Research On Vision-based SNNPD Vibration Control Reduction Of Tail-supported Model In Wind Tunnel

The wind tunnel test is a necessary method for the development of the aircraft.The tail support structure is the most commonly used in wind tunnel test,which has little influence on the flow field around the model.In this way,under the conditions of high sound speed and high angle of attack,it is prone to large vibration,which seriously affects the test range of the performance of the aircraft and even endangers the wind tunnel test equipment.The vibration can be reduced by an active vibration control method using a piezoelectric actuator.However,current methods mostly use contact measurement such as accelerometers,and inevitably introduce additional quality.In terms of control algorithm,the commonly used control method in engineering is PID control,and the control parameters are fixed.In this way,when the vibration changes dramatically at different angles of attack,the selection of control parameters is difficult to adapt to the whole process of change.There is a contradiction between control performance and stability,and the control effect and adaptive adjustment ability are poor.In this paper,research is conducted on such issues,which the model-sting system with tail support structure generates low-frequency vibration in subsonic and transonic wind tunnel tests.(1)The various unstable factors in the transonic and subsonic wind tunnels are analyzed.With the experimental process and structural characteristics of the system's variable angle of attack,the main causes of the wind-induced vibration of the system are analyzed.Combined with the principle of force and electric of piezoelectric,the closed-loop control of the vibration of the wind tunnel tail support model is expounded.Based on the core aspects of vibration detection and algorithm in closed loop,the overall scheme of vision-based vibration measurement and system-observer neural-network PD method for vibration control reduction of tail-supported model in wind tunnel is proposed.(2)The performance of the traditional vibration measurement method in the wind tunnel tail-supported model is analyzed.Acceleration sensor is not accurate enough.The non-contact direct displacement measurement method is analyzed,and the visual measurement is feasible.The vibration pixel coordinates are obtained from the image processing method of the vibration displacement information.However,there is too much redundant information,and the data cooperation processing architecture of FPGA and CPU is adopted.A high-speed extraction and transmission method based on image vibration displacement is proposed to make up for the real-time problem of visual measurement.Furthermore,the vibration measurement method of the wind tunnel tail-supported model based on machine vision is proposed,and the real-time measurement of the vibration displacement of the wind tunnel tail support model is realized.(3)The experimental data of subsonic and transonic wind tunnels with variable angle of attack is studied.The analysis shows that the vibration changes dramatically at different angles of attack.Different vibration conditions require different control parameters to give consideration to both control performance and stability.Aiming at the problem of poor adaptive adjustment ability of traditional fixed parameter control,a vibration suppression method of neural network PD based on system observer(SNNPD)is proposed.The stability is demonstrated by Lyapunov theorem.And the simulation test and excitation test are carried out to verify the results.(4)A laboratory platform is built,and the accuracy of the vision measurement system is verified with a high-precision fiber displacement sensor.The visual method is used to measure the modal and vibration mode information of the system.The hammering experiment are carried out under the controller off,PD control and SNNPD control.The vibration stabilization time is shortened to 0.92 s,and damping ratio is chosen as the evaluation index,which is increased to 0.00656 from 0.0064 compared with uncontrolled.It is verified that the SNNPD control algorithm proposed in this paper can flexibly adjust parameters and shows better control performance.