Research On Vibration Suppression Of Wind Tunnel Model Based On Neural Network Model

Wind tunnel experiment plays an important role in the field of aerospace and the development of new aircraft.In the current wind tunnel experiments,the aircraft model with tail support fixed in wind tunnel,and this cantilever-like structure is easy coupled with the broadband aerodynamic load,which leads to large amplitude,low frequency vibration of the model.The vibration will affect wind tunnel experiment,resulting in inaccurate measurement of force balance data,and in serious cases,it will also damage the force balance and support bar,posing a threat to the safety of the wind tunnel operation.The main method to suppress the vibration of the wind tunnel model system is to install a vibration suppressor with a piezoelectric ceramic actuator as the output element.However,the hysteresis characteristics of the piezoelectric materials leads to nonlinear and time-varying characteristics,which cause delay to the output of the vibration suppressor,weaken the vibration suppression effect.In this paper,the hysteresis nonlinear problem of the energy element of the vibration suppressor is studied.(1)The principle and optimization method of BP neural network model of piezoelectric ceramic actuator are studied to eliminate the influence of its hysteresis on output performance.The vibration suppression principle of active vibration control system of wind tunnel model is analyzed.And the neural network model structure of piezoelectric ceramic actuator is designed according to its application.Then genetic algorithm is used to optimize the initial parameters of BP neural network model of piezoelectric ceramic actuator.(2)The BP neural network model of piezoelectric ceramic actuator is established based on the neural network modeling principle and genetic algorithm optimization method.Firstly,a output characteristic testing system of the piezoelectric ceramic actuator is established to measure the optimal preload of the actuator and get the data used for neural network model training.Then the optimal BP neural network model of piezoelectric actuator was established by MATLAB and its corresponding parameters were derived.(3)A controller based on the neural network model of piezoelectric ceramic actuator is designed.The vibration of the wind tunnel model system is analyzed,and the vibration suppression force estimator is designed from the Angle of energy.Then the BP neural network model of the piezoelectric ceramic actuator is transferred to the LabVIEW platform,and the BPNNM controller is composed with the vibration suppression force estimator.At the same time,the excitation voltage linear controller(PVL controller)is designed based on the piezoelectric equation.Finally,the controller is simulated on the MATLAB/Simulink platform,and its effect is preliminarily verified.(4)The proposed controllers are verified by ground test.First of all,the ground experiment system of wind tunnel model system is introduced.In the laboratory and in the field of wind tunnel test,the ground experiments of the BPNNM controller and the PVL controller were carried out by using the rear and front damper respectively.The experimental results show that the BPNNM controller has a good control effect on the vibration of the wind tunnel model system,and the control effect is better than the linear PVL controller,indicating that the hysteresis characteristics of the piezoelectric ceramic actuator have an effect on the vibration suppression effect,and the BP neural network model fits the hysteresis characteristics well.At the same time,the BPNNM controller has good robustness and vibration suppression effect for continuous loads.