Research On Multivariable Control Strategy Based On Neural Network Predictive Control In Cryogenic Wind Tunnel

Cryogenic high Reynolds number continuous transonic wind tunnel(hereinafter,cryogenic wind tunnel)is mainly used to conduct the aerodynamic force test at high Reynolds number for large scale aircraft.It has many and strong coupling control parameters in flow field control.It is a typical nonlinear,multivariable,strong coupling,time-varying control and large delay system.The design of flow field control strategy in cryogenic wind tunnel is difficult.At present,the wind tunnel control strategy is difficult to meet the demand of control system of cryogenic wind tunnel.It is significant to carry out a research about the key technology on cryogenic wind tunnel's flow field control.In cryogenic wind tunnel flow field control system,it mainly control Mach number,total temperature and total pressure accurately,to meet the requirements in different working condition.Mach number,total pressure and temperature control is coupling.At low temperature and subsonic condition,in Mach number,for example,Mach number is calculated by total pressure and static pressure.When the Mach number is adjusted with the speed of the compressor,the pressure control system will be affected.The same,adjusting the total pressure with exhaust valve,will affect Mach number control.So the Mach number control system is coupling with total pressure control system.The temperature in flow field is control by liquid nitrogen injection volume,and the temperature is closely related with heat from the compressor and the cold quantity by the nitrogen removal system.Because of the wind tunnel equipment and metal body piercing thermal inertia,the temperature control will lag,and it must be in consideration in temperature control.The total pressure is controlled by the mass from the flow,so the liquid nitrogen which can control the temperature is coupling with the pressure control.In addition,relationship with liquid nitrogen and temperature is affected with the liquid nitrogen vaporizing situation,there is a nonlinear and time-varying control system.In total,there are many variables in the cryogenic wind tunnel and coupling with each other,and the system is non-linear,time-varying,it become a problem to control the cryogenic wind tunnel flow field.In this paper,0.3m cryogenic wind tunnel flow field is the object of study,the “lumped parameter approach” which is based on the conservation equations(mass equation,momentum equation,power equation)of fluid mechanics to be used to establish a dynamic flow field parameters numerical model.The model was taken a preliminary test and the result show that the control variable in the flow field of cryogenic wind tunnel is nonlinear,and with a strong coupling between variables.The study group received partial data in 0.3m cryogenic wind tunnel to preliminary verify the model,the result shows that the model's outputs is similar to 0.3m cryogenic wind tunnel's outputs,the model will provide a reference for the further cryogenic wind tunnel control system design.To solve the flow field control problem in cryogenic wind tunnel,a predictive control algorithm which is based on neural network is proposed to design cryogenic wind tunnel flow field control strategy.The principle of predictive control and the neural network on-line identification algorithm is studied,and they are combined to design control algorithm.A BP neural network predictive control algorithm is designed for cryogenic wind tunnel flow field control.Combined with the characteristics of cryogenic wind tunnel flow field,a simplified controller multivariable optimization algorithms and a variable step predictive control are proposes to improve program efficiency and improve it can be used in practice.The simulation result shows that the predictive control based on neural network can meet the wind tunnel multivariable control requirements.It provides a valuable solution for the future construction of the actual cryogenic wind tunnel.