Research On Simulation Of Aircraft Pitch Channel Control Based On QAR Data

The aircraft virtual maintenance simulator is a simulation equipment for training crew and pilots.The flight simulation function can combine the failure effect with the flight process to assists pilots in completing the normal flight procedures and failure procedure training.The pitch control method is involved in each flight stage,so the pitch channel control simulation is an important link to improve the flight simulation function.To simulate the automatic flight control law of the specified model,it is usually necessary to establish a relatively reliable mathematical model with the help of original factory design data,flight test data and wind tunnel test data.The flight simulation function is applied to the maintenance simulator,and the simulation accuracy is not high,so it can be realized by using the actual aircraft operation data.The quick access recorder(QAR)records various parameters in detail.This type of data includes mass characteristic parameters,dynamic parameters,control surface deflection parameters,atmospheric parameters,etc.required for the simulation.Therefore,based on the QAR data of a specific aircraft,this thesis explored the simulation method of pitch channel control under this aircraft.Firstly,Sorted out the relationship between various modules of flight simulation through the flight crew operation manual(FCOM),and the pitch control mode logic was divided into different selection modes in the flight guidance(FG).Selected appropriate parameters from the thousands of data recorded by QAR for pre-processing such as filtering and resampling,and solved for missing variables in the data,so that the data was reasonably available and provided reliable data support for control law simulation.Secondly,inducted general dynamic equations,and constructed the aerodynamic sample set from A320 QAR data.Since the aerodynamic characteristics of the aircraft were closely related to the actual flight state,the relationship between aerodynamic force and various state quantities was highly nonlinear.In order to improve the modeling accuracy,based on the polynomial aerodynamic modeling of the research group,the weighted K-nearest neighbor(WKNN)modeling method based on differential evolution(DE)was proposed to improve the Euclidean distance.The nonlinear aerodynamic force and moment models were established in a targeted manner,which provided the necessary model support for the pitch channel control simulation.Thirdly,analyzed various control strategies such as PID,neural network,dynamic inverse,etc.,and determined the access time of automatic pilot(AP)mode by referring to the variation law of corresponding variables in the QAR data.Designed the control loop of pitch channel according to different AP modes.Used the fminsearch function to solve the balance state of the aircraft and determined the starting point of the simulation.Applied the nonlinear dynamic inverse strategy based on state feedback to complete the simulation design of stabilization loop.At the same time,for the pitch attitude control problem with inaccurate model parameters,an adaptive compensation structure was established in the pitch angular velocity loop to correct the system inverse error caused by imprecise modeling.For variables with relatively stable transient processes such as altitude,speed,and vertical speed,the guidance loop adopted an improved neural network PID control strategy.The objective function was constructed from the deviation between the QAR data and the system output,and the objective function was minimized by the improved particle swarm optimization(IPSO)algorithm.Completed the neural network weights update,and then adjusted the controller gain.Finally,combined with the A320 QAR data,a pitch channel control module was established based on MATLAB/Simulink,and the flight simulation verification was carried out for typical AP modes corresponding to different flight stages.The simulation results show that the simulation model output is different from the corresponding QAR reference data,but the time-varying trend is similar,which meets the flight simulation function requirements in the maintenance simulator.