Research On Nonlinear Controller Design Of Turbofan Engine Based On Data-driven

Turbofan engines are widely used in civil and military aviation fields due to their high propulsion efficiency and low fuel consumption.However,with the improvement of the performance requirements of the aircraft for the propulsion system,the structure of the turbofan engine becomes more complex,and the nonlinearity of the system and the coupling characteristics between variables are also enhanced accordingly.Therefore,it is necessary to explore the advanced controller design method to meet its performance and safety requirements.In recent years,with the development of artificial intelligence technology,data-driven controller design method is gradually developed in the field of aero-engine.Therefore,this thesis relies on the key project of a ministry ’XX engine basic problem research’ to carry out the design of nonlinear controller based on data-driven for a certain type of turbofan engine.The main research contents include:Aiming at the design problem of direct controller for turbofan engine,a controller design method based on adaptive enhancement is proposed.Firstly,the least squares support vector machine is introduced into the model reference controller design architecture,and the controller design is completed based on convex optimization method,so as to ensure the asymptotic convergence of the controller.Secondly,a turbofan engine controller based on adaptive enhancement is designed by using multiple basic controllers and adaptive enhancement algorithm.The simulation results show that the controller design can reduce the steady-state control error of turbofan engine,and the over-fitting is effectively suppressed due to the introduction to adaptive enhancement algorithm.Aiming at the design problem of multivariable controller for turbofan engine,a control method based on sliding mode variable structure is proposed.Firstly,the control parameters and objectives of this turbofan engine are analyzed,and the simulation working point is determined.Secondly,the sliding mode controller of turbofan engine is designed to control fuel flow and nozzle area.The power reaching law with saturation is used to prevent the chattering of switching surface,and the gain and saturated boundary layer are optimized by Whale Optimization Algorithm(WOA).Hardware-in-the-loop verification results show that the algorithm can ensure real-time requirements,steady-state error is less than 1%,and has good control performance.In order to further to improve the performance of turbofan engine multivariable controller,this thesis proposes two design methods of turbofan engine multivariable controller based on data drive.Firstly,an RBF Neural Network sliding mode controller is proposed to compensate the unmeasured disturbance of the engine system in real time.Simulation results show that the proposed method can improve the response time of the system,and the steady-state error is smaller than that of the classical sliding mode control.Secondly,a causal convolution neural network controller is proposed,which uses regularization to reduce over-fitting and uses WOA to optimize initial weights and learning rates.The simulation results show that the proposed method can further to reduce the steady-state error and meet the control requirements of turbofan engine.