Acquisition And Verification Of Prepar3D-based Flight Aerodynamic Data

The acquisition of flight data and the modeling of aircraft systems are of great significance in the research of aircraft overall design,flight quality assessment,and accident investigation.Traditional aircraft modeling generally uses ground wind tunnel tests,aerial flight tests,etc.,and the research cost is relatively high.In order to ensure that the experiment is convenient,efficient,and low cost,this paper uses the Cessna 172 flight simulator and flight simulation software Prepar3 D to conduct flight simulation tests,and then extracts flight test data.In order to obtain the aerodynamic model,based on the idea of machine learning,this paper uses artificial neural network technology to process multi-dimensional nonlinear flight test data and verify the identification results.Firstly,the appropriate modeling method should be determined in the paper.For aerodynamic parameter identification,the six degree of freedom dynamic equations of the rigid body of the aircraft are often selected as the state equations of the system.After the coordinate transformation is done,the six degree of freedom motion equations of the system axis is obtained,such as the attitude angle equations,the centroid dynamics equations and the dynamic equations around the centroid.On this basis,several traditional mathematical expressions and typical neural network methods commonly used in modeling are briefly introduced.Secondly,the aerodynamic data are acquired based on Prepar3 D,and the identifiability analysis is performed.The common methods of aircraft parameter identification and their applicable scenarios are compared.Considering the dynamic non-linear characteristics of the aircraft system model,the recurrent neural network has a feedback structure and can store information,which can well characterize the characteristics of the aerodynamic model.Finally,the obtained aerodynamic data is processed and used to identify the aerodynamic parameters of the neural network to establish the aerodynamic model of the aircraft.And verify its fitting effect.Through experimental verification and analysis,it can be known that the aerodynamic data obtained from flight simulation experiments can have good practicability for the parameter identification of its model,and the recurrent neural network proposed in this paper can complete accurate modeling for aerodynamics.