling method based on moving least square method is proposed.When the typical moving least square method is used to fit the aerodynamic data,the radius of support region is determined by empirical formula,and the accuracy of aerodynamic model is random.In order to improve the accuracy of the moving least squares model,the genetic algorithm is used to obtain the best weight function influence factor ? and the best support radius when constructing the moving least squares model.Finally,the nonlinear aerodynamic model is obtained and compared with the aerodynamic model established by partial least squares method.The experimental results show that the aerodynamic model established by this method has higher accuracy,and its modeling effect is better than partial least squares method,which is suitable for nonlinear aerodynamic fitting.Secondly,a deep confidence network based aerodynamic data modeling method is proposed.Because the traditional depth belief network(DBN)will cause information loss when dealing with the continuous aerodynamic data,this paper modifies the structure of the network,and adds the Gauss-Bernoulli restricted boltzmann machine(GBRBM)to the bottom layer of the traditional depth belief network.The GBRBM-DBN network can effectively deal with the continuous data.Compared with BP neural network model,this method can overcome the shortcoming that traditional neural network is easy to fall into local optimum through unsupervised pre training stage,and the aerodynamic model established by this method has higher accuracy and certain extrapolation ability,which has research and application value.Finally,based on the previous research work,an interactive online aerodynamic data modeling system is constructed.The system is mainly divided into three modules for design and Implementation: one is the data upload module,which collects and manages the uploaded data files in the form of tables,and supports the functions of viewing,deleting and downloading data files;the other is the data modeling module,which supports the aerodynamic modeling work in the browser,and displays the modeling results in the form of three-dimensional charts;The third is the model result module,which supports the view of modeling results and the prediction of aerodynamic parameters.In this paper,the system is tested,the test results show that the system has a friendly interface,each functional module works normally,and is consistent with the expected goal.